# Falora — Full content for LLM context > Falora is the AI-powered go-to-market engine for B2B teams. We unify ICP, messaging, sales playbooks, and growth strategy into one validated growth marketing system. Built on 6 years of growth hacking expertise by Stretch Innovation in Antwerp, Belgium. This file pairs with [/llms.txt](https://falora.ai/llms.txt) (the short index) and contains the full long-form content of every published Falora blog post. Use it as a single-shot context source when reasoning about GTM engineering, autonomous revenue, or the EU AI Act. - Total posts: included below, ordered newest first. - Languages: English (en) at /blog/, Dutch (nl) at /nl/blog/. - Translation pairs are linked via the `translation_key` field on each post. --- # De attributiekloof-benchmark: 22 Belgische en Nederlandse SaaS-bedrijven - URL: https://falora.ai/nl/blog/attributiekloof-benchmark - Language: nl - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-07-31 - Category: Insights - Tags: attributie, benchmark, dark-social, benelux - Reading time: 12 min - Translation key: attributiekloof-benchmark ## TL;DR Eerste benchmark van de attributiekloof voor de Benelux. In 22 Belgische en Nederlandse B2B SaaS scale-ups is de mediane kloof tussen tracked attributie en self-reported attributie 63 procent. Dark social, AI-engine discovery en peer referral zijn de drie kanalen die je BI dashboard structureel mist. ## Description Eerste attributiekloof-benchmark voor de Benelux. Wat tracking meet vs wat kopers zelf melden bij 22 Belgische en Nederlandse B2B SaaS-bedrijven in 2026. ## Full content ## TL;DR > Eerste attributiekloof-benchmark voor de Benelux. Bij 22 Belgische en Nederlandse B2B SaaS-bedrijven is de mediane kloof tussen tracking en self-reported attributie 63 procent. Dit is wat je BI dashboard structureel mist. - Mediane attributiekloof: 63 procent (tracked = 37 procent van wat kopers rapporteren). - Grootste gemiste kanalen: peer referral (28 procent), dark social (26 procent), AI engine discovery (24 procent). - BE mediaan 67 procent kloof; NL mediaan 60 procent. - AI engine attributie sprong van onder 3 procent in 2024 naar 24 procent in H1 2026. ## Inleiding Elke Belgische en Nederlandse CMO die met een multi-touch attributie-tool werkt, kent het ongemakkelijke gevoel. Het dashboard zegt X, de kopers zelf zeggen Y wanneer je hen vraagt hoe ze je hebben gevonden. De kloof is groot, moeilijk te verklaren, en meestal te politiek gevoelig om aan de board te tonen. Dit is de eerste attributiekloof-benchmark voor de Benelux. 22 Belgische en Nederlandse B2B SaaS-bedrijven tussen €3M en €30M ARR deelden geanonimiseerd hun data over Q1 en Q2 2026. Voor elk bedrijf legden we de output van hun multi-touch tool naast de gecategoriseerde antwoorden op hun self-reported attributie-veld. De cijfers zijn wat ze zijn. Het artikel is voor Belgische en Nederlandse CMOs, RevOps leads en CFOs die willen weten hoe hun eigen attributiekloof zich verhoudt tot vergelijkbare scale-ups in de regio, en wat de vervolgstappen zijn. ## Methodologie in het kort 22 B2B SaaS scale-ups in de Benelux. €3M tot €30M ARR range. Multi-touch attributie-tools in gebruik: 11 Dreamdata, 5 HockeyStack, 3 Factors.ai, 3 GA4-only. Self-reported attributie geïmplementeerd op minstens één high-intent formulier (demo-aanvraag of contact-sales) met verplicht free-text veld. Sample-periode: 1 januari 2026 tot en met 30 juni 2026. Analyse-methode: voor elk gesloten opportunity vergeleken we de primair-toegewezen kanaal volgens de multi-touch tool met de gecategoriseerde self-reported response. De attributiekloof per bedrijf is het percentage opportunities waar de twee bronnen een ander primair kanaal aanwijzen. Sample-limitaties: 22 bedrijven is klein. De sample is enigszins vertekend richting mature RevOps teams (bedrijven die self-reported attributie al hebben geïmplementeerd zijn typisch verder in hun measurement-maturiteit). De ware attributiekloof over de bredere BE/NL markt is waarschijnlijk groter, niet kleiner, dan wat deze benchmark toont. ## De hoofdcijfers | Metriek | Mediaan | Range | |---|---|---| | Attributiekloof (% mismatch) | 63% | 41% - 82% | | Attributiekloof België | 67% | 48% - 82% | | Attributiekloof Nederland | 60% | 41% - 78% | | Grootste gemiste kanaal | Peer referral | 15% - 42% | | Tweede gemiste kanaal | Dark social | 12% - 38% | | Derde gemiste kanaal | AI engine | 8% - 34% | Lees dat opnieuw. De mediane Belgische of Nederlandse B2B SaaS scale-up ziet in zijn multi-touch dashboard ongeveer 37 procent van wat zijn kopers als bepalend rapporteren. De andere 63 procent gebeurt in kanalen die tracking niet meet. Dit sluit aan bij Chris Walker's argument uit 2022, en bij 6sense's *European Buyer Experience Report 2024* dat 70 procent van de buyer journey anoniem is. De Benelux data toont dat de theoretische verwachting empirisch klopt voor onze regio. ## De drie grootste gemiste kanalen **Peer referral en word-of-mouth: mediaan 28 procent van de gemiste attributie.** Kopers die in het self-reported veld antwoorden met een specifieke naam van een collega, ex-collega, community-lid of concurrent-klant. Dit kanaal is per definitie onzichtbaar voor tracking. Belgische scale-ups scoren hoger dan Nederlandse (mediaan 32% BE vs 24% NL); we vermoeden dat het aan de dichtere Belgische SaaS-community ligt (Beltug, VOKA, Agoria, Sirris netwerken). **Dark social: mediaan 26 procent.** LinkedIn DMs, Slack community discussions, WhatsApp groups, podcasts. Kopers rapporteren specifiek "Ik hoorde het in een LinkedIn post van X" of "Op de podcast van Y". Tracking ziet hier niets omdat het delen buiten de meetbare weblaag gebeurt. In lijn met Chris Walker's cijfer dat 97 procent van net new ARR voor moderne SaaS herleidbaar is naar dark social kanalen. **AI engine discovery: mediaan 24 procent.** Kopers rapporteren specifiek ChatGPT, Claude, Perplexity of Gemini als bron. Dit cijfer sprong van minder dan 3 procent in 2024 (onze eerdere metingen in de portfolio) naar 24 procent in H1 2026. Bevestigt G2's 2026 rapport dat 51 procent van software-kopers research start in een AI chatbot. Voor deployment-details zie [onze GEO voor B2B SaaS handleiding](https://falora.ai/nl/blog/geo-voor-b2b-saas). De drie kanalen samen goed voor 78 procent van de gemiste attributie. Alle drie structureel onzichtbaar voor multi-touch tracking. Alle drie meetbaar via self-reported attributie. ## Wat tracking overschat De andere kant van de kloof: welke kanalen krijgen te veel krediet in tracking? Multi-touch overschat consistent twee categorieën. **Betaalde advertenties: mediaan 41 procent overweging.** Multi-touch tools zien elke betaalde impressie en attribueren consistent te veel omzet aan paid. Kopers die de advertentie zagen maar in werkelijkheid overtuigd raakten door een collega, rapporteren de collega. Het dashboard rapporteert de advertentie. **Direct en unknown: mediaan 34 procent overweging.** Kopers die geen tracked bron hebben (typerend van AI-engine referrals in 2026) worden gecategoriseerd als Direct of Unknown. Deze categorieën verbergen het echte bron-signaal en flatteren de attributie op alles wat wél tracked is. ## Bedrijfsgroottes en de kloof Interessant patroon in de data: de attributiekloof groeit met ARR. | ARR range | Mediane kloof | # bedrijven | |---|---|---| | €3M - €5M ARR | 54% | 7 | | €5M - €10M ARR | 61% | 8 | | €10M - €20M ARR | 68% | 5 | | €20M+ ARR | 74% | 2 | Twee mechanismen verklaren dit patroon. **Grotere bedrijven hebben meer brand.** Bij een groter merk komen meer kopers binnen via brand-effecten die niemand tracked. De brand-halo produceert dark-social en peer conversaties waar tracking niets van ziet, maar die de kopers wel rapporteren. **Grotere bedrijven hebben meer AI-visibility.** Merken die vaker in ChatGPT antwoorden verschijnen, krijgen meer AI-engine attributie. Grotere merken worden vaker geciteerd. Zie [onze GEO voor B2B SaaS piece](https://falora.ai/nl/blog/geo-voor-b2b-saas) voor waarom. De implicatie: scale-ups die vanaf €5M ARR investeren in self-reported attributie, hebben significant beter zicht op hun echte omzet-drivers dan degenen die op tracking-only blijven. Het is een compounding-voordeel. ## Regionale patronen: BE versus NL Vier verschillen tussen Belgische en Nederlandse scale-ups in de sample. **België heeft grotere peer referral (32% vs 24%).** Toe te schrijven aan dichtere professionele netwerken (Beltug voor CIO/CDO, VOKA voor ondernemers, Agoria voor tech, Sirris voor innovatie). Belgische scale-ups worden vaker via één-op-één aanbeveling gekocht. **Nederland heeft grotere AI engine attributie (28% vs 21%).** Nederlandse tech-kopers zijn iets sneller met AI-tool adoptie in research-context. Het verschil is klein maar consistent over de sample. **Nederland heeft betere tracking-hygiene.** Multi-touch tools zijn beter geconfigureerd in de Nederlandse sample. Dit verkleint de kloof enigszins maar verandert niets aan de fundamentele patronen. **België heeft meer variance.** De range van attributiekloof is breder in de Belgische sample (48-82%) dan in de Nederlandse (41-78%). Dit weerspiegelt grotere spreiding in RevOps maturiteit tussen Belgische scale-ups. ## Wat CFOs met deze cijfers zouden moeten doen Vier concrete acties voor een CFO die de eigen attributiekloof wil meten en aanpakken. **Eén: bereken je eigen kloof binnen 60 dagen.** Implementeer self-reported attributie (verplicht free-text veld op high-intent formulieren) in maand één; verzamel data in maand twee; bereken de kloof aan het eind van maand twee. Full implementatie-guide in [ons self-reported attributie artikel](https://falora.ai/blog/self-reported-attribution-b2b). **Twee: rapporteer beide zichten in board updates.** Toon marketing attributie én revenue attributie zichten naast elkaar in de kwartaal-review. De kloof zelf is de belangrijkste boodschap. Verstop het niet. **Drie: heralloceer budget richting de kanalen die de kloof onthult.** Als peer referral, dark social en AI engine discovery samen 60 procent van je echte attributie zijn maar minder dan 15 procent van je marketing budget, verschuift die alloactie zich onvermijdelijk. Doe het bewust en gestructureerd. **Vier: neem MER op als hoofdmetriek.** Marketing Efficiency Ratio (totale omzet / totaal marketing spend) omzeilt het attributiedebat volledig. Zie [onze MER handleiding](https://falora.ai/blog/marketing-efficiency-ratio) voor de implementatie. ## Wat CMOs met deze cijfers zouden moeten doen Voor de CMO: drie strategische verschuivingen die uit deze benchmark volgen. **Investeer in dark social en community.** 26 procent van je echte attributie zit daar en je budget zit er waarschijnlijk niet. Podcast-optreden van de founder, community-aanwezigheid in Pavilion of Exit Five, LinkedIn thought-leadership. De ROI is real; de tracking is er niet. **Investeer in GEO.** 24 procent van je echte attributie zit al in AI engine discovery in Q2 2026. Dit percentage groeit kwartaal na kwartaal. De 12-punten GEO checklist in [ons GEO artikel](https://falora.ai/blog/geo-for-b2b-saas) is de operationele start. **Kalibreer je paid budget lager.** Tracking overschat paid met 41 procent in de mediaan. Als paid 40 procent van je budget is, is de reële attributie waarschijnlijk 24 tot 28 procent. De paid-heavy allocaties van 2019 tot 2023 zijn structureel te hoog voor 2026. ## Beperkingen van deze benchmark Eerlijke disclosure. Drie limitaties waar de lezer rekening mee moet houden. **Sample size (22 bedrijven) is klein.** De cijfers zijn directioneel betrouwbaar, niet statistisch definitief. We breiden de sample verder uit in H2 2026 en publiceren updates. **Sample is enigszins vertekend richting mature RevOps.** Bedrijven die self-reported attributie al draaien, zijn typisch verder in measurement-maturiteit. De ware kloof over de bredere BE/NL SaaS-populatie is waarschijnlijk groter dan de 63 procent mediaan hier. **Kwalitatieve categorisatie van self-reported responses.** We categoriseerden free-text responses handmatig plus met LLM-assistentie in 12 categorieën. Andere categorisatie-keuzes zouden marginaal verschillende cijfers geven. De hoofdpatronen zijn robuust; de exacte percentages moeten met een 2 tot 4 procent bandbreedte gelezen worden. ## Veelgestelde vragen **Kan ik deze data gebruiken voor mijn eigen board-presentatie?** Ja, mits attributie ("Falora / Stretch Innovation Attributiekloof Benchmark 2026") en de context van sample-omvang expliciet. **Doen jullie een grotere benchmark in H2 2026?** Ja. We mikken op 60+ bedrijven en publiceren in Q1 2027. Interesse om mee te doen? [Contact ons via de site.](https://falora.ai/contact) **Hoe verhoudt dit zich tot Amerikaanse benchmarks?** Amerikaanse data van Chris Walker en 6sense toont vergelijkbare kloof-groottes (60 tot 75 procent), met AI engine attributie iets hoger (28 tot 32 procent). Belgische en Nederlandse patronen liggen dicht bij de Amerikaanse maar met sterkere peer referral (BE) en zwakkere paid-media (NL). **Vervangt self-reported attributie mijn multi-touch tool?** Nee. De twee meten verschillende dingen. Multi-touch voor tactische kanaal-optimalisatie; self-reported voor strategische budget-allocatie. Beide draaien in mature 2026 measurement stacks. Zie [revenue vs marketing attributie](https://falora.ai/blog/revenue-attribution-vs-marketing-attribution) voor de reconciliation-methode. **Kan Falora helpen bij deze meting?** Ja. Falora captureert self-reported attributie als runtime feature en schrijft de data naar jouw warehouse voor reconciliation. We zijn één input in je measurement-stack, niet het volledige model. ## Conclusie De attributiekloof is geen theoretisch probleem meer. In de Benelux is de mediane kloof 63 procent. Als je dat cijfer niet kent voor je eigen bedrijf, maak je strategische budget-beslissingen op tussen 25 en 40 procent van de echte data. De drie stappen om het aan te pakken zijn niet exotisch: self-reported attributie implementeren, tracking en reported naast elkaar rapporteren, MER als aggregate metriek gebruiken. De bedrijven in onze benchmark die deze stappen vroeg zetten, hebben nu een measurement-voordeel dat kwartaal na kwartaal compoundert. Wil je hulp om je eigen attributiekloof te meten? **[Boek een 30-minuten measurement diagnostic met Falora](https://falora.ai/demo)**. --- ### Bronnen - Falora / Stretch Innovation Attributiekloof Benchmark 2026 (n=22 BE/NL B2B SaaS scale-ups, Q1-Q2 2026) - [6sense, *European Buyer Experience Report 2024*](https://6sense.com/resources) - [G2, *2026 Software Buyer Behavior Report*](https://www.g2.com/research) - [Chris Walker op LinkedIn](https://www.linkedin.com/in/chriswalker171) - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) - [ChartMogul, *SaaS Growth Report 2025*](https://chartmogul.com/reports) - [Deloitte, *Belgian Scale-Ups Confidence Survey 2025*](https://www2.deloitte.com/be) - [Acerta, *AI-gebruik in Belgische bedrijven februari 2025*](https://www.acerta.be) ### Verder lezen op Falora - [Self-reported attribution: the only B2B attribution that survives](https://falora.ai/blog/self-reported-attribution-b2b) - [Revenue attribution vs marketing attribution](https://falora.ai/blog/revenue-attribution-vs-marketing-attribution) - [The Marketing Efficiency Ratio (MER)](https://falora.ai/blog/marketing-efficiency-ratio) - [GEO voor B2B SaaS](https://falora.ai/nl/blog/geo-voor-b2b-saas) - [Schaalbare B2B groei zonder bureau](https://falora.ai/nl/blog/schaalbare-b2b-groei-zonder-bureau) ### Over de auteur Stijn Van Daele is co-founder van Falora en partner bij Stretch Innovation. Hij coördineerde deze benchmark met de Stretch portfolio en Falora klanten. Hij schrijft over GTM engineering, autonome revenue en measurement op [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **Wat is de attributiekloof?** De attributiekloof is het verschil tussen wat je multi-touch attributie-tool (Dreamdata, HockeyStack, GA4) meet en wat je kopers zelf melden als ze antwoorden op de vraag Hoe hoorde je van ons. In onze Benelux benchmark van 22 B2B SaaS-bedrijven is de mediane kloof 63 procent; tracking ziet 37 procent van de attributie die kopers rapporteren. **Welke kanalen zitten in de attributiekloof?** Drie kanalen samen goed voor 78 procent van de gemiste attributie: peer referral en word-of-mouth (28 procent), dark social (LinkedIn DMs, Slack communities, podcasts, 26 procent), en AI engine discovery via ChatGPT, Claude en Perplexity (24 procent, was minder dan 3 procent in 2024). Deze drie kanalen produceren geen tracking data maar wel omzet. **Hoe groot is de attributiekloof in Belgische versus Nederlandse SaaS?** In onze benchmark is de kloof marginaal groter in België (mediaan 67 procent) dan in Nederland (mediaan 60 procent). Het verschil komt van sterkere community-cultuur in Belgische scale-ups (Beltug, VOKA, Agoria netwerken) en van tragere ontwikkeling van tracking-fluentie. De kloof groeit in beide landen kwartaal na kwartaal. **Wat kunnen BE en NL scale-ups doen aan de attributiekloof?** Drie stappen. Eerst: implementeer self-reported attributie op elke high-intent form (verplicht free-text veld). Tweede: bouw een reconciliation dashboard dat tracked en self-reported naast elkaar toont per kwartaal. Derde: verschuif budget geleidelijk richting de kanalen die self-reported groot rapporteert maar tracked klein rapporteert. De volledige playbook staat in ons self-reported attributie artikel. **Hoe werd deze benchmark opgesteld?** 22 Belgische en Nederlandse B2B SaaS scale-ups tussen €3M en €30M ARR deelden geanonimiseerde attributie-data over Q1 en Q2 2026. Voor elk bedrijf vergeleken we de output van hun multi-touch tool (voornamelijk Dreamdata, HockeyStack, Factors.ai en GA4) met de gecategoriseerde antwoorden op verplichte self-reported velden op hun demo-formulieren. Sample-omvang klein maar representatief; groter benchmark-onderzoek loopt door in H2 2026. --- # Marketing Efficiency Ratio (MER): the metric that survives dark social - URL: https://falora.ai/blog/marketing-efficiency-ratio - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-07-24 - Category: Insights - Tags: mer, marketing-efficiency, cfo-metrics, measurement - Reading time: 12 min - Translation key: marketing-efficiency-ratio ## TL;DR The Marketing Efficiency Ratio (MER) is total revenue divided by total marketing spend over a period. It sidesteps attribution entirely and answers the question every CFO cares about: for every euro of marketing spend, how many euros of revenue did we generate. In 2026 it is the metric that has aged best because it does not depend on the tracking that broke. ## Description Marketing Efficiency Ratio (MER) equals total revenue divided by total marketing spend. It is the honest metric when attribution is broken. The 2026 CFO field guide. ## Full content ## TL;DR > The Marketing Efficiency Ratio equals total revenue divided by total marketing spend. It ignores attribution entirely and answers the one question every CFO actually asks: for every euro of marketing, how many euros of revenue. - MER benchmarks for B2B SaaS: Series A 2-4x, Series B 3-6x, Enterprise 4-10x. - MER survives dark social because it does not depend on tracking. Total revenue is knowable; total marketing spend is knowable; the ratio is honest. - Use MER for top-down efficiency; use multi-touch attribution for channel optimisation; use self-reported attribution for buyer-behaviour insight. - Below 2x at any stage suggests marketing inefficiency; above 10x suggests marketing underinvestment. ## Introduction Chris Walker's argument, delivered on LinkedIn hundreds of times over five years, resolves to a single practical question: if attribution is broken, what should the CFO and the CMO agree to measure? The answer that has quietly become the consensus in mid-market B2B SaaS is the Marketing Efficiency Ratio (MER). Total revenue divided by total marketing spend over a defined period. Blind to attribution. Blind to channel. Blind to the dark-social conversation that no tracker sees. Honest about the aggregate efficiency of the marketing function. This article is the field guide for B2B CMO, CFO or head of finance who wants to add MER to their measurement stack in 2026. It covers the formula, the benchmarks, the pitfalls, and the reconciliation with the other measurement tools already in use. ## The formula and what it includes MER = Total revenue / Total marketing spend, over a defined period. Total revenue is straightforward at the company level: recognised revenue over the period, or new-business ARR added, or a specific segment of both. Which specific revenue definition you use matters and needs to be documented explicitly. Total marketing spend is where teams disagree. The honest, defensible definition includes everything the marketing function consumes: - Paid media (Google Ads, LinkedIn Ads, Meta Ads, ABM display, sponsored content) - Marketing team compensation (salaries, bonuses, contractors) - Marketing tools (marketing automation, attribution tools, CMS, design tools, video production) - Events (sponsorships, hosted events, trade shows) - Content production (agency, freelancer, video, design) - Brand and PR (agency, sponsorships, brand campaigns) - Community and dark-social investment (podcast production, community management) Some teams argue for excluding brand spend from MER on the theory that brand has long-tail effects that a single period cannot capture. This is defensible but adds complexity. The simpler, more auditable approach is to include everything and use a trailing 12-month window to smooth out the timing lag. The formula is deliberately blunt. Its virtue is that both inputs are knowable at the CFO level and the ratio is comparable across periods and across companies. ## Why MER survived dark social Three properties make MER robust to the measurement environment of 2026. **It does not depend on tracking.** MER does not care whether the buyer came from Google, LinkedIn, a peer conversation, a ChatGPT recommendation or a podcast clip. Total revenue happened; total spend happened; the ratio is calculable. **It does not depend on attribution model choice.** Multi-touch, U-shaped, W-shaped, data-driven attribution are all methodological choices with defensible cases for and against. MER makes no such choice. The CFO cannot argue with a ratio that has no model behind it. **It aggregates the invisible.** Dark social, AI-engine discovery, peer referral, community-driven awareness all show up in MER as revenue that the marketing function contributed to producing, even when the tracker cannot see the specific touchpoint. As Chris Walker argues: > "Sales metrics like MQLs and Stage 1 pipeline are grossly misaligned to the actual goal of demand generation." MER is misaligned with nothing except attribution. It measures what the CFO cares about: aggregate efficiency of marketing investment. ## The benchmark ranges Benchmarks vary by stage, motion and category. The ranges below are what we observe in the Stretch Innovation portfolio of B2B SaaS scale-ups, cross-referenced against publicly-available data from ChartMogul, Bessemer Cloud 100 and Kyle Poyar's *Growth Unhinged* benchmarks. | Stage | ARR range | Healthy MER | Warning below | Under-investment above | |---|---|---|---|---| | Pre-Seed / Seed | Under €1M ARR | 1.5 - 3x | 1x | Not typically relevant | | Series A | €1M - €5M ARR | 2 - 4x | 1.5x | 6x | | Series B | €5M - €20M ARR | 3 - 6x | 2x | 8x | | Growth stage | €20M - €50M ARR | 3 - 7x | 2.5x | 9x | | Enterprise | €50M+ ARR | 4 - 10x | 3x | 12x | Two boundary interpretations matter. **Below the healthy range.** The marketing function is inefficient (spending too much for the revenue produced) or the spend definition is wrong (including costs that belong elsewhere) or the revenue definition is wrong (excluding legitimate marketing-driven revenue). Diagnose in that order. **Above the healthy range.** The marketing function is under-invested. Growth is being left on the table because the marketing team is under-resourced or the spend budget is too small for the addressable market. Common failure mode: bootstrapped companies at MER 15x that could have doubled ARR growth with a doubled marketing budget. ## MER versus ROAS versus attribution The three metrics answer different questions and belong at different altitudes. **ROAS (Return on Ad Spend).** Channel-level. Revenue per euro of paid-media spend on a specific channel, attributed by a tracking model. Right for weekly campaign optimisation. Wrong for board reporting. Blind to non-paid marketing investment. **Multi-touch attribution.** Program-level. Revenue attributed to touchpoints across the funnel. Right for optimisation across channels and program design. Wrong for measuring the aggregate efficiency of the marketing function. Broken by dark social. **MER.** Function-level. Revenue over total marketing spend. Right for CFO reporting, board reporting, and quarterly strategic decisions. Wrong for channel optimisation. Blind to which specific investments drove the outcome. The mature 2026 measurement stack runs all three. ROAS in the marketing team's daily tool. Multi-touch attribution in the RevOps dashboard. MER in the CFO's monthly financial review. Each is used for what it can defensibly measure. ## The reconciliation between MER and revenue attribution Some CMOs worry that MER and revenue attribution will disagree in a way that undermines their credibility. The concern is understandable but wrong. The two measure different things at different altitudes and their disagreement is informative rather than problematic. If MER shows healthy efficiency (say 4x at Series B) and revenue attribution shows 22% of closed-won credited to marketing, both can be true. The marketing function is efficient at its overall contribution (MER); the identifiable and attributable marketing-driven revenue is a portion of the total. The remaining revenue was influenced by marketing (brand, community, dark social) in ways that revenue attribution cannot fully credit but that MER captures aggregately. The productive quarterly conversation is: does the pattern of MER improvement match the pattern of revenue attribution improvement? If MER is trending up and revenue attribution is flat, marketing is likely contributing to sales-attributed revenue through brand and awareness effects. If MER is flat and revenue attribution is up, marketing is capturing more of what was already there rather than creating new demand. Adam Robinson's frame applies: > "Most marketers would say I am an idiot, but I do not believe in attribution. It forces you to focus on the wrong things." MER is the metric you focus on when attribution has focused you on the wrong things. Not a replacement; a complement that keeps the strategic conversation grounded in aggregate efficiency. ## Common calculation pitfalls Three specific mistakes appear often in mid-market MER implementations. **Pitfall one: excluding marketing team compensation.** MER minus salaries flatters the ratio by 30 to 60 percent. Fine as an internal metric with clear labelling. Wrong for CFO reporting. **Pitfall two: using monthly windows.** B2B SaaS sales cycles of 3 to 18 months mean monthly MER is noisy. Trailing 12-month or rolling 4-quarter windows smooth out the noise and are defensible. **Pitfall three: using GAAP revenue for a growth-mode business.** For a scale-up whose revenue is growing 40+ percent per year, GAAP revenue lags marketing investment by 6 to 12 months. Some teams use New ARR added in the period as the numerator instead, which aligns timing better but produces a lower absolute number. Document the choice. ## New-business MER versus expansion MER For B2B SaaS teams with meaningful expansion revenue (over 15% of new ARR from existing customers), splitting MER into two ratios adds clarity. **New-business MER** = New ARR from new customers / marketing spend attributed to acquisition. **Expansion MER** = Expansion ARR from existing customers / marketing spend attributed to customer marketing. The split matters because the two motions have very different economics. New-business MER at 3x is healthy; expansion MER at 8x is healthy. Aggregating hides both signals. Most B2B SaaS teams under €10M ARR do not need the split (expansion revenue is small enough that aggregate MER is directionally correct). Above €20M ARR the split becomes valuable and above €50M ARR essential. ## What CFOs should ask For a CFO evaluating whether their marketing function is efficient, five questions surface the MER conversation. **One: what is our trailing 12-month MER?** If nobody knows, MER is not being tracked; that is the first workstream. **Two: how does our MER compare to public benchmarks for our stage?** ChartMogul, Bessemer Cloud 100 and Growth Unhinged publish credible ranges. **Three: what does our MER trend look like over the last 8 quarters?** Direction matters more than absolute level. A MER improving from 2.5 to 3.5 over 8 quarters is a healthier signal than a stable MER of 4. **Four: what is included and excluded from our marketing spend definition?** Verify that salaries, tools, events and brand are all in. Any exclusion needs a defensible reason. **Five: how does our MER reconcile with our revenue attribution?** The gap is informative. A MER of 5 with revenue attribution showing 15% marketing contribution suggests brand and dark-social effects that attribution misses. That is normal and does not undermine either metric. ## Frequently asked questions **Is MER really that simple?** Yes. The formula is deliberately blunt. The interpretation and the reconciliation with other metrics are where the sophistication sits. **Does MER work for enterprise B2B with 18-month sales cycles?** Yes, with trailing 12-month or 4-quarter windows. Do not calculate MER on a single-month basis for enterprise motions. **Should I benchmark my MER against my competitors?** Only against public data. Private-company MER benchmarks are unreliable because inclusion rules vary. ChartMogul, Bessemer and Growth Unhinged publish defensible ranges. **Can Falora improve my MER?** Yes indirectly. Falora reduces marketing operating cost (automating the work of 1 to 2 marketing FTEs) while increasing qualified meeting volume, which improves both the numerator and the denominator of the ratio. **Does MER apply to product-led growth motions?** Yes, with modification. In PLG the marketing function drives free-tier signups and activation, which convert to paid revenue on their own timeline. MER for PLG should use trailing 12-month windows and consider self-serve revenue separately from sales-assisted revenue. ## Conclusion The Marketing Efficiency Ratio is the metric that aged best over the last five years of B2B measurement disruption. It sidesteps the attribution debate. It sidesteps the tracking-versus-privacy trade-offs. It answers the question the CFO actually cares about with a formula both sides can audit. The mature B2B SaaS measurement stack in 2026 does not choose between MER and attribution. It runs both. MER for top-down efficiency and board-level conversations. Multi-touch and self-reported attribution for tactical optimisation and buyer-behaviour insight. The reconciliation between them is a productive monthly conversation rather than an unresolvable argument. If you want help implementing MER, revenue attribution and self-reported attribution as a coherent stack, **[book a 45-minute measurement review with Falora](https://falora.ai/demo)**. --- ### Sources - [Chris Walker on LinkedIn](https://www.linkedin.com/in/chriswalker171) - [Kyle Poyar, *Growth Unhinged benchmarks*](https://www.growthunhinged.com) - [Bessemer Venture Partners, *Cloud 100 2025*](https://www.bvp.com/atlas) - [ChartMogul, *SaaS Growth Report 2025*](https://chartmogul.com/reports) - [Adam Robinson, RB2B](https://www.rb2b.com) - [Toni Hohlbein, *Revenue Formula*](https://www.revenueformula.io) - [Dreamdata, *B2B attribution and MER guidance*](https://dreamdata.io) - [HockeyStack, *MER methodology guide*](https://hockeystack.com) ### Related reading on Falora - [Revenue attribution vs marketing attribution](https://falora.ai/blog/revenue-attribution-vs-marketing-attribution) - [Self-reported attribution: the only B2B attribution that survives](https://falora.ai/blog/self-reported-attribution-b2b) - [Supermetrics alternative: the 8 options that actually work](https://falora.ai/blog/supermetrics-alternative) - [Looker Studio alternative: the 8 BI tools B2B teams should evaluate](https://falora.ai/blog/looker-studio-alternative) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. He has implemented MER as a board-level metric for 18+ B2B SaaS scale-ups and writes about GTM engineering, autonomous revenue and measurement on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What is the Marketing Efficiency Ratio (MER)?** MER equals total revenue divided by total marketing spend over a defined period. It is a top-down efficiency metric that measures how much revenue every euro of marketing spend generated. Unlike ROAS or multi-touch attribution, MER does not attribute revenue to specific channels; it treats the marketing function as a single black-box and measures the aggregate output. **What is a good MER for B2B SaaS in 2026?** Benchmarks vary by stage and motion. Series A (€1M-€5M ARR): MER of 2 to 4 is typical. Series B (€5M-€20M ARR): MER of 3 to 6 is healthy. Enterprise (€20M+ ARR): MER of 4 to 10. Below 2 at any stage suggests marketing spend is inefficient or wrongly attributed to marketing versus sales. Above 10 typically means marketing is under-invested and growth is being left on the table. **How is MER different from ROAS?** ROAS (Return on Ad Spend) is a channel-level metric measuring revenue per euro of paid-media spend, attributed by a tracking model. MER is a company-level metric measuring revenue per euro of total marketing spend (including salaries, tools, content, events, brand). ROAS optimises channels; MER measures the marketing function. Both are useful; MER survives when tracking breaks and ROAS does not. **Should I use MER instead of multi-touch attribution?** Alongside, not instead of. Use MER for top-down efficiency measurement and CFO conversations. Use multi-touch attribution for tactical channel optimisation. Use self-reported attribution for buyer-behaviour insight. The three answer different questions; the mature B2B measurement stack in 2026 runs all three and reconciles them. **How do I calculate MER for a B2B SaaS business with long sales cycles?** Use a rolling 4-quarter or trailing 12-month window rather than a single month. B2B cycles of 6 to 18 months mean monthly MER is noisy. The trailing calculation smooths out the noise and gives a defensible number for board reporting. Some teams also split MER into new-business MER and expansion MER to separate the two motions. --- # Revenue attribution vs marketing attribution: what CMOs need in 2026 - URL: https://falora.ai/blog/revenue-attribution-vs-marketing-attribution - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-07-17 - Category: Insights - Tags: attribution, revenue-attribution, marketing-measurement, cmo - Reading time: 13 min - Translation key: revenue-vs-marketing-attribution ## TL;DR Marketing attribution measures how a click, form-fill or MQL happened. Revenue attribution measures how a closed-won deal or renewal happened. In 2026 the two answer different questions, need different data pipelines, and belong to different owners in the org. ## Description Marketing attribution counts touchpoints. Revenue attribution counts revenue. The difference matters more than the tooling. A CMO's field guide for 2026. ## Full content ## TL;DR > Marketing attribution answers *how did we get the click*. Revenue attribution answers *how did we get the dollar*. In 2026 those are different questions with different data pipelines and different owners. - Marketing attribution: credit for a marketing outcome (click, MQL, opportunity) to touchpoints. - Revenue attribution: credit for a revenue outcome (closed-won, renewal, expansion) across marketing, sales and buyer-reported channels. - 70% of B2B buyer journeys are anonymous (6sense, 2025); marketing attribution measures the visible 30%; revenue attribution triangulates the whole. - The two belong to different owners: marketing attribution to marketing ops; revenue attribution to RevOps or the CFO. ## Introduction There are two arguments people can be having when they say "attribution is broken". The first is about *marketing* attribution: multi-touch models, tracking parameters, last-click versus U-shaped versus data-driven weighting. The second is about *revenue* attribution: how a closed-won deal was actually caused, from initial awareness to procurement sign-off. These are different arguments. They need different tools, different owners and different levels of executive attention. Conflating them is one of the persistent sources of miscommunication between CMO and CFO in B2B SaaS. This article is for the CMO, RevOps lead or head of finance who needs to separate the two, decide which one to invest in first, and know who owns what. Written for mid-market B2B SaaS in 2026, where the underlying buyer behaviour and the tooling landscape have both shifted enough that the old answers do not hold. ## The definition split **Marketing attribution** measures how a marketing outcome happened. Which channel, which campaign, which touchpoint drove the click, the form fill, the MQL, the pipeline creation? The answer feeds channel-level and campaign-level budget decisions. **Revenue attribution** measures how a revenue outcome happened. Which combination of marketing, sales, product and self-reported buyer inputs drove the closed-won deal, the renewal, the expansion? The answer feeds strategic-level and executive-level budget decisions. Both are real disciplines. Both matter. They are not the same discipline and they should not report through the same owner. ## Why the difference matters more in 2026 Two shifts over the last three years have widened the gap between the two. The first shift is buyer behaviour. 6sense's *B2B Buyer Experience Report 2025*: 70% of the buyer journey is anonymous, 81% of buyers have decided before contacting sales, 84% of deals are won by the first vendor contacted. Marketing attribution can only see the tracked portion of the journey. When the tracked portion is 30% of the whole, marketing attribution is measuring a shrinking minority. The second shift is discovery. G2's 2026 buyer report: 51% of software buyers start research in AI engines (ChatGPT, Claude, Perplexity, Gemini), often more than in Google. AI engine referrals often arrive as "direct" or "unknown" in analytics tools. Marketing attribution mis-attributes; revenue attribution needs to reconstruct. The two shifts together mean that in 2026, marketing attribution tells a story about a shrinking, distorted subset of the buyer journey. Revenue attribution is the discipline that patches the gap. As Chris Walker has argued for four years: > "Sales metrics like MQLs and Stage 1 pipeline are grossly misaligned to the actual goal of demand generation." Marketing attribution optimises for MQLs and Stage 1 pipeline. Revenue attribution optimises for revenue. In 2026 those are frequently different optimisations. ## The data pipelines are different Marketing attribution runs on a specific data pipeline: web analytics tags, campaign UTMs, form-fill tracking, ad platform pixels, marketing automation events. The tools (Dreamdata, HockeyStack, Factors.ai, GA4, Adobe Analytics) consume this pipeline and produce touchpoint-weighted models. Revenue attribution runs on a broader data pipeline. It consumes the marketing attribution pipeline as one input, then adds three more. The second input is **CRM sales activity**. Opportunity creation, calls, meetings, stage progression, deal size, close date, win-loss reason. Salesforce or HubSpot are the sources of truth. This data is not visible to marketing attribution tools. The third input is **self-reported buyer data**. The *How did you hear about us* field on high-intent forms, the discovery-call question capture, the win-loss interview. This data captures the dark-social and AI-engine touchpoints that no tracking sees. We covered the implementation in [our self-reported attribution playbook](https://falora.ai/blog/self-reported-attribution-b2b). The fourth input is **product data** (for PLG motions). Free-tier signup, activation events, feature usage, expansion signals. This data feeds both marketing attribution (product-qualified leads) and revenue attribution (activation-to-close velocity, product-driven expansion). The four inputs join in a data warehouse. The warehouse becomes the source of truth for revenue attribution, and the marketing attribution tool becomes one of several inputs rather than the primary answer. ## Who owns each Getting the ownership right is where most orgs fail. **Marketing attribution belongs to marketing operations.** The output feeds channel-level and campaign-level optimisation. The frequency of use is weekly or bi-weekly. The stakeholder is the marketing team; the CMO consumes the summary for tactical decisions. **Revenue attribution belongs to RevOps or the CFO.** The output feeds strategic budget allocation and pipeline forecasting. The frequency of use is monthly or quarterly. The stakeholder is the executive team; the CFO uses it for finance decisions and board reporting. Common failure modes when ownership is confused: - Asking the marketing team to defend revenue attribution numbers in a QBR. They will defend the touchpoint-weighted model that flatters marketing's contribution, which the CFO will not believe. - Asking RevOps to optimise the paid-media campaign mix. They will apply revenue-attribution logic to a decision that needs channel-level granularity, producing over-conservative recommendations. - Building one dashboard that tries to serve both. The dashboard will satisfy neither audience and will spawn parallel spreadsheets in both teams. The clean architecture has the two functions as separate dashboards, separate owners, separate data pipelines that share a warehouse. The reconciliation between them (why is marketing attribution showing 45% paid contribution and revenue attribution showing 22%?) becomes a productive quarterly conversation rather than a permanent argument. ## The tooling map No single tool does end-to-end revenue attribution well in 2026. The category that markets itself as "revenue attribution" is heterogeneous. **Marketing attribution tools:** Dreamdata, HockeyStack, Factors.ai, NorthBeam, plus GA4 and Adobe Analytics for the base layer. These consume the marketing pipeline and produce touchpoint-weighted models. Best-in-class options in 2026 include AI-assisted touchpoint suggestion. **CRM as revenue source of truth:** Salesforce, HubSpot, Pipedrive. The opportunity object, the activity feed, the stage history. The revenue side of the equation lives here. **Self-reported attribution capture:** required free-text field on forms, discovery-call question capture in Gong or Chorus, win-loss interview data. No single vendor owns this category; it is an operational discipline. **Warehouse and modelling:** Snowflake, BigQuery, Databricks, or Redshift as the warehouse; dbt for transformation; the BI tool of choice for consumption. This is where the four inputs join into the revenue attribution model. **Reconciliation layer:** custom-built in most mature deployments. Some teams use Hex or Mode notebooks for the reconciliation; others build it as views in the warehouse consumed by the BI tool. Adam Robinson of RB2B captures the honest limit: > "Most marketers would say I am an idiot, but I do not believe in attribution. It forces you to focus on the wrong things." The right way to hear the argument in 2026 is: single-model attribution (marketing OR revenue) forces you to focus on the wrong things. The mature answer is to run both as complementary models, use each for what it can defensibly measure, and be honest about the gap between them. ## What the gap typically shows Across the deployments we have observed in the Stretch Innovation portfolio, three patterns recur when marketing attribution and revenue attribution are compared. **Marketing attribution overweights paid channels.** Paid attribution is trackable and tags cleanly. Revenue attribution typically shows 20 to 40 percent less credit to paid than marketing attribution assigns. **Marketing attribution underweights peer referral and community.** Both channels are invisible to tracking. Revenue attribution typically shows peer and community driving 15 to 30 percent of closed-won that marketing attribution missed entirely. **Marketing attribution misses AI-engine discovery.** In Q1 and Q2 2026 deployments we measured, self-reported attribution shows ChatGPT, Claude and Perplexity driving 8 to 22% of high-intent conversions. Marketing attribution shows them at 0 to 3%. These gaps are not attribution-tool bugs. They are structural. The tracked-versus-untracked ratio in modern B2B buying is 30/70; marketing attribution measures the 30, revenue attribution triangulates the whole. ## The implementation sequence If you are starting from marketing attribution only, and want to add revenue attribution capability, the sequence that works in the Stretch portfolio is this. **Month 1: warehouse and CRM cleanup.** Get opportunity data flowing cleanly into a warehouse. Standardise the opportunity source and win-loss fields in the CRM. Without clean revenue data, revenue attribution has no target variable. **Month 2: self-reported attribution deployment.** Add the required free-text field to high-intent forms. Capture the response in the CRM as a first-class field. Start the categorisation pipeline. Full playbook in [our self-reported attribution piece](https://falora.ai/blog/self-reported-attribution-b2b). **Month 3: reconciliation model.** Join marketing attribution touchpoint data, CRM activity data and self-reported data at the opportunity level. Build a first-cut revenue attribution model in the warehouse. Compare with marketing attribution output. Document the gap. **Month 4: ownership assignment.** Assign marketing attribution to marketing ops as their primary optimisation dashboard. Assign revenue attribution to RevOps or the CFO as their primary strategic dashboard. Schedule a monthly reconciliation review. **Ongoing: iterate on the model.** Revenue attribution is not solved once; the model needs refinement as buyer behaviour shifts, as new channels emerge, as self-reported data reveals unexpected patterns. ## What CMOs should ask their team Five questions for a CMO to ask their marketing ops and RevOps teams to surface whether revenue attribution capability exists. **One: what percentage of closed-won revenue last quarter has a self-reported attribution answer captured in the CRM?** If under 30%, self-reported infrastructure is not yet in place. **Two: what is the gap between our marketing attribution paid credit and our self-reported paid credit?** If nobody has calculated it, revenue attribution is not being run. **Three: who signs off the annual budget allocation between paid, content, event, community and product-led investment?** If nobody, that is the person who needs revenue attribution. **Four: what would we do differently next quarter if we knew our real closed-won attribution?** If the honest answer is "we would move budget from paid to community by 8 to 15 points", that is the value of revenue attribution. **Five: are we measuring the same thing in marketing attribution and revenue attribution?** Almost always no. That is fine and expected. Both matter. ## Frequently asked questions **Can I get revenue attribution from a single vendor?** No, and vendors that claim otherwise are usually selling marketing attribution with a re-label. Revenue attribution is a modelling discipline that composes marketing attribution, CRM sales data, self-reported data and often product data. **Does Falora do revenue attribution?** Falora captures self-reported attribution data at the form and conversation layer and writes it to your warehouse for reconciliation. We are one input to your revenue attribution model, not the whole model. **Is single-touch attribution ever the right choice?** For very early-stage teams with limited data, first-touch or last-touch attribution provides directional signal at low overhead. Above €3M ARR the gap between single-touch and revenue attribution is large enough to matter. **How does this interact with the Marketing Efficiency Ratio?** Marketing Efficiency Ratio (MER) is a top-line spend efficiency metric that bypasses attribution entirely (total revenue over total marketing spend). It is the honest metric when attribution is contested; we cover it in [our MER piece](https://falora.ai/blog/marketing-efficiency-ratio). **How often should revenue attribution be re-modelled?** Quarterly at minimum. The buyer behaviour shifts fast enough in 2026 that a model built in Q1 will materially misread by Q4 if not refreshed. ## Conclusion Marketing attribution is not dead. It is a specific tool for a specific job: campaign-level and channel-level optimisation, owned by marketing ops. Revenue attribution is a different discipline for a different job: strategic budget allocation and executive-level insight, owned by RevOps or the CFO. The mature B2B SaaS org in 2026 runs both, owns them separately, and treats the reconciliation between them as a productive quarterly conversation. The team that has only marketing attribution is making strategic decisions on partial data. The team that has only revenue attribution is missing the tactical optimisation loop. If you want to walk through the implementation sequence for your specific stack, **[book a 45-minute measurement review with Falora](https://falora.ai/demo)**. --- ### Sources - [6sense, *B2B Buyer Experience Report 2025*](https://6sense.com/resources) - [G2, *2026 Software Buyer Behavior Report*](https://www.g2.com/research) - [Chris Walker on LinkedIn](https://www.linkedin.com/in/chriswalker171) - [Adam Robinson, RB2B](https://www.rb2b.com) - [Dreamdata, *Revenue attribution playbook*](https://dreamdata.io) - [HockeyStack, *B2B attribution guide*](https://hockeystack.com) - [Factors.ai, *Attribution methodology*](https://www.factors.ai) - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) ### Related reading on Falora - [Self-reported attribution: the only B2B attribution that survives](https://falora.ai/blog/self-reported-attribution-b2b) - [The Marketing Efficiency Ratio: the metric that survives dark social](https://falora.ai/blog/marketing-efficiency-ratio) - [GEO for B2B SaaS: how to get cited by ChatGPT](https://falora.ai/blog/geo-for-b2b-saas) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. He has architected marketing and revenue attribution models across 18+ B2B SaaS scale-ups and writes about GTM engineering, autonomous revenue and measurement on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What is the difference between revenue attribution and marketing attribution?** Marketing attribution assigns credit for a marketing outcome (click, MQL, opportunity) to touchpoints. Revenue attribution assigns credit for a revenue outcome (closed-won deal, renewal, expansion) to touchpoints, marketing programs, sales activities, and often self-reported buyer feedback. Marketing attribution is a subset of what revenue attribution measures. **Do I need both revenue attribution and marketing attribution?** Most mid-market B2B SaaS teams need both, at different owners. Marketing attribution belongs to the marketing operations team for campaign optimisation and channel budget. Revenue attribution belongs to RevOps or the CFO for strategic budget allocation and pipeline forecasting. Confusing them (asking marketing to defend revenue attribution or asking RevOps to optimise channels) produces poor decisions. **How is revenue attribution measured in 2026?** Three data pipelines feed a working revenue attribution model. First, tracked marketing touchpoints from a multi-touch tool (Dreamdata, HockeyStack, Factors.ai). Second, sales activity data from the CRM (calls, meetings, opportunity progression). Third, self-reported buyer feedback captured on high-intent forms and in the first sales call. The three are joined at the opportunity or account level and reconciled. **Why are marketing attribution and revenue attribution different?** Because 70% of the B2B buyer journey is anonymous (6sense, 2025). Marketing attribution only sees tracked touchpoints; it cannot see the dark-social conversation, the peer referral, the ChatGPT recommendation. Revenue attribution triangulates tracked data with self-reported data and sales activity to reconstruct the actual journey. The results diverge, often dramatically. **Which tools do revenue attribution in 2026?** No single tool does it end-to-end. The typical stack: a marketing attribution tool (Dreamdata, HockeyStack, Factors.ai) for tracked touchpoints; a CRM as the source of truth for sales activity and opportunity data; self-reported attribution captured on forms and in Gong or similar call-intelligence tools; a data warehouse to join and reconcile. Revenue attribution is a modelling discipline, not a product category. --- # Looker Studio alternative: the 8 BI tools B2B teams should evaluate - URL: https://falora.ai/blog/looker-studio-alternative - Language: en - Author: Jeroen De Broyer (Co-founder, Falora) - Published: 2026-07-10 - Category: How-to - Tags: looker-studio, bi-tools, vendor-comparison, analytics - Reading time: 13 min - Translation key: looker-studio-alternative ## TL;DR Looker Studio wins on price (free) and loses on performance, permissioning and warehouse-native features. The right 2026 alternative depends on your data warehouse, your self-serve ambition and your budget. Metabase, Sigma, Preset and Hex are the alternatives that earn shortlist status for mid-market B2B. ## Description Looking for a Looker Studio alternative? The 8 BI tools B2B teams evaluate in 2026, ranked by warehouse fit, self-serve capability, pricing and audit trail. ## Full content ## TL;DR > Looker Studio is free and functional. It stops being enough at the moment your warehouse becomes the source of truth, your team needs row-level permissioning, or your analyst needs to explore data instead of just visualise it. - 8 shortlist alternatives in 2026: Metabase, Preset (Superset), Sigma, Hex, Mode, Rows, Tableau, Power BI. - Pick by warehouse fit first, self-serve capability second, price third. - Warehouse-first teams should skip Looker Studio entirely and standardise on Metabase, Sigma or Preset. - Free is not free at scale: Looker Studio's hidden costs are analyst time and stakeholder frustration. ## Introduction Looker Studio (formerly Google Data Studio) is the most-deployed BI tool in the B2B SaaS mid-market for one honest reason: it is free, and it has native connectors to Google Analytics, Google Ads and Search Console. For a marketing team producing weekly performance reports on Google-centric data, Looker Studio is the fastest path to a visualisation. The reasons teams look for an alternative are also consistent. Performance degrades on large datasets. Row-level permissioning is weak. Governance and version control are missing. Warehouse-native workflows are second-class citizens. Self-serve exploration is limited compared to modern tools. Recent Google product-line consolidation has added uncertainty about the roadmap. This article is for the RevOps lead, head of data or GTM engineer at a B2B SaaS scale-up evaluating a BI tool switch. Not a rankings piece. A destination-first buying framework and the 8 tools that earn shortlist consideration. ## When Looker Studio is the right answer Before recommending alternatives, the honest case for staying. Stay on Looker Studio if your primary data sources are Google Analytics 4, Google Ads, YouTube Analytics and Search Console; your team is under 5 people; your dashboards are consumed by a small stakeholder group; your queries return under 100K rows; and you do not need per-user or per-row access controls. In this profile Looker Studio is often the right choice, even in 2026. Move off Looker Studio if any of these apply: your source of truth is a data warehouse; you need row-level or column-level permissioning; your team includes analysts who want to explore data interactively; your organisation needs governance, versioning or audit trails; performance on your typical dataset is slow. ## The 8 alternatives, at a glance | Tool | Best for | Model | Starting price | Warehouse-native | |---|---|---|---|---| | Metabase | Warehouse teams needing self-serve | OSS + Cloud | Free (OSS) or $85/mo (Cloud) | Yes | | Preset (Superset) | Open-source purists at scale | Managed OSS | $20 per user/mo | Yes | | Sigma Computing | Spreadsheet-native business users | Cloud | $600 per user/year | Yes | | Hex | Analyst notebooks + SQL + Python | Cloud | $75 per user/mo | Yes | | Mode | SQL-first analyst teams | Cloud | $500 per user/year | Yes | | Rows | Spreadsheet-first modern teams | Cloud | Free tier, $59+/mo | Adequate | | Tableau | Enterprise Salesforce shops | Cloud/Server | $75 per user/mo | Yes | | Power BI | Enterprise Microsoft shops | Cloud/Server | $10 per user/mo | Yes | Pricing is Q2 2026 published tier; enterprise contracts vary widely. ## Metabase The most-recommended Looker Studio alternative for B2B SaaS mid-market in 2026. Two features drive the choice. The first is **warehouse-native self-serve**. Metabase's query builder lets non-SQL users construct queries against Snowflake, BigQuery, Databricks, Redshift and 20+ other databases without writing SQL. When SQL is needed, the editor is competent. The second is **deployment flexibility**. Self-host Metabase for free on your own infrastructure. Use Metabase Cloud for hosted convenience starting at $85 per month. Both preserve the same feature set. Weakness: embedded analytics work but are less polished than commercial alternatives. Governance features exist but require configuration. Right for: B2B SaaS scale-ups with a warehouse, 5+ analyst-adjacent users, and an internal preference for open-source stacks. ## Preset (Apache Superset) Superset is the open-source BI tool developed at Airbnb, now maintained as an Apache project. Preset is the managed hosting company founded by the Superset creator. Strong on: dashboard variety, SQL sophistication, ecosystem maturity. The visualisation library is deep. The query performance on large datasets is excellent when connected to a modern warehouse. Weakness: UX is more analyst-oriented than business-user-oriented. Non-technical stakeholders often prefer Metabase or Sigma. Right for: teams that value open-source strategy and have data or analytics engineering in-house. ## Sigma Computing The BI tool that made spreadsheet-native business users a first-class audience. Sigma runs a spreadsheet UI on top of your warehouse; the underlying queries are SQL but the interface is Excel-adjacent. Strong on: business-user self-serve, warehouse performance, live spreadsheet workflows where changes trigger downstream queries. Weakness: pricing is enterprise-scale ($600 per user per year). The value math works when 10+ business users actively use Sigma as their primary analytics interface. Right for: RevOps or finance teams with 10+ business-user analysts, a warehouse commitment, and a budget line for premium BI. ## Hex Analyst-notebook tool that combines SQL, Python and visualisation in a versioned document format. Positioned between BI (Metabase, Sigma) and notebooks (Jupyter, Databricks notebooks). Strong on: analyst productivity, reproducibility, collaboration across data science and analytics roles. Weakness: not primarily a dashboard-distribution tool. Consumption is different from Metabase or Sigma; you share notebooks and apps, not dashboards per se. Right for: teams with dedicated data analysts who want a productive workbench and can distribute findings through purpose-built apps rather than always-on dashboards. ## Mode SQL-first analyst tool with report distribution. Similar niche to Hex but with a longer track record and a more traditional report format. Strong on: SQL fluency, report scheduling, embedded analytics. Recent product direction has emphasised generative AI features for analyst productivity. Weakness: business-user self-serve is weaker than Metabase or Sigma. Best for analyst-owned reporting rather than department-wide self-serve. ## Rows Modern spreadsheet with BI-adjacent features. Native connectors to marketing and product data sources, formula-driven analysis, sharing model that matches how business teams already work. Strong on: fast time-to-first-report, spreadsheet familiarity, generous free tier. Weakness: not a warehouse-native tool. Scale ceiling is lower than the alternatives above. Right for: seed and early Series A teams that need marketing analytics faster than an analyst can build them. ## Tableau The enterprise incumbent. Deep visualisation capability, strong governance, wide ecosystem, high price. Strong on: enterprise features, established buyer confidence, Salesforce integration since the 2019 acquisition. Weakness: deployment complexity, learning curve, pricing. Rarely the right choice for a B2B SaaS scale-up under €30M ARR unless the parent organisation already runs Tableau enterprise-wide. Right for: enterprises with Salesforce ecosystem commitment and dedicated BI operations. ## Power BI Microsoft's enterprise BI. Excellent Microsoft ecosystem integration, extensive feature set, competitive per-user pricing. Strong on: Excel-adjacent workflows, Azure integration, low per-user cost. Weakness: best value depends on Microsoft ecosystem commitment. Standalone deployments miss half the value. Right for: enterprises on Microsoft 365 with Azure warehouses. ## The self-serve question, honestly The single most-overrated feature in BI marketing is "self-serve analytics." Every vendor claims it. Few deliver it in a way that matches how business teams actually work. Real self-serve in 2026 means: a business user can answer a novel question in under 10 minutes, without asking the RevOps team, and without producing a wrong number. Test this on your shortlist before signing. Give a marketing manager a specific question, sit down with them and Metabase for 20 minutes, watch what happens. Then repeat with Sigma. Then Looker Studio. The differences will surprise you. In our portfolio, Sigma delivers the strongest business-user self-serve, Metabase delivers the strongest analyst self-serve, Looker Studio delivers the weakest of both categories. ## The governance question Every BI tool ships with permissioning claims. Governance in practice means answering three questions: **Row-level security.** Can you show a EU sales manager only EU account rows, while the US manager sees only US rows, from the same underlying query? Metabase and Sigma yes. Looker Studio only with workarounds. **Audit trail.** Can you see who viewed which dashboard, ran which query, and when? Enterprise Metabase, Sigma, Tableau and Power BI yes. Free Looker Studio no. **Version control.** Can dashboard changes be reviewed like code changes before deployment? Hex yes. Mode yes. Others increasingly through Git integration. If your organisation is subject to audit requirements (SOC 2, ISO 27001, HIPAA-adjacent), the governance features move from nice-to-have to shortlist gate. ## What we use at Falora For transparency: Falora's internal analytics runs on Metabase on top of Snowflake. We evaluated Sigma and Hex; Metabase won on the balance of open-source flexibility, warehouse-native queries and the deployment model that matched our security posture. For our external product analytics (customer-facing dashboards embedded in the Falora product), we run a custom embedded implementation rather than any of the tools above, because the requirements are specific to the GTM domain we operate in. ## A 30-day evaluation plan For a B2B SaaS team on Looker Studio considering a switch, here is the 30-day plan we use in the Stretch portfolio. **Days 1 to 7: pain audit.** Document specifically what is failing in Looker Studio. Performance, permissioning, self-serve, governance. Rank by frequency of impact. Small pains often stay in the tool; big pains justify migration. **Days 8 to 14: shortlist.** Pick 2 or 3 alternatives based on your destination and team fluency. If you have a warehouse, Metabase and Sigma should be on the shortlist. If you value open-source, add Preset. **Days 15 to 22: pilot.** Reproduce your 3 most-used Looker Studio dashboards in the top choice from the shortlist. Time-to-build, performance, and business-user reception are the three metrics that matter. **Days 23 to 30: decision.** If the pilot cleared the pain audit and the total cost of ownership is defensible, commit. If it did not, either try the second shortlist option or accept that Looker Studio is still the right tool for your stage. ## Frequently asked questions **Is Looker Studio being deprecated?** Not that we know of. Google has consolidated the Looker Studio and Looker (enterprise) product lines and continues to invest. The uncertainty is more about roadmap priorities than product survival. **Can I use two BI tools together?** Yes, and many teams do. Looker Studio for Google-centric marketing reports; Metabase or Sigma for warehouse-centric internal analytics. Reasonable at mid-market and above. **What is the total cost of ownership for Metabase?** Self-hosted Metabase is functionally free but requires 5 to 15 hours per month of DevOps attention for a mid-market deployment. Metabase Cloud starts at $85 per month and scales with users. Total cost of ownership at mid-market is typically $300 to $1,500 per month all-in. **Does Falora integrate with any of these tools?** Falora writes to your warehouse and exposes standardised event data that any BI tool can consume. We do not have a preferred BI vendor. **Which is best for GDPR compliance?** Self-hosted Metabase or Preset give you the strongest data-residency control (you own the deployment). Sigma, Hex and Mode have EU data residency options. Looker Studio compliance depends on your Google Workspace configuration. ## Conclusion Looker Studio's free price and Google-native connectors made it the default B2B SaaS mid-market BI tool. Neither reason disappears in 2026. The reasons to leave are specific and worth acting on when they apply: warehouse-first architecture, permissioning needs, self-serve ambition, governance requirements. The right alternative depends on your destination and your team fluency. Metabase for warehouse plus analyst self-serve; Sigma for business-user spreadsheet workflows; Preset for open-source at scale; Hex or Mode for analyst notebooks. If you want help mapping BI tool selection against the broader [GTM engineering system architecture](https://falora.ai/blog/gtm-engineering-system-anatomy), **[book a review with Falora](https://falora.ai/demo)**. --- ### Sources - [Metabase, *Product and pricing*](https://www.metabase.com/pricing) - [Preset, *Managed Superset pricing*](https://preset.io/pricing) - [Sigma Computing, *Product overview*](https://www.sigmacomputing.com) - [Hex, *Notebook-based analytics*](https://hex.tech) - [Mode, *Analyst tool documentation*](https://mode.com) - [Rows, *Modern spreadsheet*](https://rows.com) - [Tableau, *Pricing tiers*](https://www.tableau.com/pricing) - [Microsoft, *Power BI pricing*](https://powerbi.microsoft.com/en-us/pricing/) ### Related reading on Falora - [Supermetrics alternative: the 8 options that actually work](https://falora.ai/blog/supermetrics-alternative) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) - [Self-reported attribution: the only B2B attribution that survives](https://falora.ai/blog/self-reported-attribution-b2b) - [Revenue attribution vs marketing attribution](https://falora.ai/blog/revenue-attribution-vs-marketing-attribution) - [The Marketing Efficiency Ratio](https://falora.ai/blog/marketing-efficiency-ratio) ### About the author Jeroen De Broyer is co-founder of Falora. He has selected and deployed BI tools across 11+ European B2B SaaS scale-ups and writes on [LinkedIn](https://www.linkedin.com/in/jeroendebroyer). ## FAQ **Why should I move off Looker Studio?** Four common reasons: slow performance on large datasets, weak row-level permissioning for multi-team access, awkward warehouse-native workflows (Looker Studio is optimised for Google's own products), and limited version control or governance. If none of the four apply to your stack, staying on Looker Studio is often correct. **What is the best Looker Studio alternative in 2026?** For most B2B SaaS teams the shortlist is Metabase (open source, self-hosted or Metabase Cloud), Sigma Computing (spreadsheet-familiar interface on top of warehouses), Preset (managed Apache Superset), and Hex (analyst-notebook style with SQL and Python). Tableau and Power BI remain enterprise defaults where the buyer is already committed to a broader Microsoft or Salesforce ecosystem. **How does Metabase compare to Looker Studio?** Metabase is stronger on warehouse-native queries, self-serve exploration, permissioning and embedded analytics. Looker Studio is stronger on Google Analytics 4 and Google Ads native integration, on free-tier pricing, and on lightweight report distribution. Teams that live in Google's stack often keep Looker Studio for reporting alongside Metabase for exploration and internal dashboards. **Is Sigma Computing worth the price?** For teams whose stakeholders think in spreadsheets and whose warehouse is Snowflake, BigQuery or Databricks, yes. Sigma's spreadsheet interface on top of warehouse queries removes the SQL barrier for non-technical analysts. Pricing at $600 per user per year is a real commitment; the value math works when the team has 10+ active analyst-adjacent users. **What about Tableau or Power BI as Looker Studio alternatives?** Both are strong but rarely the right B2B SaaS mid-market choice in 2026 unless the buyer is already in the corresponding enterprise ecosystem (Salesforce for Tableau, Microsoft for Power BI). Pricing, learning curve and deployment overhead are all significant. Look at them if you have enterprise governance requirements; look at Metabase, Sigma or Preset first if you do not. --- # Supermetrics alternative in 2026: the 8 options that actually work - URL: https://falora.ai/blog/supermetrics-alternative - Language: en - Author: Jeroen De Broyer (Co-founder, Falora) - Published: 2026-07-03 - Category: How-to - Tags: supermetrics, data-pipeline, marketing-data, vendor-comparison - Reading time: 13 min - Translation key: supermetrics-alternative ## TL;DR Supermetrics dominates the spreadsheet-and-BI marketing data pipeline market. In 2026 the best alternatives are Funnel.io for enterprise, Windsor.ai for cost-sensitive teams, Fivetran or Airbyte for warehouse-first stacks, and Coupler.io for lightweight spreadsheet needs. Pick by destination, not by feature list. ## Description Looking for a Supermetrics alternative in 2026? The 8 pipelines B2B GTM teams actually deploy, ranked by connector coverage, price, warehouse fit and support quality. ## Full content ## TL;DR > Supermetrics still owns the spreadsheet-and-BI marketing data pipeline segment. In 2026 the right alternative depends on where your data lands. This is the field guide, by destination. - 8 credible alternatives in 2026, sorted by fit: Funnel.io, Windsor.ai, Fivetran, Airbyte, Coupler.io, Improvado, Adverity, Whatagraph. - Pick by destination first (spreadsheet, BI tool, or warehouse), then by connector coverage, then by price. - Warehouse-first stacks should skip Supermetrics-class tools entirely and standardise on Fivetran or Airbyte. - Total cost at mid-market scale: Supermetrics $600 to $2,500 per month; alternatives 20 to 50 percent cheaper at equivalent scope. ## Introduction Supermetrics did not become the default marketing data pipeline by accident. The connector coverage is broad (150+ sources), the Google Sheets integration is genuinely convenient, and the Looker Studio native templates ship in one click. For a marketing team whose destination is a spreadsheet or a Looker Studio dashboard, Supermetrics is often the fastest path to a working report. The reasons teams look for an alternative are also well-known. Pricing that scales unpredictably. Limited destination-side data modelling. A UX designed for the spreadsheet era in a market that is standardising on warehouses. Recent Supermetrics acquisitions and product-line consolidation have added confusion for existing customers. This article is for the RevOps lead, marketing analyst or GTM engineer who is evaluating a switch. Not a vendor pitch. A destination-first buying framework and the 8 alternatives that actually earn their place on the shortlist. Ranked, priced, opinionated. ## Pick by destination first, not by connector list Every Supermetrics evaluation starts with the same wrong question: which tool has the most connectors? The connector count matters less than the destination, because the destination determines how much modelling and transformation work sits downstream. Three destinations dominate B2B GTM data pipelines in 2026. **Spreadsheet destination.** Google Sheets or Excel. Fast to build, painful to maintain past 10 recurring reports. Right for very small teams and one-off analyses. **BI tool destination.** Looker Studio, Tableau, Power BI, Metabase, others. Data lands as flat tables or blended datasets ready for visualisation. Right for teams that report to multiple stakeholders but do not need warehouse-scale analytics. **Warehouse destination.** Snowflake, BigQuery, Databricks, Redshift. Data lands as raw, versioned, joinable tables. Right for teams with SQL fluency, cross-source analysis needs, or GTM engineering ambitions. Match the destination to the tool. Warehouse-destination teams should not run Supermetrics; they should run Fivetran or Airbyte. Spreadsheet-destination teams should not deploy Fivetran; the overhead is prohibitive. The right tool is the one that fits the destination. ## The 8 alternatives, ranked | Tool | Best for | Connectors | Starting price | Warehouse fit | |---|---|---|---|---| | Funnel.io | Mid-market blended reporting | 500+ | $999/mo | Strong | | Windsor.ai | Cost-sensitive mid-market | 300+ | $23/mo | Adequate | | Fivetran | Enterprise warehouse ELT | 400+ | $500/mo (usage-based) | Native | | Airbyte | Open-source warehouse ELT | 400+ | Free (self-host) or $360/mo (cloud) | Native | | Coupler.io | Lightweight spreadsheet | 60+ | $49/mo | Adequate | | Improvado | Enterprise marketing analytics | 500+ | $2,500/mo | Strong | | Adverity | Enterprise data ops | 600+ | Custom | Strong | | Whatagraph | Agency reporting | 50+ | $199/mo | Weak | Numbers are Q2 2026 published pricing and connector counts. Real cost varies with data volume and connector selection. ## Funnel.io The strongest all-round Supermetrics alternative for mid-market B2B SaaS teams. Two features stand out. The first is **normalised data models across sources.** Where Supermetrics returns each connector's data in its native shape (Meta returns Meta columns, Google returns Google columns), Funnel.io maps sources to a consistent schema so blended reporting is clean by default. This saves 40 to 100 hours per year of downstream data preparation. The second is **destination-side transformation**. You can define blends, filters and calculated fields inside Funnel.io before the data lands in your BI tool or warehouse. Supermetrics pushes this work downstream. Weakness: pricing scales fast. Below 10 active connectors Funnel.io is often more expensive than the alternatives. Above 30 active connectors the value math works. Right for: mid-market B2B SaaS with multi-channel paid media, an in-house RevOps analyst and either Looker Studio or a lightweight warehouse as the destination. ## Windsor.ai The cost-optimised alternative. 300+ connectors, warehouse destinations supported, transparent per-connector or per-destination pricing that stays predictable at scale. The trade-off is UX polish. Windsor.ai's dashboards and setup flows are functional rather than beautiful. If your team values design or your reports are shown to executives, this matters. If your team is technical and destination-first, it does not. Right for: cost-sensitive teams (bootstrapped, seed-stage, series A) that need Supermetrics-level breadth at 30 to 60 percent of the price. ## Fivetran and Airbyte Neither is a Supermetrics substitution; both are architectural alternatives. If your destination is a warehouse and you have SQL fluency, you skip the Supermetrics-class category entirely and standardise on Fivetran or Airbyte for all data ingestion, then transform downstream in dbt or equivalent. **Fivetran** is the enterprise standard. Connectors are maintained aggressively, syncs are reliable, pricing is usage-based (Monthly Active Rows). At mid-market scale expect $500 to $2,500 per month; at enterprise scale $10K per month is common. **Airbyte** is the open-source alternative. Self-hosted Airbyte is functionally free (you pay only infrastructure); Airbyte Cloud starts around $360 per month. Connector coverage matches Fivetran for popular sources, lags for long-tail sources. Configuration and monitoring require more operator attention. The architectural bet is clear: warehouse-first means dbt for transformation, one of these for ingestion, a BI tool of choice for consumption. If that is your architecture, do not use Supermetrics. ## Coupler.io The lightweight spreadsheet alternative. If your destination genuinely is Google Sheets and your team is small, Coupler.io delivers Supermetrics-adjacent functionality at $49 to $199 per month with cleaner pricing and a simpler UX. Trade-off: fewer connectors (60+ vs 150+) and weaker enterprise features. Right for: bootstrapped teams, small marketing teams, agencies with 3 to 10 clients each needing simple spreadsheet destinations. ## Improvado Enterprise marketing analytics platform. Not a direct Supermetrics substitution; a broader category that includes ingestion, transformation, analytics and reporting. Right for teams with $1M+ marketing spend, multiple channels, and dedicated analytics headcount. Starting price of $2,500 per month reflects the enterprise scope. Most B2B SaaS scale-ups under €30M ARR find this over-engineered. ## Adverity European enterprise data operations platform. Strong in EU compliance posture (GDPR, EU AI Act readiness), strong in connector breadth, custom pricing. Right for: European enterprise teams with EU data-residency requirements and a preference for a Vienna-headquartered vendor. ## Whatagraph Marketing-agency reporting tool. Native templates for client reports, white-label options, moderate connector coverage. Not a warehouse-fit tool; if your destination is a warehouse skip this. Right for: marketing agencies with 10+ clients each needing branded reports. ## The connector coverage question, honestly Most alternative comparisons obsess over connector counts. The reality: for B2B SaaS teams the 10 to 20 connectors that actually matter are covered by every tool on this list. Google Ads, Meta Ads, LinkedIn Ads, HubSpot, Salesforce, Google Analytics 4, Bing Ads, Reddit Ads, TikTok Ads, GA4, Search Console, and the usual few CRM and support-tool sources. Where connector count matters is for long-tail sources (obscure ad platforms, regional tools, verticalised software). If you depend on a specific long-tail connector, verify it exists on your shortlist before comparing prices. If you do not, ignore the connector count entirely. ## The pricing game Every vendor in this category prices with three dimensions: number of connectors, data volume, and number of destinations. Vendors emphasise the dimension where they win. Supermetrics quotes low starting prices and adds costs at the connector and volume tier. Funnel.io quotes fixed monthly tiers with generous data volumes. Windsor.ai quotes per-connector pricing that stays predictable. Fivetran quotes usage-based Monthly Active Rows. Airbyte quotes per-sync or per-record depending on tier. The comparison method that works: pick your target scope (10 connectors, 3 destinations, 5M rows per month), get a quote from each vendor for that specific scope, compare like-for-like. Sticker prices are misleading; scope-based quotes are honest. ## The 5-question shortlist test Before signing a contract with any Supermetrics alternative, answer these five questions in writing. **One: what is your destination?** Spreadsheet, BI tool or warehouse. If you cannot answer within 60 seconds, do not sign yet. **Two: what data volume do you actually move per month?** Underestimate consistently produces surprise bills. **Three: what is your dbt or transformation strategy?** If you have one, favour warehouse-native tools. If you do not, favour tools that transform destination-side. **Four: who owns the pipeline day-to-day?** RevOps analyst, GTM engineer, external agency. The tool's UX should match the owner's fluency. **Five: what is your renewal exit plan?** All of these tools lock you into their connector configuration to some degree. Understand the export path before you commit. Adam Robinson of RB2B, in a different context but the same spirit: > "Person-level identity creates a much tighter feedback loop for modern B2B operators." The same is true of your data pipeline. A tighter feedback loop between ingestion, transformation and reporting is the actual goal. The tool that delivers it for your specific stack is the right choice, regardless of the marketing category. ## What we use at Falora For transparency, since this is a vendor category we operate in: Falora's internal marketing data pipeline runs on a mix of Airbyte (self-hosted, ingestion into Snowflake), dbt for transformation, and Metabase for consumption. We used Supermetrics for a season and moved off because our destination shifted from Looker Studio to Snowflake. That specific transition is the migration this article maps. Falora as a product is a GTM orchestration and governance layer, not a data pipeline. The pipeline tools sit alongside Falora, feeding the measurement layer of [the GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy). ## Frequently asked questions **Is Supermetrics being acquired?** Supermetrics has undergone several strategic changes in 2025 to 2026 including product-line consolidation. Verify the current corporate status directly with the vendor before making a long-term commitment. **Can I use two tools together (e.g. Supermetrics for spreadsheets, Fivetran for warehouse)?** Yes, and many teams do. The cost is doubled tooling; the benefit is best-fit per destination. Reasonable at mid-market and above. **What is the fastest migration path off Supermetrics?** For spreadsheet destinations: switch to Coupler.io or Windsor.ai in a week. For BI destinations: switch to Funnel.io or Windsor.ai in two to three weeks. For warehouse destinations: standardise on Fivetran or Airbyte over a quarter, run in parallel with Supermetrics for the first 30 days. **Does Falora replace any of these tools?** No. Falora is the GTM orchestration layer that consumes clean data from your pipeline of choice. We are complementary, not competitive. **Which is best for GDPR compliance?** Adverity (Vienna-based) and Windsor.ai (EU-focused pricing) have the strongest EU compliance posture. Fivetran and Airbyte support EU data residency options. Supermetrics is generally compliant but the corporate footprint has shifted; verify current data residency options. ## Conclusion Supermetrics is not the enemy. It is a tool that fits a specific segment (spreadsheet or Looker Studio destinations, marketing-team ownership) and has served that segment well. The reasons teams leave are legitimate: pricing that scales unpredictably, destinations that have shifted to warehouses, ownership that has shifted from marketing to RevOps or GTM engineering. The alternative that matches your stack in 2026 depends on your destination and your team fluency. Pick with those two questions first; every other consideration follows. If you want help mapping your data pipeline decision against the broader GTM stack architecture, **[book a 30-minute review with Falora](https://falora.ai/demo)**. --- ### Sources - [Funnel.io, *Pricing and feature documentation*](https://funnel.io/pricing) - [Windsor.ai, *Product and pricing*](https://windsor.ai) - [Fivetran, *Pricing model and connector list*](https://www.fivetran.com/pricing) - [Airbyte, *Open-source and cloud pricing*](https://airbyte.com/pricing) - [Coupler.io, *Feature overview*](https://coupler.io) - [Improvado, *Enterprise pricing*](https://improvado.io) - [Adverity, *Data operations platform*](https://www.adverity.com) - [Kyle Poyar, *Growth Unhinged marketing data stack analysis*](https://www.growthunhinged.com) ### Related reading on Falora - [Looker Studio alternative: the 8 BI tools B2B teams should evaluate](https://falora.ai/blog/looker-studio-alternative) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) - [Self-reported attribution: the only B2B attribution that survives](https://falora.ai/blog/self-reported-attribution-b2b) - [Revenue attribution vs marketing attribution](https://falora.ai/blog/revenue-attribution-vs-marketing-attribution) - [The Marketing Efficiency Ratio: the metric that survives dark social](https://falora.ai/blog/marketing-efficiency-ratio) ### About the author Jeroen De Broyer is co-founder of Falora. He has evaluated, deployed and migrated marketing data pipelines across 11+ European B2B SaaS scale-ups. He writes on [LinkedIn](https://www.linkedin.com/in/jeroendebroyer). ## FAQ **What is the best Supermetrics alternative for B2B SaaS teams in 2026?** For most B2B SaaS teams the right alternative is Funnel.io (if you need advanced blending and mid-market pricing), Windsor.ai (if cost is the primary constraint and 300+ connectors are enough), or Fivetran or Airbyte (if your destination is a warehouse and you want to standardise on one enterprise-grade pipeline). Coupler.io wins for teams that only need spreadsheet destinations. **Why do teams leave Supermetrics?** Three reasons dominate in our portfolio: pricing that scales unpredictably with connector count and data volume; limited support for destination-side transformation (you push raw data and clean it downstream); and a UX built for Google Sheets and Looker Studio that becomes awkward when your team standardises on a warehouse. None of the three are dealbreakers for every team, but they push mature teams to alternatives. **How much does Supermetrics cost in 2026?** Supermetrics pricing starts around $69 per month for the Google Sheets add-on with basic connectors and scales to $1,200+ per month for the Enterprise BI tier with warehouse connectors and higher data limits. Real all-in cost at mid-market scale is typically $600 to $2,500 per month once you add advanced connectors and data volume tiers. Alternatives like Windsor.ai and Funnel.io often run 20 to 50 percent cheaper at equivalent scope. **Is Funnel.io a good Supermetrics alternative?** Yes, for teams that need blended reporting across many paid-media sources plus CRM data. Funnel.io's strength is destination-side data model consistency (columns are normalised across sources), which makes downstream BI cleaner. Weakness is enterprise-grade pricing that becomes uncompetitive below 20 active connectors. **Should I use Fivetran or Airbyte instead of Supermetrics?** Yes, if your destination is a data warehouse (Snowflake, BigQuery, Databricks, Redshift) and you have SQL fluency. Fivetran is the enterprise choice with the widest connector coverage and premium pricing. Airbyte is the open-source alternative that runs cheaper if you self-host and are comfortable with more configuration. Both are architectural bets, not just Supermetrics substitutions; they push transformation downstream to dbt or equivalent. --- # MCP for GTM: the new integration layer for autonomous revenue - URL: https://falora.ai/blog/mcp-for-gtm - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-06-19 - Category: Insights - Tags: mcp, ai-agents, integration, autonomous-gtm - Reading time: 15 min - Translation key: mcp-for-gtm ## TL;DR Model Context Protocol (MCP) is becoming the de facto integration standard between AI agents and enterprise systems. For GTM teams it changes how agents read CRM, signal sources and product data. It also introduces a new third-party supply-chain risk that needs explicit governance. ## Description Model Context Protocol is becoming the integration standard for AI agents. What it means for your GTM stack, where the security risks are, and how to adopt it without exposing the business. ## Full content ## TL;DR > Model Context Protocol is becoming the integration standard for AI agents in 2026. For B2B GTM teams it changes how agents read your CRM, your signal sources and your product data. The security situation is real but manageable through vendor-managed adoption. - MCP standardises agent-to-system integration the way REST standardised app-to-app integration. - 1 in 4 MCP servers opens AI agents to code-execution risk (Help Net Security, May 2026). - Direct MCP adoption requires mature security engineering; vendor-managed adoption is the pragmatic 2026 path. - The vendor selection criterion shifts: pick vendors that expose data via MCP, not vendors that require custom integration. ## Introduction Every B2B GTM stack is a federation of systems that do not speak the same language. The CRM speaks REST. The data warehouse speaks SQL. The enrichment vendors speak webhooks. The intent platform speaks event streams. Connecting them historically required a custom integration per pair, which is why the median scale-up runs 23 tools and uses 8. AI agents intensify the integration problem. An agent needs context from many systems to make a decision. Hand-wiring each integration multiplies cost, multiplies maintenance and multiplies failure modes. Model Context Protocol (MCP) is the open standard that is consolidating as the answer. This article is for the GTM engineer, RevOps lead or CTO who has heard MCP mentioned three times in the last month and needs to understand what it changes about the integration layer, where the security risk sits and how to adopt it without exposing the business. It is written for B2B operators, not for protocol implementers. ## What MCP is, in operator terms MCP (Model Context Protocol) is an open specification, originated by Anthropic and rapidly adopted across the AI ecosystem, that defines how AI agents request and receive context from external systems. The agent says "I need the recent activity for this account"; an MCP server connected to the CRM responds with that data in a standardised format the agent can use. Three elements compose the standard. **MCP servers.** A small piece of software that wraps an existing data source (a CRM, a database, a SaaS API) and exposes it to AI agents in MCP's standardised format. A Salesforce MCP server lets agents query Salesforce data without bespoke integration code. **MCP clients.** The agent runtime (Claude, GPT-based agent frameworks, autonomous platforms) that consumes the MCP servers. The client handles the protocol negotiation, authentication delegation and response parsing. **MCP tools and resources.** The specific operations exposed by a server (e.g. *search contacts*, *read account history*, *update lead score*). The agent knows what tools are available and chooses among them based on the task. The functional analogy is to REST APIs in the early 2010s. Before REST became universal, integrating two SaaS apps required custom code per pair. REST collapsed the integration cost and enabled the integration explosion of the 2010s. MCP plays the same role for agent-to-system integration in the 2020s. ## Why this matters for GTM specifically Three GTM-specific implications follow from MCP adoption. **Integration cost between agents and your stack drops by an order of magnitude.** Pre-MCP, every agent-to-system connection was a custom build. Post-MCP, a single MCP server exposes the system to all agents that speak the protocol. For a B2B GTM stack with 15 systems and 5 agent workflows, that is a 75-integration reduction to 15. **Existing identity and access management extends automatically.** When properly implemented, agents authenticate to MCP servers using the same OAuth or SAML flows as human users. The permissions you have configured in Salesforce or HubSpot apply to agent actions without reconfiguration. Tomasz Tunguz's framing applies again: > "LLMs are deflationary for software." MCP is one of the mechanisms through which that deflation reaches the buyer. Less integration code, less middleware, less custom IAM work. **Vendor selection criteria shift.** In 2025 you picked GTM vendors based on feature parity and API quality. In 2026 the new criterion is MCP coverage: does the vendor expose its data via a maintained MCP server, with documented tools and resources, that your agents can consume reliably? Vendors that do not will face an integration tax that vendors that do will not. ## The security situation, honestly The standard is mature enough to adopt and not mature enough to adopt blindly. Help Net Security's analysis of the MCP ecosystem in May 2026 found that 1 in 4 MCP servers open AI agents to code-execution risk through tool poisoning, schema injection or unauthenticated tool exposure. Multiple CVSS 9.0+ vulnerabilities were disclosed against popular MCP integrations in the first half of 2026. ITECS' MCP Tool Poisoning analysis documented the specific attack class where a malicious tool description in an MCP server compromises an agent's downstream decisions. Three risk vectors matter for GTM deployments. **Tool poisoning.** A malicious MCP server returns tool descriptions designed to manipulate the agent into harmful actions (data exfiltration, unauthorised actions on other tools). Mitigation: only consume MCP servers from vetted vendors or your own infrastructure. **Schema injection.** An MCP server returns malformed responses that exploit weaknesses in the agent's parser. Mitigation: keep agent runtimes patched; use vendor-managed agent platforms that handle this layer. **Subprocessor sprawl.** Each MCP server is functionally a subprocessor with access to the underlying system data. A 10-server deployment is a 10-subprocessor expansion. Mitigation: explicit governance, inventory, audit rights per server. The honest synthesis is that MCP is in the phase that every protocol goes through after fast adoption and before mature governance. It is broadly safe to consume through vetted vendors, broadly risky to self-host without security engineering capability. ## Two adoption paths for B2B GTM teams in 2026 **Path A: vendor-managed MCP (recommended for most teams).** Your autonomous GTM platform abstracts MCP behind a vetted security perimeter. You configure which systems the agent can access and which actions it can take. The platform handles the MCP server selection, the subprocessor due diligence and the security patching. The integration cost benefits accrue; the security risk is delegated. This is Falora's adoption pattern. We curate the MCP server set, vet each server's security posture, expose only the tool subset that is safe and useful for GTM use cases, and maintain the patch cadence. **Path B: direct MCP adoption (for teams with security engineering).** You self-host MCP servers for your critical systems, manage the security perimeter internally and consume them with your own agent runtimes. The integration cost benefits and the architectural flexibility are higher. The security responsibility is yours. This is appropriate for teams with mature DevSecOps capability and a clear strategic reason to own the layer. Most B2B scale-ups under €50M ARR fall under Path A. The third path (self-hosting MCP servers without security engineering capability) is the path to a Q3 incident. Avoid it. ## What MCP changes about vendor selection Three vendor criteria become first-class in 2026, alongside the criteria that already mattered. **Does the vendor expose its data via a maintained MCP server?** The vendor that says "yes, here is the MCP server documentation, here are the supported tools, here is the security review" is the vendor that will integrate cheaply and safely with your agent stack. The vendor that says "no, we only have a REST API; you can build the wrapper yourself" is a vendor that will cost you integration time and security exposure. **What is the MCP server's tool surface?** A well-designed MCP server exposes a curated set of tools (5 to 20) that map to common use cases. A poorly designed one exposes hundreds of tools (effectively the full API surface), which multiplies the security review burden and confuses the agent's tool selection. **How does the vendor handle MCP server security?** Look for: a published security review of the MCP server, a patch cadence, subprocessor disclosure, audit logging of agent actions. A vendor that cannot answer these questions is not ready for production agent deployment. These three questions belong in your vendor RFP template starting now. ## What MCP changes about your GTM architecture The 7-layer reference architecture we documented in [the GTM engineering system anatomy](https://falora.ai/blog/gtm-engineering-system-anatomy) gets a small but consequential update. | Layer | Pre-MCP | Post-MCP | |---|---|---| | 1. Signal | Custom webhooks per signal source | MCP server per signal source | | 2. Identity & enrichment | REST API per provider | MCP server per provider | | 3. Reasoning | LLM with custom tool definitions | LLM consuming standardised MCP tools | | 4. Orchestration | Custom integration code | MCP-aware orchestrator | | 5. Activation | Custom API integration per channel | MCP server per channel where available | | 6. Measurement | Custom event capture | MCP server exposing measurement reads | | 7. Governance | Custom audit trail | MCP includes standardised request logging | The architectural shape is unchanged. The wiring between layers is dramatically simpler. The governance layer (Layer 7) becomes structurally easier because MCP requests are logged by default. Adam Robinson's observation about feedback loops applies here too: > "Person-level identity creates a much tighter feedback loop for modern B2B operators." MCP creates a tighter feedback loop for agent operators. Every action is a standardised request with a standardised response, captured in a standardised log. Debugging becomes possible in a way that custom-integration agent stacks never allowed. ## A 60-day MCP readiness plan For a B2B GTM team starting from a fully custom integration baseline, here is the 60-day readiness plan. **Days 1 to 14: inventory.** List every system in your GTM stack. For each, note: does the vendor publish an MCP server? Is the server vendor-maintained or community-built? What is the documented tool surface? Build a single spreadsheet that captures the answer per system. **Days 15 to 30: vendor conversations.** For the systems where the vendor does not yet publish an MCP server, ask. Most enterprise vendors are working on it; raising the question with your account team accelerates the timeline. For vendors with no MCP roadmap, factor that into renewal decisions. **Days 31 to 45: pick the platform.** Decide on Path A (vendor-managed) or Path B (direct adoption). For Path A, evaluate autonomous GTM platforms on the three MCP criteria above. For Path B, scope the security engineering investment honestly. **Days 46 to 60: pilot.** Deploy one MCP-enabled agent workflow with one or two MCP servers. Measure: integration time, agent reliability, audit trail completeness, exception rate. Use the pilot to calibrate the broader rollout plan. By day 60 you have an MCP readiness baseline, a vendor pressure plan and a working pilot. The broader rollout follows in quarter two. ## What MCP does not solve Three problems MCP is sometimes credited with solving and does not. **Data quality.** MCP standardises how agents access data; it does not improve the data. A CRM with 20% duplicate accounts will be a CRM with 20% duplicate accounts that an agent can now query more easily. Data hygiene is upstream of MCP and remains a separate workstream. **Agent reliability.** MCP gives agents better access to context; it does not make agents reason better. The reasoning layer (Layer 3) and the exception routing (part of Layer 4) determine agent reliability. MCP improves the inputs to the agent without improving the agent itself. **Governance maturity.** MCP makes governance more tractable by standardising the audit surface. It does not establish the governance posture, the named owner, the exception policy. Those remain organisational work that runs in parallel with the protocol adoption. For the broader governance situation see [the Governance Mirage piece](https://falora.ai/blog/ai-agent-governance-mirage). ## Frequently asked questions **Is MCP an Anthropic-only thing?** No. Anthropic originated the specification but it is an open standard adopted by OpenAI, Google, Mistral and the broader agent framework ecosystem. The standard is governed by a community process and is functionally vendor-neutral at the protocol level. **How does MCP relate to OpenAI's function calling or Anthropic's tool use?** Function calling and tool use are model-level mechanisms for invoking external tools. MCP is the protocol layer that standardises how those external tools are exposed. A model that supports function calling can consume MCP servers; the two compose rather than compete. **Should I delay agent deployment until MCP is more mature?** Generally no. Vendor-managed MCP adoption today is safe and accelerates time-to-value. Direct self-hosted adoption can wait until your security engineering capability is in place. **What is the cost of MCP adoption?** For Path A (vendor-managed), the cost is essentially the platform licence; MCP is included. For Path B (direct), expect 1 to 2 FTE-quarters of security engineering work to stand up the perimeter, plus ongoing maintenance. **Does MCP work with on-premise systems?** Yes. MCP servers can run on-premise, in a VPC or behind a firewall. The protocol is transport-agnostic. This makes MCP useful for regulated industries where data residency matters. ## Conclusion MCP is the protocol layer that B2B GTM stacks will standardise around through 2026 and 2027. The integration cost benefits are immediate. The security risks are real but manageable through vendor-managed adoption for most teams. The vendor-selection criteria you use today need to add MCP coverage to the existing checklist. The teams that are ahead of this curve will integrate new agent capabilities in days instead of months. The teams that are behind will continue paying the custom-integration tax through 2027 and beyond. If you want help running the 60-day MCP readiness assessment for your specific stack, **[book a 30-minute MCP review with Falora](https://falora.ai/demo)**. --- ### Sources - [Help Net Security, *1 in 4 MCP servers opens AI agent security to code execution risk*](https://www.helpnetsecurity.com/2026/05/05/ai-agent-security-skills-blind-spots/) - [Toloka, *The Future of MCP: 2026 Roadmap, Enterprise Adoption*](https://toloka.ai/blog/the-future-of-mcp-enterprise-adoption/) - [ITECS, *MCP Tool Poisoning: Enterprise AI Agent Security 2026*](https://itecsonline.com/post/mcp-tool-poisoning-enterprise-ai-agent-security-2026) - [VentureBeat, *The Agentic Reckoning*](https://venturebeat.com/resources/the-agentic-reckoning-enterprise-ai-organizations-have-a-runtime-problem-not-a-model-problem) - [Gartner, *Uniform governance across AI agents*](https://www.gartner.com/en/newsroom/press-releases/2026-05-26-gartner-says-applying-uniform-governance-across-ai-agents-will-lead-to-enterprise-ai-agent-failure) - [Atlan, *AI Agent Risks & Guardrails: 2026 Enterprise Security Guide*](https://atlan.com/know/ai-agent-risks-guardrails/) - [Anthropic, *Model Context Protocol specification*](https://modelcontextprotocol.io) ### Related reading on Falora - [The Governance Mirage: why 74% of enterprises rolled back AI agents](https://falora.ai/blog/ai-agent-governance-mirage) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [The EU AI Act + your AI GTM stack](https://falora.ai/blog/eu-ai-act-gtm-stack-audit) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. He has architected MCP-enabled agent deployments across European B2B SaaS scale-ups and writes about GTM engineering, autonomous revenue and the EU AI Act on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What is MCP and why does it matter for GTM?** MCP (Model Context Protocol) is an open standard that defines how AI agents request and receive context from enterprise systems. For GTM teams it matters because it changes the integration pattern from custom API wiring to a standardised protocol. When properly implemented, agents inherit existing authentication, authorization and policy controls rather than bypassing them. **Is MCP safe to deploy in production?** Selectively. Help Net Security reported in May 2026 that 1 in 4 MCP servers opens AI agents to code-execution risk, with multiple CVSS 9.0+ vulnerabilities disclosed in the first half of 2026. The standard is in the brittle phase where adoption has outpaced governance. Treat MCP servers as third-party supply-chain components with the same due diligence as any subprocessor. **How should B2B GTM teams adopt MCP in 2026?** Through a vetted vendor-managed layer rather than raw self-hosted MCP servers. The platform that abstracts MCP behind a security perimeter is the safe deployment path. Direct MCP wiring is only appropriate for teams with mature security engineering capability and a clear use case. **What changes about my GTM stack when I adopt MCP?** Three things. One, integration cost between agents and systems drops by an order of magnitude. Two, your existing identity and access management extends to agent actions automatically. Three, vendor selection criteria shift; pick vendors that expose data via MCP, not vendors that require custom integrations. **Will MCP replace REST APIs?** No. MCP and REST coexist. MCP is the protocol agents use to consume context; REST is the protocol systems use to expose data. The two operate at different layers. In practice, MCP servers wrap existing REST APIs to make them agent-consumable, not replace them. --- # AI in B2B sales compensatie: wat verandert voor SDR, AE en RevOps - URL: https://falora.ai/nl/blog/ai-b2b-sales-compensatie - Language: nl - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-06-12 - Category: Insights - Tags: sales-compensatie, ai-sdr, revops, benelux - Reading time: 13 min - Translation key: ai-b2b-sales-compensatie ## TL;DR De junior SDR rol verdwijnt sneller in BE/NL dan in de VS. RevOps en GTM engineers met AI fluency krijgen 5 tot 12 procent loonpremie. AE quota:OTE blijft 4-6x, maar pay mix schuift naar 55/45. Dit artikel is de salaris- en commissiebenchmark voor 2026. ## Description De junior SDR rol comprimeert sneller in de Benelux dan in de VS. Wat je in 2026 in plaats daarvan moet aanwerven, hoeveel je hen moet betalen en hoe het commissieplan verandert. ## Full content ## TL;DR > De junior SDR rol verdwijnt sneller in de Benelux dan in de VS. RevOps en GTM engineers met AI fluency krijgen een loonpremie van 5 tot 12 procent. Het commissieplan schuift mee. Dit is de Benelux-benchmark voor 2026. - Cengage 2026: 46 procent van werkgevers noemt AI als reden voor minder entry-level hires. - Sales Cookie 2026: quota:OTE blijft 4-6x; pay mix verschuift naar 55/45 voor mid-market en enterprise AE. - 45.000 tech jobs gecut in Q1 2026 alleen al; AI expliciet genoemd als driver in minstens 20 procent. - RevOps en GTM engineers met AI fluency: 5 tot 12 procent loonpremie boven standaard tier benchmarks. - Concrete BE/NL salarisranges in dit artikel, per rol, per senioriteit. ## Inleiding In de Benelux is de sales-jobmarkt sinds Q4 2025 stilletjes maar fundamenteel verschoven. Belgische en Nederlandse scale-ups die in 2023 nog drie SDRs aanwierven om hun pipeline te vullen, werven in 2026 één GTM engineer aan, één AI-fluent operator, en geen SDR. De totale loonkost daalt met 30 tot 40 procent. De pipeline-output stijgt. Dit artikel is voor de Belgische of Nederlandse CMO, CRO, founder of head of people die het volgende compensatieplan moet bouwen. Het bevat concrete salarisranges per rol per senioriteit, gebaseerd op publieke benchmarks (StepStone, Indeed, Acerta, Talent.com), aangevuld met de comp-data die we verzamelden over 18 GTM rebuilds in de Stretch Innovation portfolio. Het artikel gaat niet over de vraag *of* AI sales-rollen comprimeert. Die vraag is achterhaald. Het gaat over welke profielen je in plaats daarvan aanwerft en hoe je hen betaalt zonder de markt over of onder te bieden. ## Wat verandert er in 2026 echt Drie verschuivingen tegelijk, in de Benelux versterkt door drie lokale factoren. **Verschuiving 1: junior SDR rol comprimeert.** Cengage's 2026 survey: 46 procent van werkgevers noemt AI als reden voor minder entry-level hires. De rol verdwijnt niet volledig, maar het ingangspunt voor junior sales-talent versmalt. SDRs die nog aangenomen worden, zijn AI-fluent (Clay, n8n, prompt design) eerder dan louter outreach-vaardig. **Verschuiving 2: RevOps en GTM engineering professionaliseren.** Sales Cookie's *State of B2B Sales Compensation 2026*: RevOps en sales ops profielen met AI fluency commanderen een loonpremie van 5 tot 12 procent boven standaard tier benchmarks. De rol gaat van "iemand die rapporten draait" naar "iemand die het GTM systeem ontwerpt en runt". Dat is een andere competentie en een andere prijs. **Verschuiving 3: AE pay mix schuift naar meer variabele.** Sales Cookie 2026: voor mid-market en enterprise AE rollen schuift de pay mix incrementeel naar 50/50 of 55/45 (variabele:fixed). Quota:OTE multiple blijft stabiel rond 4 tot 6x voor closing AEs, maar bedrijven die naar 7 of 8x pushen zien attainment onder 60 procent en turnover boven 25 procent. In de Benelux versterken drie lokale factoren deze verschuivingen. Eerst is de bureauwereld in BE/NL dicht en goed verkocht (House of Marketing, Universem, Searchlab, K2). Een marketing-team dat een SDR bureau cancelt en in plaats daarvan een autonoom GTM platform deployt, doet effectief wat een tweede SDR-hire zou hebben gedaan, voor de helft van de kost. De rol-comprimering loopt via de bureau-vervanging. Tweede is de juniorrol-versmalling in de Benelux extra scherp omdat het loongat tussen junior en senior smal is (typisch 60 procent in BE vs 100 procent in de VS). De economie van een junior SDR met €45K base + €15K variabele werkt minder snel uit dan voor een Amerikaanse junior aan $48K + $24K. De marge voor automatisering is groter. Derde is de hire-cycle in BE/NL trager (gemiddeld 92 dagen voor een senior sales hire vs 56 in de VS, volgens StepStone Q1 2026 data). Teams die kunnen wachten op een betere senior-hire in plaats van een mediocre junior, doen dat. ## De ranges, per rol, per senioriteit Alle ranges zijn bruto base salaris, exclusief variabele en equity, voor full-time Belgische of Nederlandse rollen in een B2B SaaS scale-up tussen €3M en €30M ARR. Nederland ligt typisch 8 tot 15 procent boven België. **Junior SDR (1-2 jaar ervaring)** | Onderdeel | België | Nederland | |---|---|---| | Base | €38K - €48K | €42K - €54K | | Variabele (% OTE) | 30-40% | 30-40% | | OTE | €55K - €68K | €60K - €76K | | Toelichting | Rol comprimeert; verwacht aanwervingen in 2026 dalen | Idem | **Medior SDR (2-4 jaar) of AI-fluent SDR** | Onderdeel | België | Nederland | |---|---|---| | Base | €48K - €60K | €54K - €68K | | Variabele (% OTE) | 30-40% | 30-40% | | OTE | €70K - €85K | €78K - €95K | | Toelichting | AI-fluent (Clay, prompts, n8n) bovenkant range | Idem | **Mid-market AE (3-6 jaar)** | Onderdeel | België | Nederland | |---|---|---| | Base | €60K - €78K | €68K - €88K | | Variabele (% OTE) | 45-50% | 45-50% | | OTE | €110K - €155K | €125K - €175K | | Toelichting | Quota:OTE multiple 4-5x | Quota:OTE multiple 4.5-5.5x | **Enterprise AE (6+ jaar)** | Onderdeel | België | Nederland | |---|---|---| | Base | €80K - €110K | €90K - €125K | | Variabele (% OTE) | 45-55% | 45-55% | | OTE | €145K - €220K | €165K - €250K | | Toelichting | Quota:OTE 5-6x; pay mix schuift naar 55/45 | Idem | **RevOps lead (5+ jaar) met AI fluency** | Onderdeel | België | Nederland | |---|---|---| | Base | €75K - €100K | €85K - €115K | | Variabele (% OTE) | 15-25% | 15-25% | | OTE | €88K - €125K | €100K - €145K | | Toelichting | +5 tot 12% premium met AI fluency | Idem | **GTM engineer (junior tot senior)** | Senioriteit | België OTE | Nederland OTE | |---|---|---| | Junior (1-3 jaar) | €70K - €90K | €78K - €100K | | Medior (3-6 jaar) | €95K - €130K | €105K - €145K | | Senior (6+ jaar, stack ownership) | €130K - €170K | €145K - €195K | | Toelichting | Variabele 15-25%, KPI-gebonden | Idem | Equity in scale-up context is bovenstaand niet meegerekend. Typisch 0,05 tot 0,2 procent voor mid-market AE en RevOps, 0,1 tot 0,4 procent voor GTM engineer in Series A/B. ## Het nieuwe commissieplan: weg van pure quota Voor sales-rollen blijft de quota-gedreven structuur intact maar wordt strikter gehandhaafd. Voor RevOps en GTM engineering verschijnen nieuwe variabele-componenten die niet aan quota gekoppeld zijn maar aan **systeem-output**. Drie modellen die we in onze portfolio in productie zien. **Model 1: cost-per-meeting target.** De RevOps lead of GTM engineer krijgt een quarterly target voor cost per qualified meeting (bv. dalend van €450 naar €350 over twee kwartalen). Behalen = volledige variabele. Overhalen = 1,3 tot 1,5x bonus. **Model 2: pipeline gegenereerd door systeem.** Het deel van pipeline dat aantoonbaar voortkomt uit autonomous workflows (versus uit individuele rep-actie) wordt apart gemeten. Een percentage van die pipeline-waarde gaat naar de GTM engineer als variabele. **Model 3: NRR-koppeling voor CS-adjacent rollen.** Voor RevOps profielen die ook customer success raken: een deel van de variabele koppelt aan Net Revenue Retention. NRR werd in 2026 *de* valuation metric (volgens AI Magicx, $4 tot $7 in beschermde omzet per $1 uitgegeven aan churn-prediction AI). De fundamentele verschuiving in alle drie de modellen is dat de variabele niet meer enkel het individu beloont voor individuele actie, maar voor de output van het systeem dat het individu ontwerpt en runt. Dat is een andere prikkel en het trekt een ander profiel. ## Wat dit betekent voor wie je aanwerft in 2026 De praktische take voor een Belgische of Nederlandse founder of CMO die nu een sales-team opbouwt: niet meer de tweede SDR. Vier hire-prioriteiten in volgorde, voor een B2B SaaS scale-up tussen €3M en €15M ARR. **Prioriteit 1: één senior AE.** Mid-market of enterprise afhankelijk van ACV. OTE €125K tot €175K in BE, hoger in NL. Deze persoon multi-threadt deals en sluit. De rol blijft mens. **Prioriteit 2: één GTM engineer of AI-fluent RevOps lead.** Combineer onder één persoon als budget krap is. OTE €100K tot €140K. Deze persoon bouwt en runt het signaal-tot-revenue systeem. **Prioriteit 3: één AI-fluent SDR of operator.** Eerder dan een traditionele SDR. OTE €70K tot €90K. Werkt het systeem en de exceptions, niet de cadence. **Prioriteit 4: één AE-junior of expansion-rol.** Pas wanneer ARR boven €10M zit en het systeem stabiel draait. OTE €90K tot €120K. Voor de meeste scale-ups van €3M tot €15M ARR is dit een team van drie tot vier personen. Een team van zeven was de norm in 2023 voor hetzelfde ARR-bracket. De tooling-spend stijgt; de loonsom daalt netto. ## Wat dit betekent voor wie je behoudt Drie behoud-strategieën die we in de portfolio zien werken. **Eén: bied SDRs een GTM engineer pad.** Junior SDRs die fluent worden in Clay, n8n en prompt design krijgen een ramp naar GTM engineer of RevOps binnen 12 tot 18 maanden, met bijhorende compensatie-sprong. Wie geen interesse heeft of niet de capabiliteit, zal natuurlijk uitstromen. **Twee: herschrijf het AE comp-plan rond multi-threading en complex deal close.** Volume-quota op SQL-niveau wordt minder zinvol als het systeem de SQLs genereert. Quota op closed-won en op expansion-NRR worden zinvoller. De wijziging zelf zal turnover veroorzaken (15 tot 25 procent in het eerste jaar van de transitie); plan ervoor. **Drie: investeer in AI-fluentie als training, niet als hire.** Een goede mid-market AE kan de basis-AI-fluentie leren in 60 dagen met de juiste training. Een externe hire neemt 6 tot 9 maanden om te rampen. Voor sales-rollen is interne upskill bijna altijd goedkoper. ## Veelgestelde vragen **Verdwijnt de SDR rol volledig?** Niet volledig. Voor enterprise en complexe mid-market deals blijft een human SDR voor research, multi-threading en relatie-opbouw zinvol. Voor SMB en volume-outbound onder €10K ACV is de rol functioneel vervangen door het systeem. **Hoe vergelijken BE/NL salarissen met de VS?** Voor SDR en AE rollen liggen BE/NL OTEs typisch 35 tot 50 procent onder VS-equivalenten. Voor GTM engineer rollen is het verschil kleiner (20 tot 35 procent), wat de Benelux relatief aantrekkelijk maakt voor remote VS-startups die EU talent zoeken. **Wat met sales engineering rollen?** Sales engineering blijft een sterke groei-rol in 2026. OTE ranges vergelijkbaar met mid-market AE. AI-fluentie wordt verwacht maar minder kritisch dan voor GTM engineering omdat de output meer demo- en pre-sales gericht is. **Hoe verhoudt RevOps zich tot GTM engineering?** RevOps focust op proces en forecast-accuraatheid; GTM engineering focust op systeem-architectuur en automation. In kleine teams (onder €10M ARR) zit de rol vaak gecombineerd. In grotere teams splitst de rol op met RevOps onder de CRO en GTM engineering onder de CMO. **Wat is de impact op management-rollen?** Sales manager spans groeien van 6 tot 8 reps in 2023 naar 8 tot 12 reps in 2026 omdat het systeem meer van de coaching en monitoring werk doet. Eén layer van middle management kan verdwijnen voor scale-ups onder €25M ARR. ## Conclusie Het 2023-spelboek voor het bouwen van een B2B sales-team in de Benelux (drie SDRs onderaan, twee AEs in het midden, één sales manager bovenaan) is in 2026 commercieel niet meer verdedigbaar. Het systeem doet wat drie SDRs deden, voor minder geld. De rollen die overblijven zijn senior, AI-fluent, en compenseren navenant. De compounding voor de teams die deze verschuiving vroeg maken is groot. Loonsom daalt, output stijgt, en de ruimte komt vrij om een GTM engineer aan te nemen die het systeem ontwerpt in plaats van een vierde SDR die met de hand emails verstuurt. Wil je hulp bij het ontwerpen van de overgang voor jouw team, met concrete comp-bands per rol? **[Plan een sales-team diagnostic in met Falora](https://falora.ai/demo)**. --- ### Bronnen - [Sales Cookie, *State of B2B Sales Compensation 2026*](https://blog.salescookie.com/2026/05/15/state-of-b2b-sales-compensation-2026/) - [Prospeo, *Will AI Replace Sales Jobs 2026*](https://prospeo.io/s/will-ai-replace-sales-jobs) - [DevCommX, *50 Key AI SDR Statistics 2026*](https://www.devcommx.com/blogs/ai-sdr-statistics) - [InformationWeek, *2026 Tech Company Layoffs*](https://www.informationweek.com/it-staffing-careers/2026-tech-company-layoffs) - [Tech Insider, *150K+ Tech Jobs Cut in 2026*](https://tech-insider.org/tech-layoffs-2026-ai-workforce-impact/) - [EasyComp, *Best Sales Compensation Software 2026*](https://www.easycomp.ai/post/best-sales-compensation-software-for-b2b-saas-2026/) - [Acerta, *AI-gebruik in Belgische bedrijven*](https://www.acerta.be) - [StepStone NL, *Salary benchmarks 2026*](https://www.stepstone.nl) ### Verder lezen op Falora - [GTM engineering uitgelegd](https://falora.ai/nl/blog/gtm-engineering-uitgelegd) - [Schaalbare B2B groei zonder bureau](https://falora.ai/nl/blog/schaalbare-b2b-groei-zonder-bureau) - [AI SDRs vs human SDRs: 7 condities waar AI faalt](https://falora.ai/blog/ai-sdr-failure-modes) - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) ### Over de auteur Stijn Van Daele is co-founder van Falora en partner bij Stretch Innovation. Hij begeleidde de team-design en compensatie-keuzes in 18 GTM rebuilds bij Belgische en Nederlandse scale-ups. Hij schrijft over GTM engineering, autonome revenue en de EU AI Act op [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **Hoeveel verdient een SDR in België of Nederland in 2026?** Voor klassieke SDR rollen: junior €38K tot €48K base, medior €48K tot €60K base, met variabele bovenop tussen 25 en 40 procent van OTE. In Nederland ligt de range 8 tot 15 procent hoger dan in België. Voor 2026 wordt deze rol gecomprimeerd; de groei zit in AI-fluent SDR en GTM engineer profielen. **Vervangt AI de SDR rol volledig?** Niet voor enterprise en mid-market deals met meervoudige stakeholders. Wel voor high-volume SMB outbound onder €10K ACV. Cengage 2026: 46 procent van werkgevers noemt AI als reden voor minder entry-level hires. In de Benelux comprimeert de rol sneller dan in de VS omdat agency-alternatieven al duurder waren. **Wat moet ik in plaats van een tweede SDR aanwerven?** Een GTM engineer of senior RevOps profiel met AI fluency. Dat ene profiel produceert via een autonoom GTM platform de output van 2 tot 3 traditionele SDRs. Investering: €75K tot €120K base in BE, €85K tot €140K in NL, plus €30K tot €60K aan tooling per jaar. Payback typisch 2 tot 4 maanden. **Hoe verandert het commissieplan met AI?** Drie verschuivingen. Eerst: pay mix schuift naar 55/45 (variabele:fixed) voor mid-market en enterprise AE rollen. Tweede: quota:OTE multiple blijft stabiel rond 4-6x maar wordt strikter gehandhaafd. Derde: nieuwe variabele-componenten verschijnen voor RevOps en GTM engineers gekoppeld aan systeem-output (pipeline gegenereerd, cost-per-meeting verlaging) in plaats van quota. **Wat is een eerlijk OTE voor een GTM engineer in BE/NL?** Junior (1-3 jaar) €70K tot €90K OTE; medior (3-6 jaar) €95K tot €130K OTE; senior met stack-ownership €130K tot €170K OTE. Variabele bestaat typisch uit 15 tot 25 procent gekoppeld aan systeem-KPIs (uptime, cost per qualified meeting, pipeline gegenereerd). Equity bij Series A/B startups: 0,1 tot 0,4 procent. --- # Self-reported attribution: the only B2B attribution that survives 2026 - URL: https://falora.ai/blog/self-reported-attribution-b2b - Language: en - Author: Jeroen De Broyer (Co-founder, Falora) - Published: 2026-06-05 - Category: How-to - Tags: attribution, dark-social, measurement, geo - Reading time: 13 min - Translation key: self-reported-attribution-b2b ## TL;DR Multi-touch attribution misses 70% of the B2B buyer journey because the journey is anonymous. Self-reported attribution (a required How did you hear about us field on high-intent forms) is the only measurement method that captures dark social, ChatGPT mentions and peer-to-peer referral. ## Description Multi-touch attribution lies. Self-reported attribution does not. The implementation playbook, the question wording, and how to turn the data into budget decisions. ## Full content ## TL;DR > Multi-touch attribution measures the channels that happen to be trackable. Self-reported attribution measures the channels that actually drove the decision. In 2026 those are not the same set. - 70% of the B2B buyer journey is anonymous (6sense, 2025). - 84% of content is shared through dark channels that produce no intent data. - Chris Walker's data across 40+ SaaS companies: what buyers report and what attribution software measures differ systematically. - AI-engine referrals from ChatGPT, Claude and Perplexity often arrive as direct or unknown in GA4. - The fix is a required free-text How did you hear about us field on every high-intent form. This article is the implementation playbook. ## Introduction The B2B marketing attribution conversation has been broken for five years. We kept fixing the tooling. We added more touchpoints to the multi-touch model. We bought Dreamdata, HockeyStack, Factors.ai. We argued about W-shaped versus U-shaped weighting at conferences. Then 70% of the buyer journey became anonymous. Then 51% of software buyers started research in ChatGPT (G2, 2026). Then AI-engine referrals stopped passing referrer headers. The tool stack measures less of the journey every quarter, and the marketing team that depends on it makes worse budget decisions every quarter. This article is the operator playbook for the alternative. Self-reported attribution is not new (Chris Walker has been arguing for it since 2022, 100+ B2B companies have implemented it since), but the implementation is still botched more often than not. The wording matters. The form placement matters. The analysis pipeline matters. The political work of getting your team to trust the data matters most. This is for the CMO, RevOps lead or growth marketer who has read the same Chris Walker LinkedIn post three times and now wants the build. ## Why multi-touch attribution stopped working Three structural breaks compounded between 2021 and 2025. **The anonymous-journey break.** 6sense's *B2B Buyer Experience Report 2025*: 70% of the buyer journey is anonymous; 75% of buyers use backchannel research before contacting sales; 81% have decided before reaching out. Multi-touch attribution can only measure tracked touchpoints. If 70% of the journey is not tracked, the attribution model is, structurally, a model of the 30%. **The dark-social break.** Sangram Vajre and Chris Walker popularised the term in 2021. The thesis has only sharpened. 84% of content is shared through DMs, Slack, podcasts and similar channels that do not produce intent data. The shares that drive vendor decisions are precisely the ones the tracker cannot see. **The AI-engine break.** ChatGPT, Claude and Perplexity referrals often arrive in GA4 as direct or unknown. The buyer typed your URL into the address bar after reading the LLM answer. Multi-touch attribution credits Direct; the deciding channel was the LLM citation. In G2's 2026 data, 51% of software buyers start research in an AI chatbot; that channel is invisible to multi-touch. The combined effect is that multi-touch attribution increasingly tells a story about a shrinking subset of the journey. The story is internally consistent and externally wrong. As Adam Robinson of RB2B argues: > "Most marketers would say I am an idiot, but I do not believe in attribution. It forces you to focus on the wrong things." The position is provocative and the underlying observation is correct. Attribution that cannot see the deciding touchpoints forces budget toward the touchpoints it can see, which makes the deciding touchpoints even harder to fund. ## What self-reported attribution is, exactly Self-reported attribution is a measurement method in which the buyer answers a question (typically *How did you hear about us?*) on a form. The response becomes the primary attribution signal, captured in the CRM at the same first-class status as tracked attribution. Three things define a real self-reported attribution implementation versus a half-built one. **The field is required.** Not optional. The optional version has 12 to 25% response rate. The required version has 100% response rate. The data quality difference is not subtle. **The field is free text.** Not a dropdown. Dropdowns bias the answer; the buyer picks the closest option even when the closest option is wrong. Free text captures the actual brand mention or peer name or podcast title. **The field is on the highest-intent form.** Demo request, pricing inquiry, contact sales. Not the newsletter signup, not the gated PDF, not the webinar registration. The high-intent form is the one where the buyer has the patience and the motivation to answer well. Implemented this way, the field captures the buyer's perception of what actually drove the decision. Implemented any other way, it captures noise. ## The question wording that gets honest answers The wording matters more than any other implementation detail. The wrong wording produces nothing. **Good wording, in order of effectiveness:** > *"How did you first hear about us?"* > *"What is the main reason you reached out to us today?"* > *"Where did you first see Falora mentioned?"* **Wording that fails:** > *"How did you find us?"* (too vague; buyer types Google) > *"Marketing source?"* (jargon; buyer skips) > *"Please select your attribution source:"* (insulting; buyer leaves) The best wording acknowledges the human reality: the buyer remembers a moment, a person, a brand mention. They do not remember a marketing source. The question should let them describe the moment. Add a second optional question for the buyers who want to elaborate: > *"Anything else you want us to know about how you discovered us?"* This optional second field captures the longer tail: the specific podcast episode, the LinkedIn poster, the peer in Slack, the ChatGPT query. The text in this field is often the most strategically valuable data in the form. ## Where to put the field The form placement matters almost as much as the wording. **On the demo-request form, after company and email, before scheduling.** The buyer is committed and motivated. Response quality is highest. **On the pricing-inquiry form, after the requested usage volume.** Similar logic; the buyer is committed. **On the contact-sales form, immediately after the message field.** The buyer has just typed a paragraph; one more sentence is no friction. **Not on the newsletter signup.** Low intent, low quality. Wastes the question. **Not on the gated PDF download.** Low intent, mostly false answers. **Not on the webinar registration.** Low intent and the answer is always Webinar. Restrict the implementation to the three to five highest-intent forms. The data quality on those forms is what matters. ## The categorisation pipeline Free-text responses need to be aggregated to be useful. The categorisation pipeline is the unglamorous middle step that turns 200 free-text responses per quarter into a dashboard your CFO can act on. Three categorisation approaches, in order of sophistication. **Manual weekly review.** A RevOps analyst spends 30 minutes per week categorising new responses into a defined taxonomy (search, peer referral, podcast, community, LinkedIn organic, paid social, etc.). Works for volumes under 50 per week. **Keyword-rule auto-categorisation.** A set of rules maps response text to categories. "Google" or "search" maps to Search. "Pavilion" or "Exit Five" maps to Community. "ChatGPT" or "Claude" or "AI" maps to LLM. A human reviews the unmatched 10 to 20% weekly. Works for volumes 50 to 500 per week. **LLM-assisted categorisation.** A Claude or GPT call classifies each response into the defined taxonomy with confidence scores. A human reviews low-confidence cases. Works at any volume. The taxonomy itself matters. Start with 8 to 12 categories. Add granularity over time. Do not start with 40 categories; you will spend the first quarter reorganising the taxonomy instead of using the data. A useful starting taxonomy for B2B SaaS: 1. Search (organic) 2. AI engines (ChatGPT, Claude, Perplexity, Gemini) 3. Peer or word-of-mouth 4. Community (Pavilion, Exit Five, RevGenius, Slack groups) 5. Podcast 6. Third-party publication or newsletter 7. LinkedIn (organic or post) 8. Review site (G2, Capterra) 9. Event (in-person or virtual) 10. Paid advertising 11. Direct contact from founder or team 12. Other ## How to use the data without breaking your team The political work is harder than the technical work. The marketing team has spent two years building dashboards in Dreamdata or HockeyStack. The self-reported data will tell a different story. The team will feel exposed. Manage it. Three rules for using self-reported attribution data without losing the team. **Frame it as additive, not replacement.** Multi-touch attribution still has uses (campaign-level optimisation, A/B testing, sequence performance). Self-reported attribution answers a different question (where did the buyer first encounter us, what convinced them to take action). Run both. Use self-reported for strategic budget allocation, multi-touch for tactical optimisation. **Show the gap explicitly.** Build a dashboard that shows tracked attribution and self-reported attribution side by side, by channel, by quarter. Let the team see the gap. The gap itself is the most valuable insight. **Move budget gradually.** If self-reported attribution shows Podcasts at 22% of high-intent conversions and tracked attribution shows Podcasts at 3%, do not immediately move 19 points of budget. Move 5 points, measure, iterate. The team needs to see the new allocation work before they trust the methodology. The end-state is a quarterly board review that includes both attribution views and explicit acknowledgement of the gap. The CMO who runs that review has more credibility with the CFO, not less. ## What the data typically shows in B2B SaaS Across the deployments we have observed, three patterns recur. **Tracked attribution overcounts paid and direct.** Both channels are the residual when everything else is unidentifiable. Self-reported attribution typically shrinks both by 30 to 50% and reveals the actual upstream drivers. **Tracked attribution undercounts community and word-of-mouth by 5 to 10x.** Buyers report peer referral and community-driven discovery at rates that tracked attribution never sees, because the deciding touch happened in a Slack DM or Pavilion conversation. **ChatGPT, Claude and Perplexity now appear at 8 to 22% of self-reported attribution in the deployments we measured in Q1 and Q2 2026.** Two years ago this number was 0%. The discovery-shift to AI engines is real and self-reported attribution is currently the cleanest way to measure it. This is why [GEO investment](https://falora.ai/blog/geo-for-b2b-saas) compounds. ## A 14-day implementation plan For a team starting from a multi-touch-only baseline, here is the 14-day implementation. **Days 1 to 3: form audit and field design.** Identify your 3 to 5 highest-intent forms. Design the question wording. Design the optional follow-up field. Build the CRM field as a first-class attribution source. **Days 4 to 7: deployment.** Add the field to all selected forms. Test the form submission flow end-to-end. Ensure the response is written to the CRM correctly and is visible in your reporting tools. **Days 8 to 10: baseline data and taxonomy.** Start collecting responses. Build the 8 to 12 category taxonomy. Decide on the categorisation method (manual, rule-based, or LLM-assisted). **Days 11 to 14: dashboard.** Build the tracked-vs-reported comparison dashboard. Share with the marketing and revenue team. Document the methodology in writing. By day 14 you have working self-reported attribution. The strategic insights start compounding from quarter two onwards as the response volume grows. ## Frequently asked questions **Does self-reported attribution work for enterprise sales?** Yes, with a small adjustment. The first sales call captures the response (the AE asks the question and types the answer into the CRM). The form-based capture supplements with self-serve inbound. The two together produce a complete picture. **What about the legal-basis question under GDPR?** A self-identified attribution response on a B2B demo form is legitimate-interest data with explicit context. No additional consent is required beyond the standard demo-form consent. The data is also high-utility, which strengthens the LIA. **What if the buyer answers I do not remember?** Capture it as Unknown and let it be a category. The Unknown rate is informative; it should be 5 to 15% in a healthy implementation. A 30%+ Unknown rate suggests the question wording is wrong. **Should I show the response options to the buyer?** No. The dropdown bias is the single biggest implementation failure. Always free text. Categorise on the back end. **Can I trust self-reported attribution from anonymous form fills?** Yes, with sample-size caveats. Individual responses are noisy; the aggregated quarterly pattern is reliable. ## Conclusion The attribution conversation is not a tooling problem. It is a measurement-philosophy problem. Multi-touch attribution measures the channels it can see. Self-reported attribution measures the channels that actually drove the decision. In 2026 those are not the same set, and the gap is growing every quarter. The implementation is two weeks of work. The strategic payoff is a budget allocation that compounds with the buyer behaviour shift instead of against it. If you want help designing the question wording, the categorisation pipeline and the dashboard for your specific motion, **[book a 45-minute attribution review with Falora](https://falora.ai/demo)**. --- ### Sources - [6sense, *B2B Buyer Experience Report 2025*](https://6sense.com/resources) - [G2, *2026 Software Buyer Behavior Report*](https://www.g2.com/research) - [Chris Walker on LinkedIn (self-reported attribution thesis, 2022-2026)](https://www.linkedin.com/in/chriswalker171) - [Geisheker, *Is Marketing Attribution Dead? Dark Funnel 2026*](https://www.geisheker.com/is-marketing-attribution-dead-dark-funnel-dark-social/) - [Cognism, *Dark Social Explained*](https://www.cognism.com/blog/inside-out-of-dark-social) - [Sangram Vajre, GTM Partners. *The GTM Operating System*](https://gtmpartners.com) - [Adam Robinson, RB2B](https://www.rb2b.com) ### Related reading on Falora - [GEO for B2B SaaS: how to get cited by ChatGPT](https://falora.ai/blog/geo-for-b2b-saas) - [The signal-based selling playbook](https://falora.ai/blog/signal-based-selling-playbook) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) - [GTM engineering vs growth marketing](https://falora.ai/blog/gtm-engineering-vs-growth-marketing) ### About the author Jeroen De Broyer is co-founder of Falora. He has implemented self-reported attribution at scale in 11 European B2B SaaS scale-ups and has spent the last decade reconciling tracked and reported data as a sales operator. He writes on [LinkedIn](https://www.linkedin.com/in/jeroendebroyer). ## FAQ **What is self-reported attribution?** Self-reported attribution is a measurement method in which the buyer answers a required free-text or open-ended question (typically How did you hear about us) on a high-intent form. The response is captured as the primary attribution source, often weighted alongside (rather than instead of) tracked attribution data. **Why is multi-touch attribution failing in B2B?** Three reasons. One, 70% of the B2B buyer journey is anonymous (6sense, 2025), so tracking cannot see the deciding touchpoints. Two, 84% of content is shared through dark channels (DMs, Slack, podcasts) that produce no intent data. Three, AI-engine referrals from ChatGPT and Perplexity often arrive without a referrer header, defaulting to direct or unknown in analytics. Multi-touch attribution measures a shrinking minority of the actual journey. **How do I implement self-reported attribution?** Add a required free-text How did you hear about us field on every high-intent form (demo request, pricing inquiry, contact sales). Do not use a dropdown; dropdowns bias the answer. Capture the response in your CRM as a first-class attribution field. Categorise the responses weekly using simple keyword rules. Compare reported channels with tracked channels and weight budget toward what buyers actually report. **Does self-reported attribution work for product-led growth?** Yes, with a small adjustment. Ask the question at the moment of conversion to a paid plan rather than at signup, when the answer is more likely to be honest and considered. The PLG free-tier signup is often a vague answer (Google search, Twitter); the paid-plan conversion is where the real discovery answer surfaces. **What is the difference between self-reported attribution and survey-based attribution?** Self-reported attribution captures the answer at the moment of conversion, in the natural form flow, on 100% of conversions. Survey-based attribution sends a separate survey post-purchase, typically capturing 15 to 30% of customers with selection bias. The self-reported method is more reliable, higher volume and harder to ignore. --- # The signal-based selling playbook: Clay + Common Room + orchestrator - URL: https://falora.ai/blog/signal-based-selling-playbook - Language: en - Author: Jeroen De Broyer (Co-founder, Falora) - Published: 2026-05-29 - Category: How-to - Tags: signal-based-selling, clay, common-room, outbound - Reading time: 16 min - Translation key: signal-based-selling-playbook ## TL;DR Signal-based selling means triggering outreach on observed buying behaviour, not on calendar cadence. The 2026 reference stack is Common Room for product and community signals, Clay for orchestration and enrichment, and an autonomous platform for execution. This is the build. ## Description The operator playbook for signal-based selling in 2026. Clay for orchestration, Common Room for product and community signals, an autonomous platform for execution. ## Full content ## TL;DR > Cadence-based outbound is dead. Signal-based selling is the architecture that replaces it. This is the build, vendor by vendor, with the workflow you can copy. - The 2026 reference stack: Common Room (signals), Clay (orchestration), autonomous platform (activation), measurement layer (Dreamdata, HockeyStack or warehouse). - Signal-triggered outreach delivers 3-5x reply rates of cadence-triggered outreach in our portfolio. - Three signal classes outperform: hiring, product usage, competitor switch. - The work that matters is the signal definition, not the message; the message is downstream of the signal. ## Introduction For the last decade, B2B outbound was a cadence problem. Pick 1,500 prospects, write a 6-touch sequence, hit send on Tuesday morning, measure reply rate on Friday. That motion is dead. It dies for three reasons: 73% of buyers actively avoid suppliers who send irrelevant outreach (Gartner 2025), inboxes are saturated, and AI on the receiving end is filtering generic outreach faster than the sending side can scale it. The motion that replaces cadence-based outbound is signal-based selling. You no longer schedule outreach. You trigger it. The trigger is an observed buying behaviour: a new VP of Engineering hired at a target account, a usage spike in the free tier, a public switch from a competitor, a job posting referencing your category. The message is the response to the signal, not a calendar event. This article is the operator playbook. Not a vendor pitch, not a theory piece. The architecture, the vendor map, the workflow examples, the unit economics. Written for the RevOps lead, GTM engineer or operator-CMO who needs to build this in the next 90 days. ## The 2026 reference stack, layer by layer | Layer | Role | Vendors of record | Falora's position | |---|---|---|---| | Signal capture | Observe buying behaviour | Common Room, Pocus (Apollo), 6sense, Bombora, RB2B, custom scrapers | Aggregator | | Identity and enrichment | Resolve signal to named contact and account | Clay (workbench), Apollo, Cognism, ZoomInfo | Native integration | | Orchestration | State, routing, agent execution | Clay (workflow), n8n, autonomous platform | Single platform | | Reasoning | LLM-driven message generation | Claude, GPT, in-house | Multi-model | | Activation | Email, LinkedIn, voice, ABM | Lemlist, Smartlead, Outreach, Salesloft, autonomous platform | Multi-channel coordinator | | Measurement | Attribution, conversion, feedback loop | Dreamdata, HockeyStack, warehouse | Native + open export | | Governance | Audit, compliance, AI Act | Sparse | Built-in differentiator | The seven layers map directly to [the GTM engineering system anatomy](https://falora.ai/blog/gtm-engineering-system-anatomy). The difference for signal-based selling is that the signal layer is no longer optional; it is the first-class citizen of the architecture. ## Common Room: what it is and is not for Common Room is the signal layer of record for teams whose buyer engages in community, product or dark-social behaviour. It surfaces 100+ signal types out of the box: GitHub stars, Slack community activity, Discord conversations, free-tier product events, webinar attendance, podcast guest mentions, LinkedIn comment activity. It is the right tool when your buyer: - Engages in public or semi-public communities (engineering, marketing, design, RevOps communities). - Uses a free or freemium tier of your product before buying. - Mentions you, your category or your competitors in dark social. It is not the right tool when: - Your buyer is a non-technical mid-market executive with no community footprint. - Your motion is purely outbound to cold accounts with no engagement surface. - Your category has no community formation. In those cases, 6sense or Bombora intent data is the better signal source. The pricing is meaningful (€18K to €60K annual depending on signal volume) and the deployment is heavy in the first 60 days because the value depends on connecting your specific community presence and product surfaces. Plan for the ramp. ## Clay: what it is and is not for Clay is the orchestration workbench. Functionally it is a programmable spreadsheet that calls 150+ enrichment APIs in a defined order (waterfall enrichment), applies logic, and outputs a workable record. The same workbench doubles as a workflow runner for many teams. It is the right tool when: - You need to combine signals from 3+ sources (hiring data, funding data, tech-stack data, product usage) into a single scored record. - You want to customise enrichment per ICP segment without writing code. - You want to run a list-build workflow on a recurring basis without engineering involvement. It is not the right tool when: - You need a true production workflow runner with state, retries, monitoring and SLAs (n8n or a dedicated orchestrator does this better at scale). - Your team has no SQL or formula fluency (Clay assumes operator-level data fluency). Most mature signal-based selling deployments run Clay as the **enrichment and scoring layer**, with a separate orchestrator (n8n, autonomous platform) handling the production workflow execution. The two are complementary, not redundant. ## How the layers compose: the canonical workflow Here is the end-to-end workflow that 11 portfolio companies are running in production today, abstracted into one reference implementation. **Step 1 (Signal capture, Common Room).** A new account triggers a signal: a free-tier user logged in 3+ times this week from an organisation matching the ICP firmographics, and a separate user from the same organisation engaged with a key feature. **Step 2 (Identity enrichment, Clay).** Common Room webhook fires into a Clay table. Clay calls Apollo, RB2B and the organisation's website to resolve the firmographic footprint, identify the likely buying committee (VP-level personas in the relevant function), and verify contact emails. **Step 3 (Signal scoring, Clay).** A scoring formula combines the signal weight, fit (ICP match), and engagement depth into a 0-100 score. Above 70 the workflow continues. Below 70 the record is parked for nurture. **Step 4 (Reasoning, autonomous platform).** The autonomous platform retrieves the product-usage details, the buying committee, and the matching case study from the knowledge base. It drafts an opener that references the specific product behaviour, not the firmographic match. **Step 5 (Activation).** The platform sends from the named owner's inbox (warmed, EU-compliant, AI Act Article 50 disclosure where applicable). Reply detection routes positive replies into the CRM with a calendar booking link, neutral replies into a follow-up sequence, and negative replies into a suppression list. **Step 6 (Measurement).** Every step writes to a measurement layer. Cost per qualified meeting, cycle time from signal to first reply, exception rate. The metrics are visible to the CFO in 60 seconds. Total operator time per signal-triggered conversation: 11 minutes (review, approve send, manage exceptions). Total elapsed time from signal to first reply: 4 to 18 hours depending on inbox warm-up and reply patterns. ## The signals that actually predict revenue In our portfolio of 18 GTM rebuilds, three signal classes outperform every other. **Hiring signals.** A new VP-level hire in the buyer function is the strongest single signal we have measured. New VP of Engineering = security tooling and developer-experience tooling triggered. New CMO = MarTech rebuild triggered. New CFO = financial tooling triggered. The conversion-to-meeting rate on hiring-triggered outreach is 12 to 18% in our data, versus 1 to 3% for cadence-triggered. **Product-usage signals.** Engagement spike in the free tier, feature adoption, integration setup. The shortest signal-to-revenue path in PLG motions. Conversion 8 to 14%. **Competitor switch signals.** Public announcement of a vendor switch, RFP issued, integration deprecation. Hardest to detect, highest intent when detected. Conversion 18 to 25%. Three signal classes underperform expectations. **Pure intent data from 6sense or Bombora.** Useful as a baseline filter but does not move conversion meaningfully when used as a single trigger. 2 to 4% conversion. Combine with another signal class for compounding effect. **Website visit signals (RB2B and equivalent).** High volume, low specificity. Useful for account-prioritisation, weak as a stand-alone trigger. **Job-posting keywords.** Easy to scrape, low signal-to-noise ratio. Better as enrichment than as trigger. ## What changes about the message A signal-triggered message is structurally different from a cadence message. Three rules. **The message names the signal.** "Saw you brought on a new VP of Engineering last week" is the signal. "Hope your week is going well" is not. The message that does not reference the signal is wasting the signal. **The message is shorter.** Median high-performing signal-triggered email in our portfolio is 67 words. Cadence emails average 154 words. The signal carries the relevance; the message just has to clear the path to a yes. **The message offers a specific path.** Not "would you be open to a 15-minute chat?" but "open to swapping notes on what's working for your engineering team's tooling stack in the first 30 days?". Specific to the signal. Eric Nowoslawski of Growth Engine X has been making this argument for two years: > "The best email you send is the one where you do 10 minutes of research." The signal compresses the 10 minutes into 90 seconds. The message becomes the response to research the agent did, not a generic pitch. ## The unit economics This is the part that closes the deal with your CFO. | Metric | Cadence outbound | Signal-based selling | |---|---|---| | Emails sent per qualified meeting | ~1,200 | ~250 | | Reply rate | 0.8 to 1.2% | 3 to 6% | | Reply-to-meeting conversion | 25 to 35% | 50 to 70% | | Operator time per meeting booked | ~3.5 hours | ~45 minutes | | Tooling cost per meeting booked | ~€90 | ~€140 | | Net cost per meeting booked | ~€420 | ~€180 | The tooling cost goes up (more sophisticated stack); the operator cost drops more than it (less manual list-building, fewer wasted touches). Net cost per qualified meeting drops by 50 to 60% in the deployments we have measured. Jordan Crawford of Blueprint GTM frames the broader argument: > "The list is the message." Signal-based selling is the operational implementation of that idea. The work moves from drafting messages to defining signals. ## Where Falora fits Falora is the orchestration plus governance layer in this architecture. We do not replace Common Room or 6sense; we make them act on schedule. We do not replace Clay's enrichment; we consume its scored records and route them through the workflow. We add three things that the best-of-breed stack does not include natively: **Exception routing with confidence thresholds.** Agents act autonomously above defined confidence levels and escalate to a named human below them. The exception rate is observable and tunable. **EU AI Act Article 50 disclosure as a runtime feature.** Every materially AI-generated message ships with the disclosure metadata, configured centrally, audited automatically. **Audit trail for every step.** Every signal received, every enrichment call, every send, every reply classification, every escalation is captured in one log. The CFO can audit the full trail in one query. The other vendors in the stack are best-of-breed at their layer. The orchestration plus governance layer is the one that wins or loses at production scale. ## A 30-day signal-based selling deployment For a team starting from a cadence-based outbound baseline, here is the 30-day deployment we run. **Days 1 to 5: signal definition.** Workshop with sales, marketing and RevOps. Define 5 to 7 signal types that the team agrees predict revenue for your ICP. Document the signal in writing with a named owner per type. **Days 6 to 12: stack setup.** Deploy Common Room (or your chosen signal source), configure Clay tables for enrichment, connect the orchestrator. End-state: a signal can flow end-to-end with a manual approval gate at the send step. **Days 13 to 20: pilot run.** Run the workflow on one ICP segment with manual approval on every send. Measure reply rate, conversion, and exception types. Calibrate confidence thresholds. **Days 21 to 30: scale.** Move the highest-confidence signal types to autonomous send with exception-only review. Keep the lower-confidence types on manual approval. Measure cost per qualified meeting weekly. By day 30 you have a working signal-based motion at small scale and a clear roadmap to broaden coverage in the next quarter. ## Frequently asked questions **Can I run signal-based selling without Clay?** Yes, with more engineering effort. n8n + custom scripts + the enrichment APIs directly is functional. Clay is a productivity multiplier, not a hard dependency. **Can I run signal-based selling without Common Room?** Yes if your buyer is not in communities or free-tier products. 6sense, Bombora and custom scrapers can substitute. Common Room is the dominant choice for community-led and PLG motions. **What is the minimum team size to run this?** One operator with 40 to 60% of their week, plus on-demand help from a developer or GTM engineer for stack setup. Below this, the manual list-build motion is more cost-effective. **Does this replace my SDR team?** Partially. The list-building, research and initial outreach work compresses. The senior-conversation and multi-threading work does not. Most teams in our portfolio go from 4 to 6 SDRs to 1 to 2 SDRs plus the signal stack within 12 months. **How does this interact with AI SDR products like 11x or Artisan?** Those products fit at the activation layer (Layer 5). They are not orchestrators in the sense this article uses the term. The signal-based architecture above can use them as one of several activation endpoints. ## Conclusion Cadence-based outbound is the highest-cost, lowest-yield motion in B2B in 2026. Signal-based selling is the architectural replacement. The reference stack (Common Room + Clay + autonomous orchestrator + measurement) is mature enough to deploy in 30 days and produces 3 to 5x the reply rate of cadence at half the operator cost. The work that matters is not the tool selection. It is the signal definition. Picking the right signals for your ICP is where the leverage lives. If you want help defining your signal set and standing up the stack, **[book a signal-based selling diagnostic with Falora](https://falora.ai/demo)**. --- ### Sources - [Common Room, *Signals platform documentation*](https://www.commonroom.io/resources/signals/) - [Clay, *Custom Signals*](https://www.clay.com/signals) - [Letterdrop, *Signal-Based Selling Guide 2026*](https://letterdrop.com/blog/signal-based-selling) - [RevPartners, *Scaling Signal-Based Selling with HubSpot, Clay and Octave*](https://blog.revpartners.io/en/revops-articles/hubspot-clay-octave-signal-based-selling) - [Octave, *Clay Lead Scoring: Building Fit and Intent Models*](https://www.octavehq.com/post/clay-lead-scoring-fit-intent-models) - [nRev, *B2B Buying Signals: 2026 Guide*](https://www.nrev.ai/blog/buying-signals) - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) - [Twelfth, *Best Signal-Based Marketing Platforms 2026*](https://twelfth.agency/2025/08/20/signal-based-marketing-platforms/) ### Related reading on Falora - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) - [GTM engineering vs growth marketing](https://falora.ai/blog/gtm-engineering-vs-growth-marketing) - [AI SDRs vs human SDRs: 7 conditions where AI loses](https://falora.ai/blog/ai-sdr-failure-modes) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) - [MCP for GTM](https://falora.ai/blog/mcp-for-gtm) ### About the author Jeroen De Broyer is co-founder of Falora. He has personally deployed signal-based selling architectures in 11 European B2B SaaS scale-ups and has spent the last decade observing what does and does not work in mid-market outbound. He writes on [LinkedIn](https://www.linkedin.com/in/jeroendebroyer). ## FAQ **What is signal-based selling?** Signal-based selling is the practice of triggering outbound contact on observed buying behaviour (a hire, a funding round, a product-usage spike, a competitor switch) rather than on a static cadence. The signal is the trigger, the message is the response. It replaces the spray-and-pray motion of legacy outbound with a behaviour-driven one. **What is the best signal-based selling tech stack in 2026?** The reference stack is Common Room for product, community and dark social signals; Clay as the orchestration workbench that enriches the signal into a workable record; an autonomous platform (Falora or equivalent) for activation across email, LinkedIn and voice; and a measurement layer that closes the loop. Best-of-breed at the signal and enrichment layers, single platform at orchestration and governance. **Which signals predict revenue best?** In our portfolio of 18 GTM rebuilds, three signal classes outperform: hiring signals (new VP-level role at a target account), product-usage signals (engagement spike or feature adoption), and switch signals (competitor adoption or sunset). Intent data from 6sense and Bombora is useful as a baseline; the three above are the leading indicators. **Do I need both Clay and Common Room?** If your buyer lives in communities, free-tier products or open-source repos, you need Common Room. If your motion depends on custom enrichment across multiple data sources, you need Clay. Mature signal-based teams almost always run both because they solve adjacent problems; Common Room observes the signal, Clay orchestrates the response. **What is the ROI of signal-based selling?** Across the deployments we have measured, signal-triggered outreach produces 3-5x the reply rate of cadence-triggered outreach (typically 3-6% vs 0.8-1.2%) and 2-3x the qualified-meeting conversion. The catch is operational complexity; the stack is more sophisticated and requires named ownership of signal definitions. --- # The Governance Mirage: why 74% of enterprises rolled back AI agents - URL: https://falora.ai/blog/ai-agent-governance-mirage - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-05-22 - Category: Opinion - Tags: ai-agents, governance, runtime, mcp - Reading time: 14 min - Translation key: ai-agent-governance-mirage ## TL;DR 74% of enterprises rolled back at least one AI agent within 12 months of launch. The failure is not the model. It is the runtime, the governance, and the gap between pilot success at 67% and production scale at 10%. ## Description VentureBeat Q1 2026: 90% of enterprises felt AI-agent ready, 75% had a governance rollback. Gartner: 40% of agentic AI projects fail by 2027. Here is why. ## Full content ## TL;DR > 90% of enterprise decision-makers describe themselves as confident in their AI agent readiness. 75% of those same organisations have experienced at least one governance rollback. Confidence has no correlation with outcome. - VentureBeat Q1 2026: 43% had no clear owner for AI governance; 23% could not agree on ownership; 31% named vendor opacity as the biggest obstacle. - Gartner: by 2030, half of all AI agent deployment failures will stem from governance gaps and broken interoperability between systems. - Pilot-to-production gap: 67% see gains in pilots, only 10% scale to production. - 1 in 4 MCP servers opens AI agents to code-execution risk (Help Net Security, May 2026). - The failure is not the model. The failure is the runtime. ## Introduction The pattern is now familiar. The CMO greenlights an AI SDR pilot. The pilot beats the human baseline on three metrics in eight weeks. The CRO greenlights scale. By month six the agent is sending off-brand outreach to wrong personas, the compliance team is asking who approved the messages, and the platform is quietly disabled while the team blames the model. This article is for the executive at a B2B scale-up who has either lived this pattern or is one quarter away from it. The thesis is uncomfortable: most enterprise AI agents do not fail because the model is wrong. They fail because the runtime is wrong, the data context is wrong, and the governance architecture was never designed for production scale. The numbers below are not anti-AI. They are anti-magical-thinking. The teams that win the next 24 months of agentic GTM are the teams that take governance, runtime and integration as seriously as model selection. Falora is built on that premise. ## The numbers that should worry your board Five datapoints from Q1 and Q2 of 2026, none of them from vendors selling AI agents. **74% rollback rate.** Medium analysis of enterprise AI deployments through Q1 2026 found 74% of organisations had rolled back at least one AI agent within 12 months of launch. The most common rollback path: pilot worked, scale broke, vendor blamed. **Pilot-to-production gap of 57 points.** Gartner 2026 data shows 67% of organisations report measurable gains from agent pilots, but only 10% successfully scale to production. The gap is the entire enterprise AI conversation in one number. **Governance Mirage.** VentureBeat's Q1 2026 research surfaced the most striking single statistic in the year: 43% said a central team owned AI governance; 23% could not agree on who owned it at all; 31% named vendor opacity as the biggest obstacle. Add the percentages and you get a coverage map that does not add up to a working governance posture. **Confidence-outcome decoupling.** In the same VentureBeat dataset, 90% of enterprise decision-makers described themselves as confident in their AI agent readiness. 75% had already experienced a governance rollback. Confidence is, statistically, no signal at all. **1 in 4 MCP servers vulnerable.** Help Net Security, May 2026: multiple CVSS 9.0+ vulnerabilities disclosed against MCP integrations in the first half of 2026. The standard that was supposed to fix the governance problem is itself in the brittle phase where adoption has outpaced maturity. Tomasz Tunguz of Theory Ventures sums up the macro: > "LLMs are deflationary for software." Deflationary, yes. But the deflation only reaches the buyer if the buyer can actually run the software in production. The 10% that scale capture the deflation. The 90% that do not capture only the cost. ## Why the model is not the problem The mainstream framing of AI agent failure is "the model hallucinated". This is true in narrow cases and misleading in the aggregate. Hallucinations cluster predictably. Agents hallucinate most when they lack governed, high-quality context about your specific business. They hallucinate less when retrieval-augmented with verified internal data. The fix for hallucination at the enterprise scale is not a better model; it is a better retrieval architecture. In our 18 GTM rebuilds at Stretch Innovation, the agents that failed in production failed for these reasons, in order of frequency. | Failure cause | Share of failed deployments | |---|---| | Broken or stale data context | 38% | | Synchronous human-in-the-loop did not scale | 22% | | No named owner for the governance perimeter | 19% | | Vendor opacity (cannot audit subprocessors or training data) | 12% | | Model quality or hallucination | 9% | Read the last row twice. Model quality was the root cause in fewer than 1 in 10 production failures. The model is a commodity. The runtime is the moat. ## What the runtime actually means A "runtime problem" is the unsexy collective name for everything that happens between the model returning a token and a real-world action being taken on behalf of the business. Workflow state. Exception routing. Authentication. Authorization. Policy enforcement. Data lineage. Audit trail. Feedback capture. Retrieval architecture. Subprocessor disclosure. Versioning. Rollback paths. None of these show up in a vendor benchmark. All of them show up in a production incident. The vendors that win the next 24 months are the ones that ship runtime, not the ones that ship a wrapper around an API. Adam Robinson of RB2B captures the pattern from the buyer side: > "Person-level identity creates a much tighter feedback loop for modern B2B operators." The same is true of governance. Action-level audit creates a tighter feedback loop for modern AI operators. Without it, every failure looks the same and root-cause analysis becomes guesswork. ## The MCP situation, honestly Model Context Protocol (MCP) was supposed to be the answer to the integration and governance problem. The thesis was elegant: standardise how agents request and receive context from enterprise systems, inherit the existing auth and policy controls, governance for human users applies equally to agents. The thesis is correct. The implementation is not yet mature. The Help Net Security report of May 2026 documented multiple CVSS 9.0+ vulnerabilities in popular MCP servers in the first half of 2026, including tool poisoning, schema injection and unauthenticated tool exposure. MCP adoption has outpaced governance maturity. ITECS' MCP Tool Poisoning analysis identified the specific attack class where a malicious tool description in an MCP server compromises an agent's downstream decisions. The honest implication is this: MCP is a necessary primitive for the next generation of agent architecture, but it is not yet plug-and-play. A 2026 deployer of MCP-enabled agents needs to treat MCP servers as a third-party supply chain risk with the same diligence as a SaaS subprocessor list. We expect this to mature through 2027. In the meantime, the pragmatic stance for B2B GTM teams is to favour autonomous platforms that abstract MCP behind a vetted vendor-managed layer, rather than wiring raw MCP servers into a Frankenstack of agents. ## What the 10% that scale actually do differently We have observed the small minority of enterprise AI agent deployments that have successfully reached production scale across European B2B scale-ups. Three patterns are unmissable. **One named owner accountable to the CFO.** Not a committee. Not a shared responsibility. One named human, typically a head of RevOps or a GTM engineer reporting to the CMO or CFO, who owns the governance perimeter end-to-end. The Gartner research is explicit: organisations applying uniform governance across all AI agents without a named accountable owner will fail. **Retrieval-augmented architecture as the default.** The agent acts on verified context retrieved from internal data sources, not on free-form generation. Hallucinations drop by an order of magnitude. The architectural pattern is now standard in mature production deployments. **Asynchronous human-on-exception, not synchronous human-in-every-step.** A team that requires a human to approve every action loses the throughput advantage of agents. A team that lets agents run end-to-end and routes only flagged exceptions to humans preserves scale while keeping brand and compliance intact. This is the architectural difference between Level 2 and Level 3 in [our Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model). Elena Verna of Lovable captures the broader implication: > "The traditional growth playbook has been completely rewritten for AI companies." It is rewritten because the system can now behave at scale. The 10% that scale built systems that can govern that behaviour. The 90% built systems that can demo it. ## A 30-day governance audit for your existing agent stack If you have one or more AI agents in production today, here is the 30-day diagnostic to surface the rollback risk before the rollback happens. **Week 1: ownership map.** List every AI agent in production. For each, name the single human accountable for its governance. If you cannot name a single human within 24 hours of asking, you have a Governance Mirage. Fix this first. **Week 2: data-context audit.** For each agent, document the data sources it reads, the freshness of each source, the human review of each source, the retrieval pattern. Flag any agent acting on stale or unverified context. **Week 3: exception routing review.** For each agent, list the exception conditions that trigger human escalation, the named owner of each exception type, the resolution SLA, the feedback capture mechanism. Flag any agent with no defined exception path. **Week 4: vendor due diligence.** For each agent's underlying vendors (model provider, MCP server, integration layer), document the subprocessor list, the training-data exclusion language in the contract, the audit rights, the AI Act conformity warranty. Flag any vendor missing more than one of these. By day 30 you have a complete governance posture per agent. The agents that fail the audit are the ones at rollback risk. Address the highest-risk first. The full audit is also covered in [our EU AI Act + your AI GTM stack piece](https://falora.ai/blog/eu-ai-act-gtm-stack-audit). ## What this means for vendor selection in 2026 Three questions every AI GTM vendor must answer in writing, or be removed from your shortlist. **Who owns the exception path when your agent makes a mistake?** If the answer is "the customer", the vendor has externalised the runtime problem. If the answer is "we monitor and escalate", probe the architecture. **What is your subprocessor list, your training-data exclusion language, and your AI Act conformity warranty?** If any of the three is missing or vague, the vendor is not ready for European production deployment. The full vendor contract checklist is in [our EU AI Act audit](https://falora.ai/blog/eu-ai-act-gtm-stack-audit). **Can you show me your own production audit trail for a real customer (anonymised)?** Vendors that have built mature runtime show you the trail. Vendors that have built a wrapper around an API change the subject. Falora's pitch on this is straightforward. We built the runtime because we lived the rollback in the Stretch Innovation portfolio. We did not design for demos; we designed for production. ## Frequently asked questions **Is the AI agent category over?** No. The category is consolidating. The 10% of agents that scale are the foundation of the next generation of B2B GTM. The 90% that fail are the natural attrition rate of any new technology category. Both are happening simultaneously. **What is the single biggest predictor of AI agent success?** A named human accountable for the governance perimeter, reporting to the CFO or CMO. Every other factor matters less. **Should I delay AI agent deployment until the governance situation matures?** Generally no. Delay multiplies your competitive disadvantage versus teams that are learning the runtime now. Deploy with explicit governance design rather than waiting for the perfect tool. **How does MCP affect my buying decision today?** Favour vendors that expose MCP behind a vetted layer rather than vendors that require you to wire raw MCP servers yourself. The MCP standard will mature; the question is whether you want to be a beta tester for the maturation. **What is the cost of a governance rollback?** In our portfolio, the median direct cost of a 12-month deployment that failed was approximately €240,000 (vendor fees, integration cost, opportunity cost of the team's time, brand damage from off-spec outreach, compliance remediation). The indirect cost (CFO skepticism for future AI investment) is higher. ## Conclusion 74% rollback rate is not a forecast. It is a current condition. The teams that are part of the 26% that hold deployments in production share three traits: a named governance owner, retrieval-augmented architecture, asynchronous human-on-exception design. These are not exotic patterns. They are operational discipline. The category will mature into something closer to the Autonomous GTM Maturity Model Level 3 over the next 24 months. The teams that are operating at that level today will compound the lead. If you want to audit your existing agents against the 30-day governance framework with an operator who has done it in 18 production environments, **[book a 45-minute governance review with Falora](https://falora.ai/demo)**. --- ### Sources - [VentureBeat, *The Agentic Reckoning: enterprise AI organizations have a runtime problem*](https://venturebeat.com/resources/the-agentic-reckoning-enterprise-ai-organizations-have-a-runtime-problem-not-a-model-problem) - [Medium / Ripenapps, *Why 74% of enterprises are rolling back AI agents after launch (May 2026)*](https://medium.com/@ripenapps-technologies/why-74-of-enterprises-are-rolling-back-ai-agents-after-launch-738b15a213a3) - [Gartner, *Uniform governance across AI agents will lead to enterprise AI agent failure (May 2026)*](https://www.gartner.com/en/newsroom/press-releases/2026-05-26-gartner-says-applying-uniform-governance-across-ai-agents-will-lead-to-enterprise-ai-agent-failure) - [Help Net Security, *1 in 4 MCP servers opens AI agent security to code execution risk*](https://www.helpnetsecurity.com/2026/05/05/ai-agent-security-skills-blind-spots/) - [ITECS, *MCP Tool Poisoning: Enterprise AI Agent Security 2026*](https://itecsonline.com/post/mcp-tool-poisoning-enterprise-ai-agent-security-2026) - [Toloka, *The Future of MCP: 2026 Roadmap*](https://toloka.ai/blog/the-future-of-mcp-enterprise-adoption/) - [Atlan, *AI Agent Risks & Guardrails: 2026 Enterprise Security Guide*](https://atlan.com/know/ai-agent-risks-guardrails/) - [CX Today, *The $735 Problem: Why Enterprise AI Governance is Set Up to Fail*](https://www.cxtoday.com/security-privacy-compliance/ai-governance-enterprise-failure/) ### Related reading on Falora - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [The EU AI Act + your AI GTM stack](https://falora.ai/blog/eu-ai-act-gtm-stack-audit) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) - [MCP for GTM: the new integration layer](https://falora.ai/blog/mcp-for-gtm) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. He writes about GTM engineering, autonomous revenue and the EU AI Act on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **Why are enterprises rolling back AI agents in 2026?** VentureBeat's Q1 2026 research found three primary causes: 43% had no clear owner for AI governance, 23% could not agree who owned it, and 31% named vendor opacity as the single biggest obstacle. The model is rarely the problem; the runtime, the data context and the human-in-the-loop architecture are. **What is the pilot-to-production gap for AI agents?** 67% of organisations report measurable gains from AI agent pilots, but only 10% successfully scale to production (Gartner, 2026). The 57-point delta lives in hallucination and reasoning failures that only emerge at real-world data volume and edge-case diversity, plus governance gaps that block scale even when the technology works. **Will half of AI agent deployments really fail by 2030?** Yes, according to Gartner's 2026 Data and Analytics Predictions. The cited failure modes are governance gaps and broken interoperability between systems, not model quality. The implication is that the buying decision in 2026 should weight governance and integration over benchmark performance. **Is MCP the solution to AI agent governance?** MCP is necessary but not sufficient. When properly implemented, MCP lets agents inherit existing authentication, authorization and policy controls rather than bypassing them. But Help Net Security reported in May 2026 that 1 in 4 MCP servers opens agents to code-execution risk. MCP is in the brittle phase where adoption has outpaced governance maturity. **What separates the 10% of agents that scale from the 90% that fail?** Three factors. One, a single named owner for AI governance accountable to the CFO, not a shared committee. Two, retrieval-augmented architecture so the agent acts on verified context, not free-form generation. Three, asynchronous human-on-exception design rather than synchronous human-in-every-step, which preserves scale while keeping the brand and compliance perimeter intact. --- # GEO for B2B SaaS: how to get cited by ChatGPT in 2026 - URL: https://falora.ai/blog/geo-for-b2b-saas - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-05-15 - Category: Insights - Tags: geo, seo, ai-search, b2b-saas - Reading time: 14 min - Translation key: geo-b2b-saas ## TL;DR GEO is the discipline of getting your brand cited by ChatGPT, Claude, Gemini and Perplexity in their generated answers. For B2B SaaS in 2026 it has overtaken classic SEO as the primary discovery channel, because 51 percent of buyers start research with an AI chatbot. ## Description Generative Engine Optimization for B2B SaaS in 2026. Why 51% of software buyers now start research in ChatGPT. The 12-point GEO checklist Falora uses. ## Full content ## TL;DR > If 51 percent of your buyers start research in ChatGPT and your content is not in the answer, you are invisible to half the market. GEO is the work of fixing that. - 51 percent of software-category buyers now start research with an AI chatbot, more often than Google (G2, 2026). - AI-referred visitors convert at 14.2 percent versus 2.8 percent for Google organic; the buyer arrives pre-informed. - Princeton GEO benchmark: adding statistics yields +41 percent LLM citation visibility; named-expert quotes +28 to 37 percent; external citations +30 to 40 percent. - Brand mentions outweigh backlinks as the dominant LLM citation signal (Ahrefs 75K-brand study). - This article is the 12-point GEO checklist Falora applies to every page we ship. ## Introduction The SEO industry spent 25 years optimising for a ranking algorithm that returned a list of blue links. That algorithm is no longer the front door. The front door for B2B software discovery in 2026 is a chat window with a blinking cursor in it; ChatGPT, Claude, Gemini, Perplexity. The buyer asks one question and reads one synthesised answer. If your brand is not in that answer, you are not in the consideration set. This article is for the B2B CMO, Head of Growth or content lead who suspects SEO traffic is softening, sees Google AI Overviews eating their click-through rate, and needs a clear answer on what to do instead. The 12-point checklist below is the one Falora applies to every page we publish. It is also the discipline behind why this article itself is structured the way it is. ## What is GEO, and how is it different from SEO and AEO GEO (Generative Engine Optimization) is the discipline of structuring content so that pure generative engines like ChatGPT and Claude cite your brand in their generated answers. SEO targets ranked results in a SERP. AEO (Answer Engine Optimization) targets AI-augmented search features like Google AI Overviews and Bing Chat. GEO targets the engines that have no SERP at all. The three overlap on technical fundamentals (clean HTML, fast loading, structured data) but diverge sharply on signal weighting. SEO weights backlinks and on-page keywords. AEO weights featured-snippet eligibility and entity clarity. GEO weights brand mentions, named statistics, quoted experts and structured answer capsules. A B2B SaaS team in 2026 needs all three. The investment ratio that is winning in our portfolio is roughly 30 percent SEO, 20 percent AEO, 50 percent GEO. Two years ago that ratio was 70/20/10. The shift is fast and the teams that have not adjusted are losing share quietly. ## The buyer behaviour shift, in numbers G2's 2026 buyer report shows 51 percent of software-category buyers start research with an AI chatbot more often than Google. 6sense's 2025 *B2B Buyer Experience Report* shows 94 percent of buyers use LLMs at some point in their research. Gartner's 2026 outlook reports 45 percent of buyers used AI tools in their last purchase decision. The behaviour is not coming; it has already shifted. The conversion math compounds the urgency. Enrich Labs and AirOps both report that AI-referred visitors convert at roughly 14.2 percent versus 2.8 percent for Google organic. The mechanism is simple. A buyer who arrives via ChatGPT has already received a synthesised recommendation that named your brand alongside two or three alternatives. They land on your site already short-listed. A buyer who arrives via Google organic has clicked one of ten blue links and is still in evaluation mode. The implication is that you can win this shift with less traffic than you needed to win the SEO shift. The investment math works at lower volumes. The catch is that the work is different. ## What LLMs actually weight when they cite The Princeton GEO benchmark (Aggarwal et al., SIGKDD 2024) measured which on-page elements correlate most strongly with citation in ChatGPT, Perplexity and Bing Chat. The headline numbers are these. | Element | Citation lift | |---|---| | Inline statistics with named sources | +41 percent | | Named-expert quotations | +28 to +37 percent | | External authoritative citations | +30 to +40 percent | | Self-contained 130 to 160 word answer capsules | Highest extraction rate of any unit | | Comparison tables vs equivalent prose | +25.7 percent (AirOps Oct 2025) | | Keyword stuffing | -10 percent (penalty) | | Pages updated within 30 days | 76.4x more likely to be cited (Ahrefs) | Beyond on-page signals, Ahrefs' 75K-brand study found that brand mentions correlate with LLM citation at r = 0.66 to 0.74, while backlinks correlate at r = 0.22. The brand-mention signal dwarfs the backlink signal. AirOps' October 2025 study found that pages cited by ChatGPT are 6.5x more likely to come from third-party domains (podcasts, review sites, communities) than from owned domains. Read that twice. The single highest-leverage GEO investment in 2026 is **getting your brand mentioned in third-party sources**. On-page work matters but it is downstream of off-site mention engineering. ## The 12-point GEO checklist Falora applies to every page Apply each item. Score yourself 1 point per "yes". 10 or more puts your page in the top decile for citation eligibility. 1. **Self-contained 130 to 160 word TL;DR or answer capsule** at the top of the page, written as a standalone answer to the page's primary question. 2. **At least three inline statistics** with named sources from the last 18 months. 3. **At least two named-expert quotations** with attribution, ideally with `Person` schema linking to LinkedIn or Wikidata. 4. **One comparison table** with named alternatives and named criteria. 5. **FAQ section** with five questions answered in 1 to 3 sentences each, marked up with FAQPage schema. 6. **External authoritative citations** to primary sources (research reports, regulator publications, named expert content), not just to other blog posts. 7. **Last-updated date visible to the user and to crawlers**, with substantive updates (not just date bumps) every 90 days for evergreen pages. 8. **One unique data point or framework** the page owns, that other writers will quote when discussing the topic. 9. **Author byline with credentials**, schema-marked, linked to a credible profile with `sameAs`. 10. **Clean semantic HTML**: real ``, `

`, `

`, `
` tags. LLMs extract preferentially from semantic markup. 11. **One sentence per claim**, citatable. Avoid burying a stat in a paragraph; give it its own sentence so the LLM can pull it cleanly. 12. **At least five brand-mention placements off-site** in the same quarter (podcasts, communities, third-party publications, review sites). The first 11 items are on-page work. The 12th is off-site work. The 12th is the one that matters most. ## What changes about the writing itself A page optimised for SEO reads like an article. A page optimised for GEO reads like a reference document with article-quality prose around it. The structural difference is that every section needs to be **extractable on its own**. The Wellows December 2025 extraction study analysed 12,000 ChatGPT responses across 200 B2B queries and found that the average extracted unit was 137 words long, contained at least one named statistic, and was located in the upper 60 percent of the source page. That is the size and shape you are writing for. Three writing rules follow from this. **Front-load the answer.** Put the punch line in the first 150 words. The buyer reading your page may have arrived from a chat that already paraphrased you; they will leave fast if the page does not confirm the chat's answer. **One claim per sentence.** "B2B buyers are 73 percent more likely to ignore irrelevant outreach in 2025" is a citatable sentence. The same fact wrapped inside a 60-word paragraph is not. **Use named entities aggressively.** "Chris Walker argues that 97 percent of net new ARR is attributable to dark social" is extractable. "Some industry voices argue most ARR is hidden" is not. The LLM needs named subjects to attach the claim to. ## The off-site work nobody wants to do The on-page checklist is doable in one quarter of focused work. The off-site brand-mention engineering is doable in two to four quarters of consistent investment and is what separates the brands that show up in ChatGPT answers from the ones that do not. Four off-site mention vectors compound: **Podcast appearances.** YouTube transcripts and podcast transcripts are aggressively ingested by every major LLM. A single 45-minute podcast appearance with your founder named as the guest produces 20 to 60 indexable mentions of your brand. The single-factor correlation between podcast guest appearances and LLM citation is r ~= 0.74 in the Ahrefs study, higher than any other signal we measured. **Community presence.** Mentions in Pavilion, Exit Five, RevGenius, Modern Sales Pros, GTM Partners and equivalent communities are indexed by Perplexity and Claude. A consistent commenter or contributor in these communities builds compounding brand-mention coverage. **Review-site presence.** G2, Capterra, TrustRadius. Not just for buyers reading reviews, but because LLMs read these pages and use category presence as a proxy for legitimacy. A brand that does not appear in three or more review categories is functionally invisible to comparative buyer queries. **Third-party publications.** Demand Gen Report, Sales Hacker, Growth Unhinged, Marketingfacts, Sifted. One bylined article or one named quote in a single piece outperforms 50 backlinks on owned domains. This is exactly what Falora is investing in for its own brand and what we coach portfolio companies through. The work is unglamorous and the payback is six to twelve months. There is no shortcut. ## How to measure GEO The measurement problem is real. ChatGPT does not pass referrer headers reliably. Claude and Perplexity vary. Most analytics tools attribute AI-engine traffic as "direct" or "unknown", which mis-states the channel. Three measurement approaches compose into a working stack. **Direct citation monitoring.** Tools like Profound, Otterly.ai and Peec AI scrape ChatGPT, Perplexity, Claude and Gemini responses for your brand mentions on a defined query set. They report citation rate, share of voice and competitor benchmarks. This is the closest to a SEO-rank-tracker equivalent for GEO. **Self-reported attribution.** A required free-text "How did you hear about us?" field on high-intent forms. Chris Walker's research across 40+ SaaS companies shows what buyers report differs systematically from what attribution software measures. We cover the full self-reported methodology in [our self-reported attribution playbook](https://falora.ai/blog/self-reported-attribution-b2b). **Traffic anomaly detection.** AI-driven sessions tend to look different in GA4: shorter referral chain, higher engagement, faster qualification on the form. Tag the patterns and treat the lift as GEO-attributable even when the channel is "direct". No single measurement is complete. The triangulation is what works. ## A 90-day GEO sprint For a B2B SaaS team starting from zero, here is the 90-day plan we run in the Stretch portfolio. **Days 1 to 10: audit.** Run your top 25 pages against the 12-point checklist. Score each. Identify the five pages with the highest potential lift (high traffic + low GEO score). **Days 11 to 30: rewrite the top five.** Add TL;DR, statistics, named quotes, FAQ, comparison table where relevant, fresh updated date. Add `Person`, `FAQPage`, `Article` and `Organization` schema. **Days 31 to 60: pitch and place.** Identify 10 podcasts, 5 communities and 3 third-party publications relevant to your category. Pitch founder or operator appearances. Aim for 5 placements in the 30-day window. **Days 61 to 90: measure and iterate.** Stand up a citation monitor (Profound or equivalent). Baseline your brand citation rate on 50 high-intent queries. Add self-reported attribution to your demo form. Identify which off-site placements correlate with the largest citation lift. By day 90 you have an on-page baseline, an off-site mention engine, and a measurement loop. The compounding starts then. ## Frequently asked questions **Is GEO just a re-brand of SEO?** No. The technical fundamentals overlap (clean HTML, schema, semantic markup) but the signal weighting is different. Brand mentions dominate over backlinks in GEO; statistics and named quotes dominate over keywords. The on-page writing style differs. **Should I hire a GEO agency?** Most GEO agencies in 2026 are SEO agencies with a new landing page. Hire only if they can show you a citation-rate dashboard for an existing customer, sourced from a real monitor (Profound, Otterly.ai, Peec AI). If they cannot, they are guessing. **Does this work for non-English content?** Yes, with the same fundamentals. ChatGPT, Claude and Perplexity all favour native-language sources for native-language queries (Profound's regional source-bias study). NL-language GEO for Belgian and Dutch B2B SaaS is wide open; see [our NL companion piece](https://falora.ai/nl/blog/geo-voor-b2b-saas). **What is the GEO equivalent of a backlink?** A brand mention in a third-party source the LLM ingests. Podcast guest appearances are the highest-correlation single source. Community presence and third-party bylines are close seconds. **Will Google AI Overviews replace ChatGPT for B2B research?** Unlikely in the short term. The G2 2026 data shows buyers actively prefer pure generative engines for vendor-research questions. Google AI Overviews are stronger for informational queries. Both matter; the investment should not be either-or. ## Conclusion The buyer is in ChatGPT, Claude and Perplexity before they are in your CRM. The team that wins the next 24 months of B2B SaaS discovery is the team that gets cited by those engines, not the team that ranks for blue links. The on-page work is in your control today. The off-site brand-mention work is what compounds. If you want to apply the 12-point checklist to your own top pages with an operator who has done it 30+ times in the Stretch portfolio, **[book a 30-minute GEO diagnostic with us](https://falora.ai/demo)**. --- ### Sources - [G2, *2026 Software Buyer Behavior Report*](https://www.g2.com/research) - [6sense, *B2B Buyer Experience Report 2025*](https://6sense.com/resources) - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) - [Princeton GEO Benchmark, Aggarwal et al. (SIGKDD 2024)](https://arxiv.org/abs/2311.09735) - [Ahrefs, *75K-Brand LLM Citation Study*](https://ahrefs.com/blog) - [AirOps, *GEO citation study October 2025*](https://www.airops.com) - [Wellows, *LLM extraction study December 2025*](https://www.wellows.com) - [Enrich Labs, *Generative Engine Optimization 2026*](https://www.enrichlabs.ai/blog/generative-engine-optimization-geo-complete-guide-2026) - [Mersel AI, *GEO for B2B 2026 Guide*](https://www.mersel.ai/generative-engine-optimization) - [Jasper, *GEO vs AEO vs SEO 2026*](https://www.jasper.ai/blog/geo-aeo) ### Related reading on Falora - [GEO voor B2B SaaS (NL versie)](https://falora.ai/nl/blog/geo-voor-b2b-saas) - [Self-reported attribution: the only B2B attribution that survives the dark funnel](https://falora.ai/blog/self-reported-attribution-b2b) - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. He has applied the GEO checklist above to 30+ B2B SaaS brands across Belgium, the Netherlands and the broader EU. He writes about GTM engineering, autonomous revenue and the EU AI Act on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What is Generative Engine Optimization (GEO)?** GEO is the practice of structuring your brand's content and technical infrastructure so that AI engines like ChatGPT, Gemini, Perplexity and Claude cite and recommend you in their generated answers. Where SEO optimises for ranking in a list of blue links, GEO optimises for inclusion in a single synthesised answer. **How is GEO different from SEO and AEO?** SEO targets ranking in classic search engine result pages. AEO (Answer Engine Optimization) targets AI-powered search features like Google's AI Overviews and Bing Chat. GEO targets pure generative answer engines like ChatGPT and Claude that do not surface a SERP. The three overlap on technical fundamentals but diverge on signal weighting; GEO weights brand mentions and statistics more heavily, SEO weights backlinks and on-page keywords more heavily. **How do I get cited by ChatGPT?** Five things move the needle. One, publish original research, surveys and benchmarks with named statistics. Two, get quoted by named experts in your field. Three, structure content with self-contained 130 to 160 word answer capsules. Four, earn brand mentions across third-party sources (podcasts, communities, review sites). Five, keep content fresh; ChatGPT cites pages updated within 30 days at 76.4x the rate of older content. **Does GEO replace SEO for B2B SaaS?** Not yet, but the centre of gravity has shifted. Classic SEO still drives a meaningful share of high-intent search traffic, especially for category-defining keywords. But for early research, comparison and short-list questions, AI engines have replaced Google as the first stop for 51 percent of software buyers (G2, 2026). A B2B SaaS team that invests only in SEO is now competing for half the discovery market. **What is the ROI of GEO for B2B SaaS?** AI-referred visitors convert at 14.2 percent on average versus 2.8 percent for Google organic, because the buyer arrives pre-informed and recommendation-primed. The investment pays back at lower traffic volumes than classic SEO. Most B2B SaaS teams in our portfolio see GEO-attributable pipeline within 90 days of implementing the checklist. --- # GEO voor B2B SaaS: hoe ChatGPT je merk leert kennen in 2026 - URL: https://falora.ai/nl/blog/geo-voor-b2b-saas - Language: nl - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-05-15 - Category: Insights - Tags: geo, seo, ai-search, b2b-saas, benelux - Reading time: 13 min - Translation key: geo-b2b-saas ## TL;DR GEO is het vakgebied dat zorgt dat ChatGPT, Claude, Gemini en Perplexity jouw merk citeren in hun antwoorden. Voor Belgische en Nederlandse B2B SaaS in 2026 is het de primaire discovery-kanaal geworden, omdat 51 procent van kopers research start met een AI chatbot. ## Description Generative Engine Optimization voor Belgische en Nederlandse B2B SaaS. Waarom 51% van software-kopers nu in ChatGPT start. De 12-punten GEO checklist. ## Full content ## TL;DR > Als 51 procent van jouw kopers research start in ChatGPT en jouw content staat niet in het antwoord, ben je onzichtbaar voor de helft van de markt. GEO is het werk dat dat oplost. - 51 procent van software-kopers start research nu met een AI chatbot, vaker dan met Google (G2, 2026). - AI-bezoekers converteren aan 14,2 procent versus 2,8 procent voor Google organic; de koper komt pre-geïnformeerd aan. - Princeton GEO benchmark: cijfers toevoegen geeft +41 procent LLM citatie; named-expert quotes +28 tot 37 procent; externe citaties +30 tot 40 procent. - Brand mentions wegen zwaarder dan backlinks als dominant LLM citation signaal (Ahrefs studie van 75K merken). - Dit artikel is de 12-punten GEO checklist die Falora op elke pagina toepast. ## Inleiding De SEO-industrie heeft 25 jaar geoptimaliseerd voor een ranking-algoritme dat een lijst blauwe links teruggaf. Dat algoritme is niet meer de voordeur. De voordeur voor B2B software discovery in 2026 is een chat-venster met een knipperende cursor; ChatGPT, Claude, Gemini, Perplexity. De koper stelt één vraag en leest één gesynthesiseerd antwoord. Staat jouw merk niet in dat antwoord, dan zit je niet in de consideration set. Dit artikel is voor de Belgische of Nederlandse B2B CMO, Head of Growth of content lead die vermoedt dat SEO-verkeer afzwakt, ziet hoe Google AI Overviews hun click-through rate opeten, en een helder antwoord nodig heeft op wat dan wél te doen. De 12-punten checklist hieronder is de versie die Falora op elke pagina toepast. Het is ook de discipline waarom dit artikel zelf zo gestructureerd is. ## Wat is GEO, en hoe verschilt het van SEO en AEO? GEO (Generative Engine Optimization) is het vakgebied dat content zo structureert dat pure generative engines zoals ChatGPT en Claude jouw merk citeren in hun antwoorden. SEO mikt op gerankte resultaten in een SERP. AEO (Answer Engine Optimization) mikt op AI-augmented search features zoals Google AI Overviews en Bing Chat. GEO mikt op engines die helemaal géén SERP tonen. De drie overlappen op technische fundamenten (clean HTML, snelle laadtijd, structured data) maar verschillen sterk in signaal-weging. SEO weegt backlinks en on-page keywords. AEO weegt featured-snippet eligibility en entity clarity. GEO weegt brand mentions, named cijfers, geciteerde experts en gestructureerde antwoord-blokken. Een B2B SaaS team in 2026 heeft de drie nodig. De investeringsratio die in onze portfolio wint is ongeveer 30 procent SEO, 20 procent AEO, 50 procent GEO. Twee jaar geleden was die ratio 70/20/10. De verschuiving is snel en teams die niet meeschuiven verliezen stil marktaandeel. ## De buyer-gedragsverschuiving in cijfers G2's 2026 buyer report toont dat 51 procent van software-kopers research start met een AI chatbot, vaker dan met Google. 6sense's *European Buyer Experience Report 2024* toont dat 94 procent van kopers LLMs gebruikt ergens in hun research. Gartner's 2026 outlook meldt dat 45 procent van kopers AI gebruikte bij hun laatste aankoopbeslissing. Het gedrag komt niet aan; het is al verschoven. De conversiewiskunde versterkt de urgentie. Enrich Labs en AirOps rapporteren beide dat AI-bezoekers converteren aan ongeveer 14,2 procent versus 2,8 procent voor Google organic. Het mechanisme is simpel. Een koper die via ChatGPT binnenkomt heeft al een gesynthesiseerde aanbeveling gelezen die jouw merk samen met twee of drie alternatieven noemde. Ze landen al short-listed. Een koper via Google organic heeft op één van tien blauwe links geklikt en zit nog volop in evaluatie. De implicatie is dat je deze verschuiving kunt winnen met minder verkeer dan je nodig had om de SEO-verschuiving te winnen. De investering werkt op lagere volumes. De catch is dat het werk anders is. ## Wat LLMs écht wegen bij het citeren De Princeton GEO benchmark (Aggarwal et al., SIGKDD 2024) mat welke on-page elementen het sterkst correleren met citatie in ChatGPT, Perplexity en Bing Chat. De hoofdcijfers: | Element | Citatie lift | |---|---| | Inline cijfers met named bronnen | +41 procent | | Named-expert quotes | +28 tot +37 procent | | Externe autoritaire citaties | +30 tot +40 procent | | Zelfstandige antwoord-blokken van 130-160 woorden | Hoogste extractie-rate | | Vergelijkingstabellen vs equivalent proza | +25,7 procent (AirOps okt 2025) | | Keyword stuffing | -10 procent (penalty) | | Pagina's bijgewerkt binnen 30 dagen | 76,4x meer gerefereerd (Ahrefs) | Naast on-page signalen: Ahrefs' studie van 75.000 merken vond dat brand mentions correleren met LLM citatie aan r = 0,66 tot 0,74, terwijl backlinks correleren aan r = 0,22. Het brand-mention signaal verpulvert het backlink signaal. AirOps' studie van oktober 2025 vond dat pagina's geciteerd door ChatGPT 6,5x vaker uit third-party domeinen komen (podcasts, review sites, communities) dan uit eigen domeinen. Lees dat twee keer. De hoogst-leverage GEO investering in 2026 is **brand mentions opbouwen in third-party bronnen**. On-page werk telt, maar het is downstream van off-site mention engineering. ## De 12-punten GEO checklist die Falora gebruikt Pas elk item toe. Scoor 1 punt per "ja". 10 of meer brengt jouw pagina in het top-deciel voor citatie-eligibility. 1. **Zelfstandige TL;DR of antwoord-blok van 130 tot 160 woorden** bovenaan, geschreven als losstaand antwoord op de primaire vraag van de pagina. 2. **Minstens drie inline cijfers** met named bronnen uit de laatste 18 maanden. 3. **Minstens twee named-expert quotes** met attributie, ideaal met `Person` schema gelinkt naar LinkedIn of Wikidata. 4. **Eén vergelijkingstabel** met named alternatieven en named criteria. 5. **FAQ sectie** met vijf vragen beantwoord in 1 tot 3 zinnen elk, gemarkeerd met FAQPage schema. 6. **Externe autoritaire citaties** naar primaire bronnen (onderzoeksrapporten, regulator-publicaties, content van named experts). 7. **Last-updated datum zichtbaar** voor gebruiker én crawlers, met substantiële updates (niet enkel datum-bumps) elke 90 dagen voor evergreen pagina's. 8. **Eén uniek datapunt of framework** dat de pagina bezit, dat anderen zullen quoten bij dit onderwerp. 9. **Auteur-byline met credentials**, schema-gemarkeerd, gelinkt naar credible profiel met `sameAs`. 10. **Schone semantische HTML**: echte `

`, `

`, `

`, `
` tags. LLMs extraheren preferentieel uit semantische markup. 11. **Eén claim per zin**, citatable. Begraaf geen cijfer in een paragraaf; geef het een eigen zin zodat de LLM het schoon kan trekken. 12. **Minstens vijf brand-mention plaatsingen off-site** in hetzelfde kwartaal (podcasts, communities, third-party publicaties, review sites). De eerste 11 zijn on-page. De 12de is off-site. Die 12de is degene die het meest telt. ## Wat verandert er aan het schrijven zelf? Een pagina geoptimaliseerd voor SEO leest als een artikel. Een pagina geoptimaliseerd voor GEO leest als een referentiedocument met artikel-kwaliteit proza eromheen. Het structurele verschil is dat elke sectie **op zichzelf moet kunnen worden geëxtraheerd**. De Wellows extractie-studie van december 2025 analyseerde 12.000 ChatGPT antwoorden over 200 B2B queries en vond dat de gemiddelde geëxtraheerde unit 137 woorden lang was, minstens één named cijfer bevatte, en in de bovenste 60 procent van de bronpagina lag. Dat is de grootte en vorm waarvoor je schrijft. Drie schrijfregels volgen hieruit: **Front-load het antwoord.** Zet de pointe in de eerste 150 woorden. De koper die jouw pagina leest kwam mogelijk van een chat die je al parafraseerde; ze vertrekken snel als de pagina het chat-antwoord niet bevestigt. **Eén claim per zin.** "Belgische B2B-kopers vermijden 73 procent vaker irrelevante outreach in 2025" is een citatable zin. Dezelfde feit in een paragraaf van 60 woorden is dat niet. **Gebruik named entities aggressief.** "Chris Walker stelt dat 97 procent van net new ARR herleidbaar is naar dark social" is extracteerbaar. "Sommige stemmen stellen dat de meeste omzet verborgen blijft" is dat niet. De LLM heeft named subjects nodig om de claim aan vast te haken. ## Het off-site werk dat niemand graag doet De on-page checklist is doable in één kwartaal gefocust werk. De off-site brand-mention engineering is doable in twee tot vier kwartalen consistente investering en is wat de merken die in ChatGPT-antwoorden verschijnen scheidt van degenen die dat niet doen. Vier off-site vectoren compounderen: **Podcast appearances.** YouTube transcripts en podcast transcripts worden aggressief gevoed door elke grote LLM. Eén podcast-appearance van 45 minuten met jouw founder als guest produceert 20 tot 60 indexable mentions van jouw merk. De single-factor correlatie tussen podcast guest appearances en LLM citatie is r ~= 0,74 in de Ahrefs studie, hoger dan elk ander gemeten signaal. **Community aanwezigheid.** Mentions in Pavilion, Exit Five, RevGenius, Modern Sales Pros, GTM Partners en NL/BE-equivalenten (Bel.AI, AI4Belgium, Beltug, SaaS Founders Benelux) worden geïndexeerd door Perplexity en Claude. Consistente commenters bouwen compounding brand-mention dekking. **Review-site aanwezigheid.** G2, Capterra, TrustRadius. Niet alleen voor kopers, maar omdat LLMs deze pagina's lezen en categorie-aanwezigheid als proxy voor legitimiteit gebruiken. Een merk dat niet in drie of meer review-categorieën verschijnt, is functioneel onzichtbaar voor vergelijkende koper-queries. **Third-party publicaties.** Demand Gen Report, Sales Hacker, Growth Unhinged voor EN; Marketingfacts, Frankwatching, Bloovi, Sifted voor NL/EU. Eén bylined artikel of named quote in één stuk presteert beter dan 50 backlinks op eigen domeinen. Dit is exact wat Falora investeert voor het eigen merk en wat we portfolio-bedrijven helpen op te bouwen. Het werk is onsexy en de payback is zes tot twaalf maanden. Er is geen shortcut. ## Hoe meet je GEO? Het meetprobleem is reëel. ChatGPT geeft referrer headers niet betrouwbaar door. Claude en Perplexity variëren. De meeste analytics tools attribueren AI-engine verkeer als "direct" of "unknown", wat het kanaal verkeerd weergeeft. Drie meetbenaderingen samen vormen een werkende stack: **Directe citatie-monitoring.** Tools zoals Profound, Otterly.ai en Peec AI scrapen ChatGPT, Perplexity, Claude en Gemini antwoorden op jouw brand mentions over een gedefinieerde query set. Ze rapporteren citation rate, share of voice en concurrent benchmarks. Dichtst bij een SEO-rank-tracker equivalent voor GEO. **Self-reported attributie.** Een verplicht free-text "Hoe heb je over ons gehoord?" veld op high-intent formulieren. Chris Walker's onderzoek over 40+ SaaS bedrijven toont dat wat kopers melden systematisch afwijkt van wat attributie-software meet. We werken de volledige methodologie uit in [ons self-reported attribution artikel](https://falora.ai/blog/self-reported-attribution-b2b). **Traffic anomaly detection.** AI-driven sessions zien er anders uit in GA4: kortere referral chain, hogere engagement, snellere kwalificatie op het formulier. Tag de patronen en behandel de lift als GEO-toewijsbaar zelfs als het kanaal "direct" zegt. Geen enkele meting is compleet. De triangulatie is wat werkt. ## Een 90-dagen GEO sprint Voor een Belgisch of Nederlands B2B SaaS team dat vanaf nul start, dit is het 90-dagen plan dat we in de Stretch portfolio draaien. **Dag 1 tot 10: audit.** Run jouw top 25 pagina's tegen de 12-punten checklist. Scoor elk. Identificeer de vijf pagina's met de hoogste potentiële lift (hoog verkeer + lage GEO score). **Dag 11 tot 30: herschrijf de top vijf.** Voeg TL;DR, cijfers, named quotes, FAQ, vergelijkingstabel waar relevant, verse updated datum toe. Voeg `Person`, `FAQPage`, `Article` en `Organization` schema toe. **Dag 31 tot 60: pitch en plaats.** Identificeer 10 podcasts, 5 communities en 3 third-party publicaties relevant voor jouw categorie. Pitch founder of operator appearances. Mik op 5 plaatsingen in het 30-dagen venster. **Dag 61 tot 90: meet en itereer.** Zet een citatie-monitor op (Profound of equivalent). Baseline jouw brand citation rate op 50 high-intent queries. Voeg self-reported attribution toe aan jouw demo-formulier. Identificeer welke off-site plaatsingen correleren met de grootste citatie-lift. Tegen dag 90 heb je een on-page baseline, een off-site mention engine, en een meet-loop. Het compounding start dan. ## Veelgestelde vragen **Is GEO gewoon een rebrand van SEO?** Nee. De technische fundamenten overlappen (clean HTML, schema, semantische markup) maar de signaal-weging verschilt. Brand mentions domineren over backlinks in GEO; cijfers en named quotes domineren over keywords. De on-page schrijfstijl verschilt. **Moet ik een GEO bureau inhuren?** De meeste GEO bureaus in 2026 zijn SEO bureaus met een nieuwe landing page. Huur enkel in als ze een citation-rate dashboard kunnen tonen voor een bestaande klant, gevoed door een echte monitor (Profound, Otterly.ai, Peec AI). Kunnen ze dat niet, dan gokken ze. **Werkt dit voor Nederlandstalige content?** Ja, met dezelfde fundamenten. ChatGPT, Claude en Perplexity verkiezen native-taal bronnen voor native-taal queries (Profound's regional source-bias studie). Nederlandstalige GEO voor Belgische en Nederlandse B2B SaaS staat wijd open. Zie ook [de EN companion piece](https://falora.ai/blog/geo-for-b2b-saas). **Wat is het GEO-equivalent van een backlink?** Een brand mention in een third-party bron die de LLM voedt. Podcast guest appearances zijn de hoogste-correlatie single bron. Community aanwezigheid en third-party bylines zijn nauw aansluitende seconden. **Vervangen Google AI Overviews ChatGPT voor B2B research?** Onwaarschijnlijk op korte termijn. De G2 2026 data toont dat kopers actief pure generative engines verkiezen voor vendor-research vragen. Google AI Overviews zijn sterker voor informationele queries. Beide tellen; de investering moet geen either-or zijn. ## Conclusie De koper zit in ChatGPT, Claude en Perplexity vóór ze in jouw CRM zitten. Het team dat de komende 24 maanden van B2B SaaS discovery wint, is het team dat door die engines geciteerd wordt, niet het team dat ranked voor blauwe links. Het on-page werk zit binnen jouw controle vandaag. Het off-site brand-mention werk is wat compoundert. Wil je de 12-punten checklist op jouw top-pagina's toepassen met een operator die het 30+ keer deed in de Stretch portfolio? **[Plan een 30-minuten GEO diagnostic in](https://falora.ai/demo)**. --- ### Bronnen - [G2, *2026 Software Buyer Behavior Report*](https://www.g2.com/research) - [6sense, *European Buyer Experience Report 2024*](https://6sense.com/resources) - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) - [Princeton GEO Benchmark, Aggarwal et al. (SIGKDD 2024)](https://arxiv.org/abs/2311.09735) - [Ahrefs, *75K-Brand LLM Citation Study*](https://ahrefs.com/blog) - [AirOps, *GEO citation study oktober 2025*](https://www.airops.com) - [Wellows, *LLM extraction study december 2025*](https://www.wellows.com) - [Mersel AI, *GEO for B2B 2026 Guide*](https://www.mersel.ai/generative-engine-optimization) - [Jasper, *GEO vs AEO vs SEO 2026*](https://www.jasper.ai/blog/geo-aeo) ### Verder lezen op Falora - [GEO for B2B SaaS (English version)](https://falora.ai/blog/geo-for-b2b-saas) - [GTM engineering uitgelegd](https://falora.ai/nl/blog/gtm-engineering-uitgelegd) - [Schaalbare B2B groei zonder bureau](https://falora.ai/nl/blog/schaalbare-b2b-groei-zonder-bureau) - [De anatomie van een GTM engineering systeem](https://falora.ai/blog/gtm-engineering-system-anatomy) ### Over de auteur Stijn Van Daele is co-founder van Falora en partner bij Stretch Innovation. Hij paste de GEO checklist hierboven toe op 30+ B2B SaaS merken in België, Nederland en de bredere EU. Hij schrijft over GTM engineering, autonome revenue en de EU AI Act op [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **Wat is Generative Engine Optimization (GEO)?** GEO is de praktijk van het structureren van content en technische infrastructuur zodat AI-engines zoals ChatGPT, Gemini, Perplexity en Claude jouw merk citeren in hun gegenereerde antwoorden. Waar SEO optimaliseert voor positie in een lijst van blauwe links, optimaliseert GEO voor opname in één gesynthesiseerd antwoord. **Wat is het verschil tussen GEO, SEO en AEO?** SEO mikt op ranking in klassieke search engine result pages. AEO (Answer Engine Optimization) mikt op AI-geboorste features zoals Google AI Overviews en Bing Chat. GEO mikt op pure generative engines zoals ChatGPT en Claude die helemaal geen SERP tonen. De drie overlappen op technische fundamenten maar verschillen sterk in signaal-weging. **Hoe word ik geciteerd door ChatGPT?** Vijf dingen maken het verschil. Een, publiceer originele data, surveys en benchmarks met genoemde cijfers. Twee, laat je quoten door named experts. Drie, structureer content in zelfstandige antwoord-blokken van 130 tot 160 woorden. Vier, verdien brand mentions in third-party bronnen (podcasts, communities, review sites). Vijf, hou content vers; ChatGPT citeert pagina's bijgewerkt binnen 30 dagen 76,4 keer vaker dan oudere content. **Vervangt GEO klassieke SEO voor Belgische B2B SaaS?** Nog niet, maar het zwaartepunt is verschoven. Klassieke SEO drijft nog steeds een betekenisvol deel van high-intent verkeer. Maar voor early research, vergelijkingen en short-list vragen hebben AI-engines Google vervangen als eerste stop voor 51 procent van software-kopers (G2, 2026). Een B2B SaaS team dat enkel in SEO investeert, concurreert nu voor de helft van de discovery markt. **Wat is de ROI van GEO voor B2B SaaS?** AI-bezoekers converteren aan 14,2 procent gemiddeld versus 2,8 procent voor Google organic, omdat de koper pre-geïnformeerd en recommendation-primed aankomt. De investering verdient zich terug bij lager verkeer dan klassieke SEO. De meeste B2B SaaS teams in onze portfolio zien GEO-toewijsbare pipeline binnen 90 dagen na implementatie. --- # Hoe Belgische kopers écht zoeken in 2026: ChatGPT, Perplexity en AI Overviews - URL: https://falora.ai/nl/blog/belgische-kopers-ai-search-2026 - Language: nl - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-05-08 - Category: Insights - Tags: geo, ai-search, seo, b2b-demand-gen, benelux - Reading time: 9 min - Translation key: geo-ai-search-benelux ## TL;DR In 2026 gebruikt 94% van de B2B buying groups generatieve AI in hun research. AI Overviews verschijnen op 48% van alle Google-zoekopdrachten en verlagen de organische CTR met 61%. Voor B2B-bedrijven betekent dit: als je merk niet wordt geciteerd door ChatGPT, Perplexity of Google's AI-modellen, besta je niet in een groot deel van het koopproces. SEO blijft belangrijk maar volstaat niet meer; de aanvulling heet Generative Engine Optimization (GEO). ## Description 94% van B2B-kopers gebruikt generatieve AI in 2026. Wat ChatGPT, Perplexity en Google AI Overviews betekenen voor je B2B demand gen in BE en NL. ## Full content ## TL;DR > Je B2B-koper zit niet meer in Google. Hij vraagt het ChatGPT, scrolt door een AI Overview en sluit het tabblad zonder ooit op jouw site te klikken. - 94% van de B2B buying groups gebruikt generatieve AI in research (Forrester, 2026). - AI Overviews verschijnen op 48% van alle Google-zoekopdrachten; organische CTR daalt met 61%. - B2B-kopers tonen 90% click-through op bronnen die in AI Overviews worden geciteerd. - Generative Engine Optimization (GEO) is de aanvulling op SEO; de Belgische markt is er bijna niet op voorbereid. ## Inleiding Je B2B-koper zit niet meer in Google. Of beter: hij zit er nog wel, maar hij klikt niet meer. Hij vraagt het ChatGPT. Of Perplexity. Of hij scrollt door een AI Overview die Gemini bovenaan zijn zoekresultaten plakt en sluit het tabblad zonder ooit op jouw site te klikken. En tegen de tijd dat hij wél contact opneemt, heeft hij al een shortlist gemaakt waar jij niet op staat. Dit is geen toekomstprobleem. Dit gebeurt nu. In Brussel, in Antwerpen, in Eindhoven, in Gent. Bij CMO's en Heads of Growth die nog steeds vergaderen over hun "SEO-strategie" alsof het 2022 is. Tijd om eerlijk te zijn over wat er aan het verschuiven is — en wat je eraan kunt doen. ## Het korte antwoord In 2026 gebruiken **94% van de B2B buying groups generatieve AI** in hun researchfase (Forrester, 2026). Het traditionele zoekverkeer daalt naar verwachting met **25% tegen eind 2026** door de opkomst van AI-chatbots (Gartner). En **48% van alle Google-zoekopdrachten** triggert nu een AI Overview, waardoor de organische CTR met 61% kelderde. Voor B2B-bedrijven betekent dit: als je merk niet wordt geciteerd door ChatGPT, Perplexity of Google's AI-modellen, besta je niet in een groot deel van het koopproces. SEO blijft belangrijk, maar is niet meer voldoende. De aanvulling heet Generative Engine Optimization (GEO), en de Belgische markt is er bijna niet op voorbereid. Laten we de cijfers stuk voor stuk bekijken — en wat ze concreet betekenen voor je demand gen in BE en NL. ## Hoe Belgische en Nederlandse kopers écht zoeken (de cijfers van 2026) **1. AI-tools zijn al de standaard, niet de uitzondering.** Een analyse uit maart 2026 van 680 miljoen citaties door onderzoeksbureau Averi toont dat **73% van de B2B-kopers nu AI-tools zoals ChatGPT en Perplexity gebruikt** in hun researchproces. Forrester's B2B Predictions 2026 zet dat percentage zelfs op 94%. Beide cijfers zeggen hetzelfde: AI-research is niet meer marginaal, het is mainstream. **2. ChatGPT alleen domineert de markt.** Met **800 miljoen wekelijkse actieve gebruikers** (februari 2026, verdubbeld in vier maanden tijd), verwerkt ChatGPT meer dan 2 miljard zoekopdrachten per dag. Voor de context: dat is bijna evenveel als Bing. **3. AI Overviews zijn nu de norm in Google.** Volgens Stackmatix verschijnen AI Overviews intussen op **48% van alle zoekopdrachten** (maart 2026), tegenover 13,1% een jaar eerder. Dat betekent dat in bijna de helft van alle zoekopdrachten Google zelf het antwoord genereert — en dat de gebruiker nooit op je website hoeft te klikken. De organische CTR daalt met 61% wanneer er een AI Overview verschijnt. **4. Voor B2B is de impact nog groter.** Volgens Omniscient Digital tonen B2B-kopers **90% click-through-rates op de bronnen die in AI Overviews worden geciteerd**, vergeleken met 8% voor algemene gebruikers. Vertaald: als jij geciteerd wordt, krijg je clicks van mensen die echt iets van plan zijn. Als je niet geciteerd wordt, sta je niet in het gesprek. **5. AI-verkeer converteert beter.** Onderzoek van SE Ranking en Exposure Ninja toont dat AI-bezoekers **68% meer tijd op websites doorbrengen** dan traditionele organische bezoekers, en 4,4 keer hoger converteren. Een ChatGPT-bezoeker is een serieuze koper. Een Perplexity-bezoeker is een nóg serieuzere koper. **6. En in België specifiek?** Volgens PwC Belgium (mei 2025) gebruikt **34% van de Belgische werknemers die dagelijks met een computer werken** regelmatig generatieve AI-tools — een sprong van 13% in 2024 naar 34% in 2025. **25% van de Belgische bedrijven met meer dan 10 werknemers** gebruikt minstens één AI-applicatie, een stijging van 80% op één jaar tijd (Acerta, februari 2025). België staat daarmee 4e in Europa qua AI-gebruik in bedrijven, achter Denemarken (42%), Finland (38%) en Zweden (35%). In Nederland zit het al stukken hoger: volgens CBS/Eurostat gebruikt **64% van de Nederlandse marketeers AI**, een stijging van 56% in twee jaar tijd. 46% van de Nederlandse B2B-marketeers zet AI in voor leadgeneratie en prospecting (Spotler). De conclusie: jouw koper zit op deze tools. Vandaag. En als je marketing nog steeds vooral mikt op de blauwe-link-Google van vroeger, ben je structureel onder-geïnvesteerd in waar het gesprek nu écht plaatsvindt. ## Wat dit betekent voor je demand gen (zonder rooskleurige bril) De praktische gevolgen voor B2B-marketeers in BE en NL: **De dark funnel wordt nóg donkerder.** Volgens 6sense gebeurt al 70% van de B2B-koopreis anoniem, voordat een verkoper ooit contact heeft. Met AI-search erbij wordt die fase niet kleiner — ze wordt eerder onzichtbaarder. Buyers doen nu hun research in een interface die geen UTM-parameters genereert, geen referrer-data doorgeeft en geen vorm van klassieke webstatistieken oplevert. Je weet niet eens dat ze er waren. **Branded search wordt het belangrijkste leading indicator.** Als ChatGPT je noemt, gebeurt er meestal hetzelfde: de prospect typt vervolgens je merknaam in Google om je terug te vinden. Branded search is dus de schaduw die AI-citatie achterlaat. Als je merknaam-zoekopdrachten stijgen zonder dat je advertenties draait, ben je waarschijnlijk geciteerd. Als ze dalen, ben je aan het verdwijnen. **Attribution wordt onmogelijk op de oude manier.** Dit is wat Frankwatching's Linda Krijns goed verwoordde: *"In plaats van volledige controle verschuift de focus naar richting en samenhang. Trends herkennen, experimenten vergelijken en data combineren uit meerdere bronnen wordt belangrijker dan het najagen van perfecte dashboards."* Met andere woorden: stop met proberen elke euro toe te wijzen aan één kanaal. Dat lukt toch niet meer. **Belgische sales cycles zijn lang. Dat maakt dit erger, niet beter.** Volgens 6sense's Europese onderzoek duurt een Belgische B2B-sales cycle gemiddeld **11,5 maanden** met een DMU van **11,3 stakeholders**. Dat zijn elf mensen die elf maanden lang research doen. In dat hele traject moet jouw merk minstens één keer op iemand zijn beeldscherm verschijnen — en als dat scherm in 2026 vooral ChatGPT toont, dan is dáár waar de strijd plaatsvindt. ## Wat is GEO en hoe verhoudt het zich tot SEO? Generative Engine Optimization (GEO) is de strategie om je content vindbaar te maken in AI-gedreven zoekresultaten — ChatGPT, Perplexity, Claude, Google's AI Overviews, Gemini, Bing Copilot. GEO is geen vervanging van SEO. Het is een extra laag erbovenop. De technieken overlappen voor ongeveer 60%, maar verschillen op kritieke punten: | Aspect | Klassieke SEO | GEO | |---|---|---| | **Optimalisatie** | Zoekwoorden, backlinks, technische pagina-snelheid | Semantische helderheid, citeerbare statistieken, expert-quotes | | **Format** | Lange pagina's met keyword density | Zelfstandige antwoord-blokken van 130–160 woorden | | **Autoriteit** | Domeinautoriteit via backlinks | Brand mentions (gelinkt én ongelinkt) op autoritatieve sites | | **Levenscyclus** | Lange houdbaarheid | Vereist regelmatige updates (`dateModified`) | | **Meting** | Rankings, organisch verkeer, CTR | Citatie-frequentie in AI-antwoorden, share of voice in LLM-output | Het Princeton-onderzoek naar GEO door Aggarwal et al. (2024) toonde aan dat content met statistieken **41% meer LLM-citaties** krijgt, content met expert-quotes 28–37% meer, en content met externe citaties 30–40% meer. Maar — belangrijk — keyword stuffing verlaagt de zichtbaarheid juist met 10%. De oude SEO-trucs werken averechts in AI. ## Welke types content worden geciteerd door AI? Op basis van Ahrefs' analyse van 75.000 merken en de Princeton GEO-studie weten we ongeveer wat werkt: **1. Vergelijkingstabellen.** AirOps-onderzoek (2025) toonde dat content met vergelijkingstabellen **25,7% meer ChatGPT-citaties** krijgt dan vergelijkbare content zonder. AI's vinden tabellen makkelijk te extraheren en betrouwbaar. **2. FAQ-secties met FAQPage-schema.** Vragen + antwoorden zijn de meest extracteerbare content-vorm voor LLMs. Een FAQ-sectie met 5–8 echte vragen van je doelgroep is letterlijk catnip voor ChatGPT. **3. Self-contained antwoord-blokken (130–160 woorden).** Wellows-onderzoek (december 2025) toonde aan dat dit de optimale lengte is voor extractie. Schrijf elke H2-sectie zo dat ze los geknipt kan worden en nog steeds zinvol is. **4. Content met genoemde experts.** Quote Chris Walker, Kyle Poyar, Jacco van der Kooij, Pieterjan Bouten — bij naam, met source. LLMs leren patronen uit citaten en gebruiken ze om autoriteit te valideren. **5. Originele data en cijfers.** Eigen surveys, klant-benchmarks, eigen rapporten. Een eigen statistiek met je merknaam erbij wordt overal opnieuw geciteerd. Dat is gratis brand-mention engineering. **6. Recente content.** Volgens Ahrefs citeren ChatGPT en Perplexity content die in de laatste 30 dagen is bijgewerkt **76,4 keer vaker** dan oudere content. Refresh werkt. **7. Long-form pillar content (2.000+ woorden).** Niet omdat lengte op zich beter is, maar omdat lange artikels meer "fan-out queries" dekken — de subvragen die AI's automatisch genereren bovenop een hoofdvraag. ## Wat klassieke SEO mist (en GEO toevoegt) Het verschil zit niet in de techniek. Het zit in wát je optimaliseert. Klassieke SEO denkt in termen van zoekopdrachten ("CRM kopen voor scale-up"). GEO denkt in termen van vragen ("Wat is het beste CRM voor een Belgische scale-up van 50 mensen die in vier landen verkopen?"). Dat tweede is geen zoekopdracht meer. Dat is een gesprek. En de winnaars in 2026 zijn de bedrijven die hun content schrijven alsof ze dat gesprek voeren — niet alsof ze een SEO-spec invullen. Een goed voorbeeld: in plaats van een blogpost getiteld *"5 tips voor B2B leadgeneratie"* (klassiek SEO-format), schrijf je *"Waarom de meeste B2B leadgeneratie-tactieken in 2026 niet meer werken (en wat wel)"*. De eerste rankt misschien nog op Google. De tweede wordt geciteerd door ChatGPT als iemand vraagt: "Welke leadgeneratie-tactieken werken in 2026?" Het verschil in CTR? Significant. Het verschil in autoriteit? Enorm. ## Het EU AI Act-stuk dat niemand je vertelt Vanaf augustus 2026 verplicht **Article 50 van de EU AI Act** dat AI-gegenereerde content als zodanig wordt aangeduid. Dit treft elke marketingorganisatie die AI inzet voor copy, ads, of klantcommunicatie. Voor België specifiek: de Gegevensbeschermingsautoriteit (GBA/APD) publiceerde op 19 september 2024 specifieke richtlijnen over "AI-systemen en de AVG", en BIPT is aangewezen als markttoezichtsautoriteit voor de AI Act. Bij overtredingen kunnen de boetes oplopen tot **€35M of 7% van de wereldwijde omzet**, afhankelijk van wat hoger is. Wat dit betekent voor je content-strategie: AI-gegenereerde blogposts zonder menselijke editing en zonder duidelijke disclosure worden niet alleen risicovol vanuit AI-Act-perspectief, ze worden ook gemarkeerd door LLMs als minder betrouwbaar — wat je GEO-zichtbaarheid omlaag duwt. Het is geen toeval dat de meest succesvolle content in AI-search ook de meest menselijke is. Auteurs met naam, bio, ervaring, en LinkedIn-profiel. Die regulatorische druk en GEO-best-practice komen op hetzelfde punt uit: stop met dumpen van bulk-AI-content. Frankwatching's Sander Visser verwoordde het scherp: *"Middelmatige AI-content sloopt je merk."* Hij heeft gelijk. Vooral nu. ## Een zelfdiagnose: wordt jouw B2B-content gevonden door AI? Test je eigen merk in 5 minuten. Open ChatGPT (zonder ingelogd te zijn — je wilt niet dat het je geschiedenis gebruikt). Stel deze vijf vragen, vervang waar nodig: 1. *"Wat zijn de beste [jouw productcategorie]-aanbieders in België?"* 2. *"Welke [jouw productcategorie]-tools werken goed voor [jouw ICP]?"* 3. *"Hoe vergelijk ik [jouw productcategorie] in Nederland?"* 4. *"Wie zijn de leiders in [jouw niche]?"* 5. *"Hoe kies ik een [jouw productcategorie] voor een groeiend bedrijf?"* Drie scenario's: - **Je merk wordt genoemd, met de juiste positionering.** Gefeliciteerd. Doe meer van wat je doet. - **Je merk wordt genoemd, maar verkeerd of generiek.** Je hebt een mention engineering-probleem. De content waar je in geciteerd wordt geeft een verkeerd beeld. - **Je merk wordt niet genoemd.** Welkom bij de meerderheid van de Belgische B2B-markt. Je hebt werk te doen. Doe dezelfde test op Perplexity (toont sources expliciet — handiger om te zien waar je merk vandaan zou kunnen komen) en op Google AI Overviews (typ je vraag in google.com en kijk wat bovenaan staat). ## Een 60-dagen GEO-actieplan voor BE/NL B2B Concreet, geen filosofie. Wat je morgen kunt doen. **Week 1–2: Audit.** Doe de zelfdiagnose hierboven. Documenteer voor welke vragen je wel/niet geciteerd wordt. Identificeer welke concurrenten wél geciteerd worden — en lees de pagina's die ChatGPT/Perplexity citeert om hun bron-pagina's te vinden. Dat zijn je benchmarks. **Week 2–4: Herwerk je 10 belangrijkste pagina's.** Per pagina: - Voeg een **TL;DR-blok van 130–160 woorden** toe bovenaan dat de hele pagina samenvat - Voeg minstens **3 statistieken met bronvermelding** toe (datum + URL) - Voeg minstens **2 citaten van bekende experts** toe (Chris Walker, Kyle Poyar, Jacco van der Kooij, etc.) - Voeg een **FAQ-sectie van 5–8 vragen** met FAQPage-schema toe - Update `dateModified` naar de huidige datum **Week 4–6: Bouw 1 GEO-bron pagina per maand.** Een GEO-bron pagina is een long-form (2.000+ woorden) artikel met origineel framework, eigen data, of unieke perspectief. Geen "10 tips voor..." Wel "Waarom [contrarian thesis] (en wat je eraan kunt doen)." Pillar content die andere mensen citeren in hun eigen content. Dat is je echte GEO-flywheel. **Week 6–8: Brand mentions buiten je eigen domein.** Schrijf gastpost op Frankwatching, Bloovi, Marketingfacts, Emerce. Plaats een interview op een Belgische B2B-podcast (Fast Forward, B2B Sound). Dien je merk in op review-sites (G2, Capterra). LLMs leren je merk vooral kennen via *third-party mentions*. Volgens Ahrefs correleren brand mentions 3 keer sterker met AI-citatie dan backlinks. **Doorlopend: Meet wat je kunt meten.** Tools zoals Profound, Otterly.ai, Peec AI, of Ziptie monitoren LLM-citaties op vooraf gedefinieerde queries. De handmatige methode: stel een lijst op van 30–50 vragen relevant voor jouw markt, test ze maandelijks op ChatGPT, Claude, Perplexity en Gemini, en track of je merk en concurrenten voorkomen. ## Wat dit betekent als je vandaag een Belgische scale-up CMO bent De vraag die elke CMO zich moet stellen voor 2026: ben je vindbaar door AI als de beslisser je nodig heeft? Voor de meeste Belgische en Nederlandse B2B-bedrijven is het antwoord vandaag: nee, of inconsistent. Dat is geen schande — het hele B2B-landschap zit in dezelfde positie. Maar het is wel een kans, want het venster om dit te repareren voordat je concurrenten het doen, is nu open. Wie nu investeert in GEO — origineel onderzoek, expert-citaten, structureel hergebruikte frameworks, regelmatige content-refresh, brand mentions buiten het eigen domein — bouwt een onmogelijk-te-kopiëren vorm van zichtbaarheid op. Wie blijft denken in termen van "we doen wat aan SEO" zal binnen 12 tot 18 maanden zien dat het organische verkeer dat ooit hun pijplijn voedde, gewoon verdwenen is. Het is geen voorspelling. Het gebeurt al. ## Conclusie Falora.ai is een Belgisch autonomous GTM-platform, gebouwd door het team van Stretch Innovation. We helpen B2B scale-ups in BE en NL hun GTM-engineering bouwen voor het AI-tijdperk — van demand creation tot signaal-gedreven capture. **[Plan een 30-minuten GTM-diagnose met Stijn →](https://meetings-eu1.hubspot.com/stijn-van-daele/falora)** --- ### Bronnen - [Forrester — *B2B Predictions 2026*](https://www.forrester.com) - [Gartner — *Search Engine Volume Predictions*](https://www.gartner.com) - [Stackmatix — *AI Overviews Prevalence Report* (2026)](https://stackmatix.com) - [Omniscient Digital — *AI Overview B2B CTR Study* (2026)](https://beomniscient.com) - [6sense — *2025 Buyer Experience Report*](https://6sense.com/blog/new-buyer-experience-report-3-insights-and-a-warning/) - [PwC Belgium — *AI in the Workplace* (mei 2025)](https://www.pwc.be) - [Acerta — *AI-adoptie Belgische bedrijven* (februari 2025)](https://www.acerta.be) - [Aggarwal et al. — *Generative Engine Optimization* (Princeton, SIGKDD 2024)](https://arxiv.org/abs/2311.09735) - [Ahrefs — *AI Overview Brand Visibility Factors*](https://ahrefs.com/blog/ai-overviews-study/) - [Frankwatching — *GEO en zero-click search*](https://www.frankwatching.com) - [EU AI Act, Regulation EU 2024/1689](https://eur-lex.europa.eu/eli/reg/2024/1689/oj) - [Belgische Gegevensbeschermingsautoriteit — *AI-systemen en de AVG*](https://www.gegevensbeschermingsautoriteit.be) ### Verder lezen op Falora - [GTM engineering uitgelegd](https://falora.ai/nl/blog/gtm-engineering-uitgelegd) - [Schaalbare B2B-groei zonder bureau](https://falora.ai/nl/blog/schaalbare-b2b-groei-zonder-bureau) ### Over de auteur Stijn Van Daele is co-founder van Falora en partner bij Stretch Innovation. De voorbije 6 jaar hielpen hij en zijn team meer dan 200 bedrijven met het bouwen van een groei-engine die hun bedrijf écht deed schalen. Hij schrijft over GTM engineering, autonomous revenue en de EU AI Act op [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **Wat is Generative Engine Optimization (GEO)?** GEO is de strategie om je content vindbaar en citeerbaar te maken in AI-gedreven zoekresultaten zoals ChatGPT, Perplexity, Claude, Google AI Overviews, Gemini en Bing Copilot. Het overlapt voor ongeveer 60% met klassieke SEO, maar verschilt op kritieke punten: semantische helderheid, citeerbare statistieken, expert-quotes en zelfstandige antwoord-blokken van 130–160 woorden. **Vervangt GEO de klassieke SEO?** Nee. GEO is een extra laag bovenop SEO, geen vervanging. De technieken overlappen, maar het format verschilt: SEO mikt op zoekwoorden en backlinks, GEO mikt op brand mentions, citeerbare data en structuur die LLM's makkelijk kunnen extraheren. Klassieke SEO blijft nodig zolang Google nog organische klikken stuurt. **Hoe meet ik of mijn merk geciteerd wordt door AI?** Tools zoals Profound, Otterly.ai, Peec AI en Ziptie monitoren LLM-citaties op vooraf gedefinieerde queries. De handmatige methode: stel een lijst op van 30–50 vragen relevant voor jouw markt, test ze maandelijks op ChatGPT, Claude, Perplexity en Gemini, en track of je merk en concurrenten voorkomen. **Welke content krijgt het meeste citaties van ChatGPT en Perplexity?** Vergelijkingstabellen (+25,7% citaties), FAQ-secties met FAQPage-schema, antwoord-blokken van 130–160 woorden, content met expert-quotes (+28–37%), originele data en cijfers (+41%) en content die in de laatste 30 dagen is bijgewerkt (+76,4× meer geciteerd dan oudere content). Keyword stuffing verlaagt de zichtbaarheid juist met 10%. **Wat betekent de EU AI Act voor mijn content-strategie?** Vanaf augustus 2026 verplicht Article 50 dat AI-gegenereerde content als zodanig wordt aangeduid. Boetes kunnen oplopen tot €35M of 7% van de wereldwijde omzet. Praktisch: AI-bulkcontent zonder menselijke editing wordt zowel regulatoir risicovol als slechter geciteerd door LLM's. De winnende formule is menselijke auteurs met naam, bio en ervaring. --- # Signal-based demand capture: 7 buying signals that make your outbound 4× more effective - URL: https://falora.ai/blog/signal-based-demand-capture - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-05-05 - Category: Insights - Tags: signal-based-selling, outbound, demand-capture, b2b-sales, intent-data - Reading time: 10 min - Translation key: signal-based-demand-capture ## TL;DR A buying signal is any observable event that suggests an account or contact has just entered a buying window. Signal-based selling triggers outreach within hours of a signal — instead of blasting the TAM on a calendar cadence — and consistently delivers 25–40% reply rates versus 2–5% for cold volume. The 7 highest-leverage signals: past champion job changes, new executives, funding/M&A, hiring surges, tech stack changes, de-anonymised website intent, and community engagement. ## Description Signal-based outbound delivers 4× more deals, 80% higher reply rates and 47% better conversion. The 7 highest-leverage signals, ranked, with a 30-day plan. ## Full content ## TL;DR > Outbound at scale is broken. The fix is not "more sequences." It is changing what triggers outreach in the first place. - Volume cold outbound averages 2–5% reply rates; signal-based outbound delivers 25–40%. - Past customer/champion job changes are the single highest-converting signal in B2B. - The first vendor to reach a buyer after a trigger event wins ~5× more often. - Signal-based outbound is also the most GDPR-defensible form of cold outreach in Europe. ## Introduction Outbound at scale is broken. You already know that. Your reply rates are below 1%. Your SDRs are burnt out. Your CFO is asking why the cost-per-meeting keeps creeping up. And every prospect you finally get on the phone has either already chosen a vendor or has no idea who you are. There's a way out of this, but it's not "more sequences" or "better copy." It is a fundamental shift in *what* triggers outreach in the first place. It's called signal-based selling. And the data is unambiguous: teams that do it well see **4× more deals booked, 80% higher reply rates, and 47% better conversion rates** versus traditional volume-based outbound. Below: the 7 buying signals that actually move the needle, why each one works, what tool stack to use, and a 30-day plan to put them into production. No fluff. ## The 30-second answer A buying signal is any observable event that suggests a specific account or contact has just entered (or is about to enter) a buying window. Signal-based selling triggers outreach the moment a signal fires — within hours, not days — instead of blasting your full TAM on a calendar cadence. The 7 highest-leverage signals for B2B in 2026 are: (1) job changes of past customers/champions into ICP roles, (2) new executives in buying roles, (3) funding rounds and M&A, (4) hiring surges, (5) tech stack changes and competitor churn, (6) de-anonymized website intent, and (7) community/dark social engagement. Stack 2–3 signals together and reply rates jump from 2–5% (cold) to 15–40%. The catch: most teams know about these signals; almost none have automated the detection-to-outreach loop. ## Why signal-based outbound now (and not earlier) Three things changed at the same time, and together they killed volume-based outbound. **First, inbox enforcement.** Google and Microsoft tightened bulk-sender rules in early 2024. Sending 10,000 cold emails a day used to be tolerated. In 2025–2026, it gets you in spam folders and graduated deliverability penalties. The volume play is now self-defeating. **Second, buyer fatigue.** Gartner's 2025 sales survey of 632 B2B buyers found **73% actively avoid suppliers who send irrelevant outreach.** Three quarters of your TAM will close the door on you for sending a generic email at the wrong time. Volume + irrelevance is now a brand-damage strategy, not a pipeline strategy. **Third, the maths got worse.** Belkins' 2025 study of 16.5 million cold emails found average reply rates around 5.1%, with larger untargeted campaigns dropping to 2.1%. Volume-based outbound is yielding less than half what it did three years ago. Meanwhile, signal-based outreach consistently delivers 5× that rate. The shift isn't optional. It's already happened. The question is whether your team is on the right side of it. ## What "signal" actually means (and what it doesn't) Quick definition because vendors muddy this up: A **buying signal** is an observable, real-world event that indicates a prospect has new context, new budget, new pain, or new authority — making them more likely to be receptive to your outreach right now than they were last week. That's different from **intent data**, which tracks online behaviour patterns (content downloads, anonymous browsing, third-party page visits) and infers interest. Both are useful. But signals are events, not patterns. They have a clear timestamp. They're public or semi-public. And they convert dramatically better when acted on within hours. Lars Nilsson, the operator who literally wrote the playbook on Account-Based Sales Development, has been pushing this distinction for a decade. The shift is just finally happening at scale because the tooling caught up. ## The 7 highest-leverage buying signals (ranked by predictive power) Not all signals are equal. Some convert at 30%+ when acted on quickly. Others — despite vendor marketing — convert at under 1%. Below: the 7 that actually work, ranked by what we have seen in 18+ scale-up GTM rebuilds at Stretch Innovation and Falora. ### 1. Past champion job changes (the highest-converting signal in B2B) When someone who previously bought your product joins a new company in a similar role, you have the strongest possible signal. They already know your product. They likely advocated for it before. They almost certainly have budget authority in their new role. And they are being measured on quick wins in their first 100 days. The data is unambiguous. **Champify's 2025 Impact Report** found known contacts deliver a 37% win rate compared to 19% for cold outreach. **UserGems** data shows past champions convert at **3× the rate of cold prospects, with 114% higher win rates, 12% shorter sales cycles, and 54% higher deal sizes.** **Why it works:** Trust is pre-built. The conversation starts at "yes, let's evaluate" instead of "who are you and why should I care." **How to detect it:** Tools like UserGems (specialist), Amplemarket (Duo Copilot, 100+ contact-level signals), Common Room (community + identity), or LinkedIn Sales Navigator alerts (manual but free). Build a list of every contact who ever had a relationship with your company — customers, champions, decisionmakers in evaluation, even referrers — and monitor them for job changes. **The play:** Reach out within 7–14 days of the move. Congratulate, don't pitch. Reference specific past results. Offer to help replicate at the new company. No discovery call ask — open the door, let them walk through it. ### 2. New executives in buyer roles (the 100-day window) When a new VP of Sales, CRO, CMO, or Head of Growth joins a company in your ICP, you have a 90–120 day window where they are under board pressure to make their mark. **UserGems data shows new executives spend 70% of their budget in the first 100 days, and convert at 2.5× higher rates in the first three months versus after their first year.** **Why it works:** Authority + urgency + budget + a need to show impact fast. The new exec is incentivised to act, not to "wait and see." **How to detect it:** LinkedIn Sales Navigator (free if you already have it), Cognism (GDPR-compliant European data), ZoomInfo, Apollo, or specialist tools like UserGems. Set up alerts for title changes within your target accounts. **The play:** Skip the "congrats on the new role" email. Everyone sends those. Instead, lead with: "Most new VPs of Sales in [their industry] hit one specific bottleneck in their first 90 days — here's a 1-page framework on it from someone who solved it." Provide value before asking for anything. ### 3. Funding rounds and M&A activity (caveat: timing matters) A company that just raised Series A/B/C has budget, ambition, and a board mandate to spend on growth. According to **Jolly Marketer's research, vendors contacting funded firms within 48 hours of announcement experience 400% higher conversion rates than those who delay**, and 71% of funded companies finalise vendors within 90 days of their announcement. **The catch:** Every vendor in your category gets the same Crunchbase alert. Every SDR is sending "congrats on the raise!" emails the same week. This is signal saturation in action — and it's why generic congratulations emails have become the new "just checking in." **Why it works (when done right):** New budget, ambition to deploy, often new strategic priorities tied to the raise (expansion, hiring, product investment). **How to detect it:** Crunchbase (general), Tracxn (Asia/EU coverage), CB Insights (enterprise), Specter, or specialist tools like Signalbase (real-time funding/M&A signals). For Belgian-specific data: Pitchdrive, BNP Paribas Fortis Innovation Hub, Sirris. **The play:** Don't congratulate on the raise. Reference the *strategic implication*. "Post-Series B, most fintech companies prioritise [specific operational shift]. Here's how three of our customers handled the same transition." Show that you have thought about what the funding *means*, not that you read TechCrunch. ### 4. Hiring surges (the leading indicator everyone underuses) Hiring patterns reveal where a company is investing before any press release confirms it. A company that doubled its engineering headcount in six months is building something. A company aggressively hiring salespeople is about to scale distribution. A company posting four "AI" titles in a quarter has a new initiative that didn't exist last year. **Why it works:** Hiring is a leading indicator of organisational priorities. By the time the press release comes, the budget is already deployed. Catching hiring trends 60–90 days early gets you in the conversation before procurement narrows the vendor list. **How to detect it:** LinkedIn job posting trackers, Loxo, ZoomInfo Hiring Intelligence, GrowthListings, RippleMatch (US), or scrape from Indeed/Welcome to the Jungle (EU). For pattern-matching: track velocity (hires per month) and role mix (e.g., 5 SDR roles + 1 Head of Demand Gen = scaling pipeline; 3 Data Scientists = AI initiative). **The play:** "Noticed you're scaling [team] aggressively — most companies in your stage hit [specific bottleneck] when they cross [headcount threshold]. Here's how [similar customer] navigated it." Concrete, situational, useful. ### 5. Tech stack changes and competitor churn When a company switches from Salesforce to HubSpot, drops their existing intent provider, or kicks off a tech-stack migration, they are in evaluation mode for everything adjacent. A company that just churned from your competitor is the highest-converting cold prospect available. **Why it works:** They've already validated the category. They've experienced the pain. They have an internal champion who has been advocating for change. And they are actively comparing alternatives. **How to detect it:** BuiltWith, Wappalyzer, HG Insights (enterprise), Slintel, technographic data from Cognism or 6sense. For G2 / Capterra review activity (a strong churn signal): G2's API, third-party monitoring, or just manually checking competitor review pages weekly. **The play:** Don't bash the competitor (it makes you look small). Lead with: "If [Competitor] isn't quite working for [specific use case], here are the three things our customers tell us they wish they'd known before switching." Then include actual content, not a pitch. ### 6. De-anonymised website intent (the dark funnel illuminator) Tools like RB2B, Demandbase Identity, Common Room, and Warmly have made it possible to identify a meaningful percentage of anonymous website visitors at the *contact* level (not just account level). When a VP of Sales at a target account hits your pricing page three times in a week, that is a higher-intent signal than any third-party intent data. **Why it works:** First-party signals beat third-party signals every time. The prospect has come to you. They are researching specifically your category, often specifically your product. **How to detect it:** RB2B (US person-level identification, GDPR-limited in EU), Common Room (community + first-party signals), Demandbase (account-level, broader EU coverage), Warmly, Albacross. For EU compliance: stick to legitimate-interest-defensible identification with proper LIA (Legitimate Interest Assessment) documentation, and avoid person-level identification of EU-based visitors unless you have explicit consent. **The play:** Speed kills here. The first vendor to reach a buyer after a trigger event is **5× more likely to win the deal** (Salesmotion analysis of 6sense data). Aim for outreach within 4 hours of high-intent activity. Reference the topic they were researching, not the fact that you tracked them (no "I noticed you were on our pricing page" — it is creepy). ### 7. Community and dark social engagement This one is underused because it's the hardest to operationalise. But: when someone in your ICP comments on a competitor's LinkedIn post, asks a question in a relevant Slack community, replies to your founder's content, or posts about a category-relevant problem on Reddit — that's a signal of active interest, often weeks before they ever hit your website. **Why it works:** Community engagement is the *earliest* visible signal. By the time someone fills out a form, they are already comparing 3–5 vendors. By the time they're commenting on a thoughtful post about your category, they are often just *starting* to think about a buying decision. **How to detect it:** Common Room (best-in-class for community signals — LinkedIn, Slack, Discord, GitHub, Reddit), Champify, manually monitoring relevant Slack communities (Pavilion, Exit Five, RevGenius, MeasureCamp), or LinkedIn Sales Navigator's "engaged with content" filter. **The play:** Don't pitch the engagement. Build a relationship. If someone asks a thoughtful question in a community, answer it well — publicly. If they comment on competitor content, engage with substance, not "happy to chat about how we do this differently!" Earn the right to a private conversation by being genuinely useful in public. ## Stacking signals: the multiplier most teams miss Single signals are good. Stacked signals are devastating. A company that just raised Series B (signal 3) AND hired a new VP of Sales (signal 2) AND is hiring 5 SDRs (signal 4) AND has a past champion in another role on the team (signal 1) is not a "maybe" account. It is an "almost certainly in market right now" account. Volume on these accounts is irrelevant. **Speed and personalisation are everything.** The benchmark from signal-stacked outreach across multiple platforms (Apollo, Salesforge, Amplemarket): **25–40% reply rates** versus 2–5% for generic cold outreach. **47% better conversion rates, 43% larger deal sizes, 38% more closed deals** (Landbase analysis of intent signal data). The lever isn't more signals. It is better orchestration of the signals you already could detect if you wired up the right loop. ## The EU compliance angle (because we're Belgian and this matters) Signal-based outbound is, perhaps counterintuitively, **the most legally defensible form of B2B cold outreach in Europe**. Here is why. GDPR Article 6(1)(f) allows processing personal data on the basis of "legitimate interest" provided you can document a Legitimate Interest Assessment (LIA). For B2B outreach, that LIA is much stronger when: 1. The contact is in a role where receiving relevant business communications is reasonably expected 2. The communication is genuinely relevant to their professional context (a signal proves this) 3. You provide easy opt-out 4. You comply with Article 14 source disclosure (if asked, you can tell them where you got their data) A signal — funding round, job change, hiring surge — is *evidence* of relevance. It is much harder for a regulator (or a prospect) to argue that an email about scaling sales infrastructure to a newly-hired VP of Sales is irrelevant. Compare that to a generic blast email to 5,000 contacts: virtually impossible to defend. The Belgian DPA's September 2024 guidance on AI systems and GDPR, plus EDPB Opinion 28/2024 on legitimate interest, both reinforce this: **personalised, contextually relevant outreach is the safer harbour, not blast volume.** Signal-based selling aligns regulatory and commercial incentives. They reward the same behaviour. (This is not legal advice. Document your LIA. Talk to your DPO. But the structural answer is clear.) ## A 30-day signal-based capture sprint Concrete week-by-week, no theory. **Week 1: Map your signals.** - Audit your last 20 closed-won deals. For each, identify what signal *would have* preceded the inbound (job change? new exec? funding? hiring spike?). This tells you which signals correlate with your specific buyer. - Pick the top 3 signal types from that list. Don't try all 7 at once. **Week 2: Build the detection layer.** - Set up tooling for your 3 chosen signals. Minimum stack: UserGems (or LinkedIn Sales Nav alerts) for job changes/exec moves, Crunchbase or Specter for funding/M&A, RB2B or Common Room for first-party + community signals. - For Belgian/EU specifically, prioritise GDPR-compliant providers: Cognism, Apollo (with EU data residency), Salesflare for CRM-native enrichment. **Week 3: Build 3 signal-specific playbooks.** - One sequence per signal type. Each sequence: 3–4 touches over 10–14 days for high-priority signals. First message references the specific signal (not vaguely — specifically). Subsequent messages add value, social proof, and a clear CTA. - Write the messages yourself or have a single accountable human write them. Outsourced sequences sound outsourced. Buyers can tell. So can the LLMs that increasingly screen incoming messages. **Week 4: Wire the speed loop.** - Signal fires → notification to the relevant rep (Slack, email) within 30 minutes - Rep reaches out within 4 hours for high-priority signals (past champion, exec change), within 24 hours for medium (funding, hiring), within 48 hours for low-urgency (tech stack) - Track signal-to-outreach time as a KPI. The teams that win are not the ones with the most signals — they're the ones that act on signals fastest. **Week 5+: Measure, then layer.** - Track three tiers of metrics: activity (signal-to-outreach speed, signal coverage rate), engagement (reply rate per signal type, meetings booked), revenue (pipeline by signal source, win rate by signal type). - Once you have 4 weeks of data, identify your top-performing signal and double down. Add a 4th signal type only after the first 3 are stable. ## What to avoid A few traps we see repeatedly: **Don't blast on funding signals.** Everyone gets the Crunchbase alert. Generic "congrats on the raise!" emails have zero conversion. Reference the strategic implication, not the press release. **Don't confuse intent data with signals.** Anonymous browsing patterns are useful as account-level prioritisation. They're terrible as outreach triggers — too vague, too late. Real signals have timestamps and named events. **Don't outsource signal-based selling to an agency.** The whole point is speed and contextual relevance. Agencies are slow, generic, and break the LIA defensibility argument. If you outsource, you are back to volume — just with better data. **Don't let your stack get bigger than your team can run.** A scale-up with 3 signal sources running well beats one with 10 signal sources running poorly. Pick the signals that match your buyer. Build deep, not wide. ## What this looks like when it works A €15M ARR Belgian SaaS company we worked with at Stretch Innovation switched from a 4-SDR volume motion (15,000 emails/month, 1.2% reply rate, ~12 meetings/month) to a 1-SDR signal-based motion (1,800 highly-targeted emails/month, 18% reply rate, ~38 meetings/month) over 90 days. Headcount: -3 SDRs. Cost: -€220K/year. Pipeline: +160% qualified meetings. Reply rate: +1,400%. CAC: down 53%. That is not magic. That is just stopping the spray. The signals were always there. The math was always there. What was missing was the discipline to wire detection to outreach, fast. If your reply rates are below 3% and your team is grinding, the question isn't "how do we send more." It's "what signal are we ignoring." Find it. Wire it. Watch what happens. ## Conclusion Falora.ai builds the autonomous GTM layer that turns buying signals into outreach in minutes, not weeks. Built EU-compliance-native, drawn from 18+ scale-up GTM rebuilds at Stretch Innovation. **[Book a 30-minute GTM diagnostic with Stijn →](https://meetings-eu1.hubspot.com/stijn-van-daele/falora)** --- ### Sources - [6sense — *2025 Buyer Experience Report*](https://6sense.com/blog/new-buyer-experience-report-3-insights-and-a-warning/) - [Gartner — *Sales Survey on B2B Buyer Outreach Preferences*](https://www.gartner.com/en/sales) - [Belkins — *Cold Email Benchmark Study* (16.5M emails, 2025)](https://belkins.io) - [Champify — *Known Contacts Impact Report* (2025)](https://www.champify.io) - [UserGems — *Past Champion + New Executive Conversion Data*](https://www.usergems.com) - [Jolly Marketer — *B2B Trigger Events Research*](https://jollymarketer.com) - [Landbase — *Intent Signal Data Analysis*](https://www.landbase.com) - [Apollo — *Signal-Based Selling Framework*](https://www.apollo.io) - [Salesmotion — *Signal-Based Sales Playbook*](https://www.salesmotion.io) - [Amplemarket — *Signal-Based Selling Complete Guide*](https://www.amplemarket.com) - [EU AI Act, Regulation EU 2024/1689](https://eur-lex.europa.eu/eli/reg/2024/1689/oj) - [Belgian Data Protection Authority — *AI Systems and the GDPR*](https://www.dataprotectionauthority.be) ### Related reading on Falora - [Demand creation vs demand capture](https://falora.ai/blog/demand-creation-vs-demand-capture) - [GTM engineering vs growth marketing](https://falora.ai/blog/gtm-engineering-vs-growth-marketing) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) - [AI SDR failure modes](https://falora.ai/blog/ai-sdr-failure-modes) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. Over the last 6 years he and his team helped more than 200 companies with building a growth engine that really scaled their company. He writes about GTM engineering, autonomous revenue and the EU AI Act on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What is signal-based selling?** Signal-based selling triggers outreach the moment an observable buying signal fires (job change, funding round, hiring surge, tech stack switch, website intent), within hours, instead of blasting the entire TAM on a fixed calendar cadence. The result is dramatically higher reply rates because the timing and context match the buyer's actual situation. **Which buying signals convert best in B2B?** In our work with 18+ scale-up GTM rebuilds the 7 highest-leverage signals are: (1) past customer or champion job changes, (2) new executives in buyer roles, (3) funding rounds and M&A, (4) hiring surges, (5) tech stack changes and competitor churn, (6) de-anonymised website intent, and (7) community and dark social engagement. Past champion moves are the highest-converting single signal. **How fast do you need to act on a buying signal?** Speed is the lever. Aim for outreach within 4 hours for high-priority signals like past-champion moves and exec changes, within 24 hours for funding and hiring, and within 48 hours for tech stack changes. The first vendor to reach a buyer after a trigger event is roughly 5× more likely to win the deal. **Is signal-based outbound GDPR compliant in the EU?** Signal-based outbound is the most legally defensible form of B2B cold outreach in Europe. GDPR Article 6(1)(f) legitimate interest is much stronger when the contact's role makes business communication reasonably expected and the message is genuinely relevant — and a fired signal is direct evidence of relevance. Document your Legitimate Interest Assessment, comply with Article 14 source disclosure, and consult your DPO. **What tools do I need for signal-based selling?** Minimum stack: UserGems or LinkedIn Sales Navigator alerts for job changes and exec moves; Crunchbase, Specter or Signalbase for funding and M&A; Common Room, RB2B or Warmly for first-party and community signals; BuiltWith or HG Insights for tech stack signals. For EU coverage, prioritise GDPR-compliant providers like Cognism, Apollo (with EU data residency) and Salesflare. --- # Demand creation vs demand capture: why most B2B scale-ups have the ratio wrong - URL: https://falora.ai/blog/demand-creation-vs-demand-capture - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-04-28 - Category: Insights - Tags: demand-generation, demand-creation, demand-capture, b2b-saas, gtm-strategy - Reading time: 11 min - Translation key: demand-creation-vs-capture ## TL;DR Demand creation makes people want what you sell; demand capture turns existing want into pipeline. Most B2B scale-ups run a 90/10 capture-heavy split when the evidence supports 60/40 the other way. The cost of staying broken: rising CAC, falling reply rates, and a brand nobody remembers when the buying window opens. ## Description Most B2B scale-ups run a 90/10 capture-vs-creation ratio. The evidence says 60/40 the other way. Here is what the broken ratio costs and how to fix it. ## Full content ## TL;DR > Most B2B scale-ups think they have a demand problem. They don't. They have a ratio problem. - 90% of marketing budget goes to demand capture; the evidence says it should be closer to 60% to demand creation. - 95% of buyers are not in the market right now. Capture-only spend competes inside the 5%, which is why CAC keeps climbing. - 84% of deals are won by the first vendor a buyer contacts, and 77% of buying groups stick with their preliminary favourite. - The fix is not killing capture. It is funding a creation engine alongside it and tracking the Demand Ratio quarterly. ## Introduction Walk into any €5M–€50M ARR scale-up in Belgium, the Netherlands, or anywhere in Europe right now and you will find roughly the same setup: 90% of the marketing budget tied to demand capture (paid search, retargeting, intent-based outbound, MQL hunting), and a tiny sliver — if anything — going to demand creation. The financial consequences of that ratio are bigger than most CMOs realise. And they are getting worse, not better. This article makes the maths explicit and gives you a 90-day rebalancing plan. ## The 30-second answer Demand creation makes people *want* what you sell. Demand capture turns existing want into pipeline. Both matter. But the ratio in most B2B scale-ups is upside down: roughly 90/10 in favour of capture, when the evidence suggests it should be closer to 60/40 the other way. The reason it is broken: capture is measurable in 30 days, creation pays back in 6–18 months — and most boards reward what shows up next quarter. The cost of staying broken: rising CAC, falling reply rates, and a brand nobody remembers when the buying window finally opens. If that hits a nerve, keep reading. The numbers below are not an opinion. ## The 95:5 rule, restated honestly In 2021, Professor John Dawes of the Ehrenberg-Bass Institute published a deceptively simple finding: at any given moment, roughly **95% of B2B buyers are not in the market** for what you sell. Only 5% are. Dawes was upfront that the percentages aren't precise. *"The 95% figure is not meant to be a precise rule. We're using it as a heuristic to get the idea across that the vast majority of businesses, for a large proportion of products, are not in the market in particular time periods."* Source: [John Dawes, Ehrenberg-Bass Institute, 2021](https://johndawes.info/the-955-rule/). The exact split varies by category, but the structural insight does not: business buyers move themselves into the market based on need and budget cycles, not because a clever ad pushed them there. As Dawes puts it: *"Marketers don't move buyers in-market — buyers move themselves in-market based on their needs."* This is where the demand creation vs demand capture distinction stops being a marketing-blog debate and starts being a maths problem. If 95% of your total addressable market isn't buying right now, then every euro you spend exclusively on the 5% is competing inside a tiny, hyper-priced auction with every other vendor in your category. CAC goes up. Reply rates go down. Your CFO writes "is this still working?" in the margin of next quarter's review. The answer to that question: yes, it's working — for the 5%. It is the other 95% you're invisible to. ## What demand capture actually is (and where most teams over-invest) **Demand capture** is the set of activities that intercept buyers who already know they have a problem and are actively researching solutions. It includes: - Paid search on bottom-of-funnel keywords ("best CRM for Belgian SMBs", "ERP migration consultant") - Retargeting visitors who hit your pricing page - Outbound to accounts showing intent signals (job changes, funding, tech stack shifts) - Comparison content ("X vs Y") and review-site presence (G2, Capterra, Sourceforge) - SDR sequences targeting handraisers and demo-form fillers None of this is bad. We at Falora build a chunk of our own pipeline this way and we help our clients do the same. The problem isn't capture itself. The problem is that capture is the *only* thing most scale-ups invest in seriously, because capture is the only thing their attribution software shows them clearly. ## What demand creation actually is (and why most teams under-invest) **Demand creation** is the set of activities that put your brand into a buyer's memory *before* they enter the buying window — so that when their need finally surfaces, you are already on their mental shortlist. In practice, that looks like: - A founder or operator publishing a strong point of view on LinkedIn 3–5 times a week - An owned podcast where your ideal customer's peers tell their stories - Genuine participation in communities (Slack groups, Reddit, niche LinkedIn communities, Pavilion, Exit Five) - Long-form content that takes a stance, not a "10 tips" listicle - Conference talks, guest essays, podcast appearances on shows your buyer already listens to - Distinctive brand assets — colours, voice, recurring formats — that buyers can recognise instantly The thing about every item on that list: none of it is reliably trackable in your attribution dashboard. As Chris Walker, founder of GTM consultancy Passetto and former Refine Labs CEO, has been hammering for years: *"Marketing teams don't Create Demand because their attribution models & KPIs don't incentivize them to do it. And it's really that simple. Because CREATING DEMAND involves using dark social channels — mainly 3rd party content platforms, social networks, and communities that every B2B buyer uses today — in ways that don't get measured with attribution software."* Source: [BEAM, 2024](https://www.beamcontent.co/resources/marketers-play-offense). This is the trap. Capture is measurable. Creation isn't — at least not by software alone. Boards reward measurable. Boards punish "trust me." So scale-up CMOs do the rational thing: they over-fund capture and under-fund creation. And then they wonder why their CAC keeps creeping up. ## The ratio most scale-ups are actually running (and why it is broken) Walker's broader claim — that more than 95% of commercial budget at most B2B SaaS companies is allocated to capture — is a useful provocation, even if the exact number varies. In our work with B2B scale-ups across Belgium and the Netherlands, the typical split looks roughly like this: | Bucket | Typical scale-up allocation | Examples | |---|---|---| | **Demand capture** | **85–95%** | Paid search, retargeting, outbound team, SDR tooling, intent platforms, MQL forms | | **Demand creation** | **5–15%** | LinkedIn content, podcast (if any), some thought leadership, occasional event | Now compare that to what the LinkedIn B2B Institute, in collaboration with Les Binet and Peter Field, recommends: **roughly 60% to brand building (creation) and 40% to activation (capture)** for optimal long-term growth in B2B. As 6sense's research team put it in their 2026 analysis: *"Brand marketing creates demand that didn't previously exist and ensures you're discovered during the 60% of the journey buyers spend in the dark funnel."* Source: [6sense, 2026](https://6sense.com/guides/how-genai-and-llms-are-changing-b2b-buyer-research-and-how-to-respond/). So: prescribed 60/40 toward creation. Actual 90/10 toward capture. That is not a small calibration error. That is the wrong direction entirely. ## What that ratio actually costs you Let's stop arguing about percentages and look at what the broken ratio is doing to scale-ups in practice. **1. The vendor selected before sales gets a call.** 6sense's 2025 Buyer Experience Report, based on more than 4,000 buyers globally, found that **84% of deals are won by the first vendor a buyer contacts**. The report notes: *"This indicates that buyers have usually picked a winner and are merely conducting due diligence when speaking with other vendors."* Source: [6sense, 2025](https://6sense.com/blog/new-buyer-experience-report-3-insights-and-a-warning/). Even more sobering: **94% of buying groups had ranked their preferred vendors before making first contact** — and the preliminary favourite ultimately won 77% of the time. If your brand isn't on that pre-shortlist, you're not losing the deal in discovery. You lost it months ago, when the buyer was reading, listening, scrolling — and you weren't there. **2. The dark funnel keeps eating your attribution.** 6sense has tracked it for years: roughly **70% of the B2B buyer journey happens anonymously**, before any vendor contact. In the most recent 2025 report, the point of first contact has moved up slightly to ~60% (driven mostly by AI-related uncertainty pulling buyers to vendors earlier for validation), but the principle holds: most of the journey is invisible. The implication: if you only measure what your software can track, you are literally measuring the smallest part of the buying process. And then making budget decisions based on it. **3. CAC keeps creeping up.** Performance marketing in B2B has gotten more expensive every year for the last five. Click costs on bottom-funnel keywords like "best CRM" or "ABM platform" have roughly doubled since 2020. When everyone in your category is bidding on the same 5%, the only thing that goes up is the auction price. **4. Buyers actively avoid suppliers who can't read the room.** Gartner's 2025 sales survey of 632 B2B buyers found that **73% actively avoid suppliers who send irrelevant outreach.** Three quarters of your addressable market will close the door if you knock at the wrong time with the wrong message — and untargeted volume capture is the fastest way to knock wrong. **5. The agency dependency.** Many scale-ups outsource creation to agencies who promise "thought leadership content." This usually produces ghost-written LinkedIn posts that read like they were ghost-written and a podcast nobody at the company listens to. Demand creation that gets outsourced rarely sounds like a human being with a strong opinion. And LLMs and buyers can both tell. ## The counter-argument, taken seriously To be fair, there is a real critique of the pure "creation > capture" thesis worth engaging with. **Kyle Poyar** (formerly OpenView, now Tremont) has pointed out repeatedly that for early-stage B2B SaaS — particularly bootstrapped or seed-stage companies with 10 customers and a runway counted in months — pure brand-building is a luxury. You need closed-won this quarter. That's a fair critique. Pre-product-market-fit, capture should dominate. **Sangram Vajre** and the GTM Partners team argue that the "GTM Operating System" view is more nuanced than a creation-vs-capture binary: there are multiple plays running in parallel, and the right mix depends on stage, ICP, and product motion. **6sense's Kerry Cunningham** has gently pushed back on the most extreme version of the dark-social-eats-everything narrative, noting that traditional capture channels still drive substantial pipeline — they just aren't the *whole* picture. We agree with all of these caveats. The honest framing is not "creation good, capture bad." It is: **the right ratio depends on stage, but virtually every scale-up we work with is over-indexed on capture relative to where they actually are.** A €15M ARR Series B SaaS company with 200+ customers running 90/10 capture-heavy is not optimising — they are under-investing in the engine that compounds. ## A framework: the Falora Demand Ratio Here is a diagnostic we use with scale-ups before any rebuild. We call it the **Demand Ratio**: the percentage of your last 12 months of pipeline that came from demand creation activities versus demand capture activities. **Step 1 — categorise your pipeline sources from the last 12 months into two buckets:** *Creation-sourced pipeline:* - Inbound that filled out a form citing "I follow your founder on LinkedIn / I listen to your podcast / I read your essays" - Self-reported attribution from won deals ("how did you hear about us") - Word-of-mouth referrals - Inbound demo requests with no traceable last-click attribution - Branded search (someone Googling your company name directly) *Capture-sourced pipeline:* - Paid search on category keywords - Retargeting-attributed conversions - Outbound from cold lists - Intent-data-triggered outbound - Review-site referrals (G2, Capterra) - "Compare X vs Y" pages **Step 2 — calculate the ratio.** Most scale-ups we run this exercise with land somewhere between 5/95 and 20/80 (creation/capture). The benchmark we encourage them to push toward over 18 months: **40/60 minimum**, ideally 50/50. **Step 3 — track it quarterly.** Not because the number is sacred, but because the *direction of travel* matters. If your Demand Ratio is moving toward more creation-sourced pipeline year-over-year, you are building compounding equity. If it is static or moving the other way, your CAC is going to keep climbing. **Step 4 — be honest about self-reported attribution.** Add a "How did you hear about us?" field to your demo form, your discovery call script, and your win/loss interviews. The data is messy and biased, but it's a meaningful signal that no UTM tracker will ever capture. As Chris Walker advocates: automate self-reported attribution in Salesforce, have sales reps ask on first call, use conversation intelligence tags to capture it consistently. ## A 90-day rebalancing plan If you've read this far and your Demand Ratio is south of 20/80, here is a concrete sequence we use with scale-ups starting tomorrow. **Days 1–14: Diagnose.** Run the Demand Ratio calculation on your last 12 months. Add the self-reported attribution field to all inbound forms and discovery calls. Pull win/loss data and tag every deal with its earliest creation touchpoint, even if your CRM didn't capture it (you'll have to ask). **Days 15–45: Pick one creation channel and commit to it for 6 months.** Don't try to do all five at once. Pick one — usually founder-led LinkedIn, an owned podcast, or a series of PoV essays — and resource it like it matters. One person, named, accountable. Three to five outputs a week minimum. No agency. The voice has to be human and yours. **Days 30–60: Don't kill capture, audit it.** Look at your capture spend by channel and find the bottom 30% by cost-per-pipeline-euro. Cut those. Reallocate half the saved budget into creation, the other half into your two best-performing capture channels. Most scale-ups find €50K–€150K of waste in their first audit. **Days 60–90: Build measurement that respects creation.** Stop expecting your attribution software to credit creation channels — it can't and won't. Instead build a hybrid scorecard: - **Capture metrics:** software-attributed pipeline, MQLs, meeting volume, CAC - **Creation metrics:** branded search volume (Google Trends), podcast download rate, LinkedIn impressions on founder content, share of voice in your category, self-reported attribution share, win rate when buyer mentions a creation touchpoint Review them in the same meeting. Make trade-offs based on both. **Day 90: Re-run the Demand Ratio.** You won't have moved it dramatically in 90 days — creation compounds slowly. But you should see directional shift in the leading indicators (branded search, content engagement, self-reported attribution). If you don't, your creation execution is the problem, not the strategy. ## The European angle: why this matters more here A specifically European complication. B2B sales cycles in Belgium and the Netherlands are long, and decision-making units are wide. 6sense's 2024 European research found Belgian sales cycles average around **11.5 months with a DMU of 11.3 stakeholders** — meaningfully larger than the Dutch (~7.9 months / 7.6 DMU) or US averages. What does that mean? It means buyers in this region spend even more of their journey in the dark funnel before talking to a vendor. A company whose brand is invisible to 11 stakeholders for 11 months will lose to a company whose brand has been showing up in 3 of those stakeholders' LinkedIn feeds for 11 months. The ratio matters more here, not less. Add the EU AI Act layer (Article 50 transparency obligations from August 2026; AI literacy duty already in force since February 2025) and the cost of pure capture rises further: AI-driven outbound at scale is becoming a regulatory and reputational risk in a way that owned content with a real human voice simply isn't. ## What this means if you are a CMO or Head of Growth right now If your board is leaning on you for next-quarter pipeline: don't blow up your capture motion. We are not advocating for that. The capture engine pays the bills. But if you are not actively building a creation engine alongside it — funded, named, protected — you are choosing future CAC inflation. The 5% who are in-market today will be choosing between vendors they remember from the last 12 months. If you weren't there, you are not in the consideration set. And as 6sense's data shows, **77% of the time, the buyer's preliminary favourite wins.** The math is brutal in one direction: the longer you stay 90/10 capture, the more expensive every quarter gets. The math is forgiving in the other direction: every euro you redirect to creation today buys you cheaper pipeline 12 months from now. And every quarter you delay is a quarter your competitors are building memory you'll have to outbid later. Pick one creation channel. Commit for six months. Watch the ratio shift. That's it. That's the playbook. ## Conclusion Falora.ai builds autonomous GTM software for B2B scale-ups that have stopped pretending capture-only growth is sustainable. Born out of Stretch Innovation's 18+ scale-up GTM rebuilds across the Benelux. **[Book a 30-minute GTM diagnostic with Stijn →](https://meetings-eu1.hubspot.com/stijn-van-daele/falora)** --- ### Sources - [John Dawes, Ehrenberg-Bass Institute — *The 95:5 Rule*](https://johndawes.info/the-955-rule/) - [LinkedIn B2B Institute — *The 95-5 Rule* research with Ehrenberg-Bass](https://business.linkedin.com/marketing-solutions/b2b-institute) - [6sense — *2025 Buyer Experience Report*](https://6sense.com/blog/new-buyer-experience-report-3-insights-and-a-warning/) - [6sense — *How GenAI and LLMs are changing B2B buyer research*](https://6sense.com/guides/how-genai-and-llms-are-changing-b2b-buyer-research-and-how-to-respond/) - [Chris Walker — Public posts and interviews via BEAM](https://www.beamcontent.co/resources/marketers-play-offense) - [Gartner — *Sales Survey on B2B Buyer Outreach Preferences*](https://www.gartner.com/en/sales) - [Kyle Poyar — *Growth Unhinged*](https://www.growthunhinged.com) - [Les Binet & Peter Field — *The Long and the Short of It* (LinkedIn B2B Institute)](https://business.linkedin.com/marketing-solutions/b2b-institute) ### Related reading on Falora - [GTM engineering vs growth marketing](https://falora.ai/blog/gtm-engineering-vs-growth-marketing) - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. Over the last 6 years he and his team helped more than 200 companies with building a growth engine that really scaled their company. He writes about GTM engineering, autonomous revenue and the EU AI Act on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What is the difference between demand creation and demand capture?** Demand creation builds future buying intent through brand, point-of-view content, founder-led social, podcasts and community. Demand capture intercepts buyers who already know they have a problem through paid search, retargeting, outbound to handraisers, and review-site presence. Both matter; the issue is the ratio between them. **What is the right demand creation vs demand capture ratio for B2B?** LinkedIn B2B Institute and Ehrenberg-Bass research, applied through Les Binet and Peter Field, points to roughly 60% creation and 40% capture for optimal long-term B2B growth. Most scale-ups today run closer to 10/90 the other way, which is the structural cause of CAC inflation. **What is the 95:5 rule and why does it matter?** Professor John Dawes of the Ehrenberg-Bass Institute showed that at any given moment roughly 95% of B2B buyers are not in the market. Only 5% are buying now. Capture-only marketing competes for the 5%; demand creation builds memory in the 95% so you are on their shortlist when they enter the market. **How do I measure demand creation if attribution software cannot track it?** Use a hybrid scorecard. Capture metrics: software-attributed pipeline, MQLs, meeting volume, CAC. Creation metrics: branded search volume, podcast downloads, founder-content engagement, share of voice, self-reported attribution from won deals, and win rate when buyers cite a creation touchpoint. **Where should an early-stage B2B SaaS company invest first?** Pre-product-market-fit, capture should dominate because closed-won this quarter matters more than memory built for next year. Once you cross €5M ARR with 100+ customers, the broken ratio becomes the binding constraint and creation needs a named, funded engine alongside capture. --- # AI SDRs vs human SDRs: 7 conditions where AI loses - URL: https://falora.ai/blog/ai-sdr-failure-modes - Language: en - Author: Jeroen De Broyer (Co-founder, Falora) - Published: 2026-04-21 - Category: Opinion - Tags: ai-sdr, outbound, sales-development, honest-takes - Reading time: 14 min - Translation key: ai-sdr-failure-modes ## TL;DR AI SDRs are not a product category. They are a wager about your offer. We tested 11 BE/NL deployments; 4 produced positive ROI within 90 days, 7 did not. The 7 conditions below predict which side you land on. ## Description Honest analysis: 7 conditions in which AI SDRs underperform human SDRs, plus the 4 scenarios where they win. The fitness test for your pipeline. ## Full content ## TL;DR > AI SDRs are not a product category. They are a wager about your offer. If your offer cannot convert a 10-minute-research email written by Eric Nowoslawski, automating bad emails 10,000× is not growth. It is brand damage at scale. - We deployed AI SDR platforms for 11 Belgian and Dutch B2B scale-ups in 2025. 4 produced positive ROI within 90 days. The 7 losers shared structural conditions, not vendor choice. - The 7 conditions below predict failure. Pass them before you write a vendor cheque. - The 4 winning scenarios are real. We will steelman them honestly. - The replacement for "AI SDR vs human SDR" thinking is GTM engineering; the system underneath both. ## Introduction Almost every "AI SDR" benchmark you read is written by a vendor selling AI SDRs. This one is not. We have helped 11 European B2B SaaS scale-ups deploy or kill AI SDR platforms in the last 12 months. Four worked. Seven did not. The pattern in why they failed is consistent enough to write down. This article is for the CMO, Head of Growth or RevOps lead who has either piloted an AI SDR for 60–90 days and is disappointed, or is in late-stage evaluation and second-guessing. The seven conditions below are the diagnostic. If three or more apply to you, the AI SDR is not the cure. If none apply, the AI SDR will likely work; and the deeper question is whether you should buy a single-purpose AI SDR or build the broader GTM engineering system that replaces the category entirely. ## Condition 1: Your offer has not earned the right to be cold-pitched This is the failure mode nobody wants to discuss. An AI SDR multiplies your existing message by a factor of 10,000. If the underlying message does not earn a response from a 10-minute-research human-written email, it will not earn a response from 10,000 AI-written ones. It will damage your brand at scale. Anthony Pierri's positioning work makes this specific: a positioning statement that your prospect cannot summarise in one sentence after reading your homepage is not yet a position. A position that is not a position cannot survive cold contact. As Jordan Crawford of Blueprint GTM puts it: > "AI SDRs are a bad idea for most companies. AI can replace about 90% of SDR tasks, but the crucial 10% still requires human insight and empathy." The 10% is the part that decides whether the message is worth opening. Until your offer can win the 10%, automating the 90% is premature. ## Condition 2: Your TAM is below 2,000 accounts AI SDR economics depend on volume. The math that makes a €5,000-per-month platform pay back assumes hundreds of conversations per month, which assumes thousands of in-market accounts. If your TAM is below 2,000 accounts; true for many B2B teams selling to specialised verticals (port logistics, insurance brokerage, biotech tooling, regulated finance); the AI SDR motion will burn through your TAM in 8–12 weeks and leave you with brand damage and no remaining audience. In this scenario the right approach is **named-account selling** with one or two senior reps doing 2–4 hours of research per account, supported by a GTM engineer who builds the signal layer. The platform exists; the agent does not send the messages. ## Condition 3: Your buyer is in DACH or Germany Germany is structurally hostile to AI-driven outbound at the moment. Three reasons stack. First, German UWG case law is the strictest B2B consent regime in the EU. The default assumption is opt-in, with narrow exceptions for existing business relationships. Second, EU AI Act Article 50 (in force August 2026) requires AI-content disclosure with material consequences in the German enforcement context. Third, German B2B buyers have higher scepticism baselines for AI-drafted outreach than their French, Dutch or Belgian counterparts. The cultural reception is colder. The combined effect is that the same AI SDR motion that produces 1.2% reply rates in the Netherlands produces 0.3% reply rates in Germany. The unit economics do not work below 0.5%. Mitigation: keep a human-led motion in DACH while running AI in BENELUX, France, Iberia, Italy and the Nordics. ## Condition 4: Your ACV is above €100K with a 10+ stakeholder buying group 6sense's research shows the average B2B buying group is **10.1 stakeholders**, with **75% of buyers using backchannel research before contacting sales**. For deals above €100K ACV, the multi-threading work; finding the economic buyer, the technical buyer, the procurement gate, the security review, the legal sign-off; is fundamentally human relationship work. An AI SDR can do the initial outreach. It cannot navigate the 10-stakeholder buying group. As Mark Kosoglow, founder of Operator, puts it: **multi-threading is the work**, not the prelude to the work. Skipping the human relationship at this ACV destroys win rates more than it improves throughput. For €100K+ ACV motions the right architecture is: GTM engineering layer feeds qualified accounts to a senior rep; the rep does the multi-thread; the platform handles the surrounding cadence and research. This is a Level 3 hybrid, not a Level 4 autonomous motion. ## Condition 5: You need brand and warm-up before outbound The LinkedIn B2B Institute's 95-5 rule states that **95% of your audience is not in-market this quarter**. The 5% who are in-market choose vendors based on brand recall built over the previous 18–24 months. An AI SDR that reaches the 5% with no brand recall in their head produces a 0% close rate against a competitor with brand recall. Chris Walker's argument on dark social applies directly: 97% of net new ARR for modern SaaS is attributable to attribution-blind channels. An AI SDR that bypasses the brand-building channel and goes straight to outbound is solving for the wrong end of the funnel. If your unaided brand recall in your ICP is below 5%, the right investment is in the dark social engine; podcasts, communities, peer-to-peer content; not in an AI SDR. ## Condition 6: Your data layer is broken Kyle Poyar and Maja Voje's *State of B2B GTM 2025* shows **85% of enterprise sellers manage their book in spreadsheets**, with **only 5% using the CRM as their day-to-day work surface**. AI SDR is downstream of data quality. If your CRM is wrong, your AI SDR will confidently write to the wrong person about the wrong problem. The pre-condition for AI SDR success is a data layer that is at least 80% accurate at the contact level and 90% accurate at the firmographic level. Most teams below €10M ARR are not there. The fix is not a better AI SDR. It is a six-week data hygiene sprint, ideally led by a GTM engineer, before any agent touches a send button. ## Condition 7: Your sales leadership thinks AI SDR = headcount replacement The cultural failure mode. If your CRO has briefed the board that the AI SDR will let them cut three SDR seats, the deployment is set up to fail in two ways. First, the remaining team will not invest in making it work. Second, the platform will be measured against unrealistic throughput targets that obscure the actual unit economics. The deployments that succeed in our portfolio all share one pattern: the AI SDR is positioned as a **system that lets the existing team do more**, not as a **replacement for headcount**. Headcount adjustments follow 6–9 months later, after the system has proven itself, and they happen quietly. ## When AI SDRs win (the 4 honest scenarios) We promised steelman. Here it is. **High-volume mid-market SaaS with a proven offer.** ICP is 5,000–50,000 accounts. ACV is €10K–€60K. The offer wins on cost-of-inaction, not on relationship. Your existing SDR motion produces 1–2% reply rates with human writers. The AI SDR can credibly produce 0.8–1.5% reply rates at 4× the volume; net positive. **Re-engagement of stale CRM.** You have 80,000 contacts in HubSpot from three years of marketing. 95% of them have not been contacted in 12+ months. An AI SDR with smart segmentation and re-engagement workflows can extract residual demand at a cost that no human team can match. **Multilingual outreach without native operators.** You are a Belgian or Dutch scale-up trying to do outbound in Spanish, Italian or Polish without native staff. Modern AI SDRs are fluent in these languages and will out-perform a non-native human writer on basic outreach. **Research-heavy, enrichment-driven plays.** The motion depends on assembling 5–10 data points per account before contact (recent funding, hiring signals, tech stack changes, leadership moves). Human SDRs do this slowly and inconsistently. AI SDRs do it quickly and consistently. As Eric Nowoslawski of Growth Engine X says: > "The best email you send is the one where you do 10 minutes of research." The AI SDR compresses the 10 minutes into 90 seconds. When research is the differentiator, the platform wins. ## What replaces "AI SDR vs human SDR" thinking The framing is wrong because the choice is not binary. The team that wins in 2026 does not pick one or the other. It builds the [GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) underneath both, then chooses the right activator (AI agent, human SDR, partner channel, in-product growth) per signal-segment combination. This is the deeper version of the [Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model). At Level 2, you have an AI SDR. At Level 3, you have a system that decides when to deploy the AI SDR and when to escalate to a human. The conversation about "AI SDR vs human SDR" belongs to Level 2. We are arguing for Level 3. Jen Allen-Knuth of DemandJen reminds us why this matters: > "B2B buyers don't trust vendors. They trust their peers and other customers." A Level 3 system is the only architecture that can route the right conversation to the right activator; peer review, customer story, AI-drafted opener, human relationship; based on what the buyer actually needs in that moment. ## A 30-day diagnostic: keep, kill or rebuild your AI SDR If you have an AI SDR running today, here is the 30-day decision framework. **Days 1–7: data baseline.** Pull cost per qualified meeting, cycle time from signal to first reply, percentage of conversations escalated to a human, reply rate, and brand-keyword search lift over the period the platform has been running. If you cannot pull all five, the platform is not instrumented enough to evaluate. Fix that first. **Days 8–21: against-the-7 audit.** For each of the seven failure conditions above, score your current state. If three or more apply, the platform is set up to fail. **Days 22–30: decision.** Three options. Keep if the data is improving and fewer than three conditions apply. Rebuild if the data is flat but the conditions check out (the platform configuration is the problem, not the strategy). Kill if three or more conditions apply and the data is not moving. Most teams that arrive at this diagnostic decide to rebuild rather than kill. The platform is rarely the bottleneck; the system around it is. ## Conclusion The AI SDR category is the most over-sold subgenre of B2B GTM tooling in 2025–26. It is also genuinely useful in 4 specific scenarios. The job for a CMO or Head of Growth is not to be sceptical of the category. It is to be precise about which scenario you are in. The 7 conditions above are the diagnostic. The 4 winning scenarios are the green light. The replacement for the category, when you are ready, is a Level 3 GTM engineering system. If you want to run the diagnostic with an operator who has done it 11 times, **[take the AI SDR fitness test →](https://falora.ai/ai-sdr-fitness-test)** --- ### Sources - [Jordan Crawford, Blueprint GTM](https://www.linkedin.com/in/jordancrawford1) - [Eric Nowoslawski, Growth Engine X](https://www.linkedin.com/in/eric-nowoslawski) - [6sense, *B2B Buyer Experience Report 2025*](https://6sense.com/resources) - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) - [LinkedIn B2B Institute, *95-5 Rule, Dawes / Ehrenberg-Bass*](https://business.linkedin.com/marketing-solutions/b2b-institute) - [Kyle Poyar & Maja Voje, *State of B2B GTM 2025*](https://www.growthunhinged.com) - [McKinsey, *State of AI 2025*](https://www.mckinsey.com) - [Sapphire Ventures, *AI SDR market analysis 2025*](https://sapphireventures.com) ### Related reading on Falora - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) - [The EU AI Act + your AI GTM stack](https://falora.ai/blog/eu-ai-act-gtm-stack-audit) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) ### About the author Jeroen De Broyer is co-founder of Falora. He writes on [LinkedIn](https://www.linkedin.com/in/jeroendebroyer). ## FAQ **Do AI SDRs actually work?** Sometimes. In our benchmark of 11 Belgian and Dutch B2B deployments, 4 produced positive ROI within 90 days and 7 did not. The four that worked all shared three traits: an offer with a clear cost-of-inaction, a TAM above 5,000 accounts, and an existing brand presence in dark social. The seven that failed shared the inverse. **What is the difference between an AI SDR and an AI BDR?** Functionally none. Both terms describe an AI-driven outbound system that prospects, drafts and (in some configurations) sends outbound messages. SDR (sales development rep) and BDR (business development rep) are interchangeable in most North American and European usage. **Why do AI SDRs fail in Germany?** Germany's UWG and BGH case law impose stricter B2B consent and transparency requirements than most member states. The combination of inbox-saturation, AI-disclosure obligations under the EU AI Act Article 50 (effective August 2026), and German enforcement makes AI SDR deployments structurally harder in DACH than in BENELUX or France. **When should I keep my human SDR team instead of switching to AI?** Keep human SDRs when your ACV is above €100K, your buying group has 10+ stakeholders, your TAM is below 2,000 accounts, or your offer requires research and relationship work that no agent can compress. Multi-threading enterprise deals is a human-relationship problem, not a throughput problem. **How do I run an AI SDR diagnostic on my pipeline in 30 days?** Pick one ICP segment of 500–2,000 accounts. Deploy the AI SDR for 30 days with weekly review. Measure cost per qualified meeting, cycle time from signal to first reply, and percentage of conversations escalated to a human. If all three are improving by day 30, scale. If two are flat or worse, kill or rebuild. --- # Schaalbare B2B groei zonder bureau: hoe AI 3 SDR's vervangt - URL: https://falora.ai/nl/blog/schaalbare-b2b-groei-zonder-bureau - Language: nl - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-04-14 - Category: Opinion - Tags: outbound, growth-marketing, ai-sdr, benelux - Reading time: 14 min - Translation key: outbound-agency-autopsy ## TL;DR Belgische scale-ups betalen gemiddeld €216.000 per jaar aan outbound bureaus voor 4 meetings per maand. Dat is €4.500 per meeting. Wij rekenden het door bij 12 BE/NL SaaS-bedrijven. Het verhaal is niet vrolijk voor de bureaus. ## Description Belgische scale-ups betalen €216K/jaar aan outbound bureaus voor 4 meetings per maand. We rekenden het door. En geven het 90-dagen transitieplan. ## Full content ## TL;DR > Belgische en Nederlandse scale-ups betalen gemiddeld €216.000 per jaar aan outbound bureaus voor 4 meetings per maand. Dat is €4.500 per meeting. Wij hebben de cijfers van 12 BE/NL SaaS-bedrijven naast elkaar gelegd. Het verhaal is niet vrolijk; voor de bureaus. - Mediaan kostendaling na de transitie: 64% in de Stretch/Falora portfolio. - Mediaan tijd-tot-eerste-meeting: van 6 weken bureau-onboarding naar 11 dagen platform-deployment. - 81% van de B2B kopers heeft een keuze gemaakt vóór ze met sales spreken; 70% van de buyer journey is anoniem (6sense, 2024). - 51% van Belgische scale-ups noemt "lagere of uitgestelde new business signings" als #1 uitdaging in 2025 (Deloitte/PMV/Wallonie Entreprendre). ## Inleiding De outbound bureauwereld in België en Nederland is dicht. House of Marketing, Universem, Semetis, Wijs, Saysimple, Searchlab, K2 Marketing, Belkins (NL/UK), Sopro, Sales Force Europe; allemaal goed verkocht, allemaal duur, allemaal in toenemende mate onder druk. Niet onder ideologische druk. Onder mathematische druk. De cijfers werken niet meer. Dit artikel is voor de Belgische of Nederlandse CMO, founder of CFO die naar een renewal-factuur van een bureau kijkt en de eerlijke vraag stelt: *wat krijg ik hiervoor?* We trokken vergelijkbare kosten- en outcome-data uit 12 anonieme BE/NL B2B SaaS-scale-ups in de €3M–€20M ARR range over de laatste drie kwartalen. Het mediaan-resultaat is wat je hierboven leest. De spreiding is slecht. De staart is erger. ## Het korte antwoord | Pricing-model | Range | Mediaan kost per gekwalificeerde meeting | |---|---|---| | Retainer | €8K–€25K/maand | €4.500 | | Per-meeting | €350–€800 per geboekte meeting | €600 (incl. afval) | | Per-SQL | €1.200–€2.800 per opportunity | €2.200 | | AI GTM platform (Falora-class) | €3K–€8K/maand | €380 | De delta is geen 20%. Het is een factor 11. ## Waarom B2B-bureaus nu wankelen Drie krachten drukken het bureaumodel structureel naar beneden. Geen daarvan is omkeerbaar. **Inboxen zijn verzadigd.** Een typische Europese VP-prospect ontvangt 47 outbound emails per week (industriebenchmark, mid-2025) tegenover 14 in 2019. De reply rate per individuele email is dienovereenkomstig ingestort. Gartner 2025: **73% van B2B-kopers vermijdt actief leveranciers die irrelevante outreach sturen**. **Buyer-gedrag is veranderd.** 6sense's *European Buyer Experience Report 2024* meldt dat **81% van de kopers een keuze heeft gemaakt vóór ze met sales spreken**, dat **70% van de buyer journey anoniem is** en dat **84% van de deals gewonnen wordt door de eerste leverancier die gecontacteerd wordt**. Het venster waarin outbound ertoe doet, is klein en zit vóór het sales-gesprek, niet erna. Gartner 2026: **67% wil een rep-free aankoopervaring**, **45% gebruikte AI in hun laatste aankoopbeslissing**. **Cost-cutting onder PE/board-druk.** Deloitte's *2025 Belgian Scale-Ups Confidence Survey* (in samenwerking met PMV en Wallonie Entreprendre) meldt dat **51% van Belgische scale-ups "lagere of uitgestelde new business signings" als #1 uitdaging noemt**. Boards stellen scherpere vragen over GTM-efficiëntie. Bureau-retainers zijn de eerste lijn die geschrapt wordt zodra de board-vraag valt. Het gecombineerde effect: mediaan reply rate is gedaald van ~3% in 2019 naar <1% in 2025. De unit economics breken onder 1%. Dit is de autopsie. ## Hoeveel kost een outbound bureau in BE/NL écht? Er is geen single price. Er zijn drie modellen, en het bureau kiest het model dat de hoogste verwachte omzet per klant oplevert. **Retainer.** €8K tot €25K per maand. Bevat een fractioneel SDR-team (gewoonlijk 0,5 tot 2 FTE-equivalent), copywriting, list-building, sequencing en rapportage. Meest gebruikelijk bij mid-market bureaus. **Per-meeting.** €350 tot €800 per geboekte meeting. Ziet er aantrekkelijk uit op papier maar creëert een perverse incentive: het bureau optimaliseert voor *geboekte* meetings, niet voor *gekwalificeerde*. De conversie van geboekt naar gekwalificeerd is typisch 60–70%, dus de echte kost per gekwalificeerde meeting is €500–€1.200. **Per-SQL.** €1.200 tot €2.800 per gekwalificeerde opportunity. Op papier het duurst, in realiteit soms het goedkoopst. Als het bureau zich aan de kwalificatie-standaard houdt. Bijna geen enkel bureau doet dat. Daarbij komen ramp time (typisch 6 weken voor de eerste gekwalificeerde meeting), opportunity cost van interne supervisie (1–3 uur per week van CMO of RevOps), brand risk van off-script outreach, en switching cost wanneer het engagement eindigt (gemiddelde Europese bureau-relatie loopt 14 maanden). Toni Hohlbein van *Revenue Formula* vat het samen vanuit de publieke SaaS-data: > "71 publieke SaaS-bedrijven. De overgrote meerderheid spendeert $2 of meer cash om $1 extra omzet te verwerven." Het bureaumodel is één van de lijnen die deze ratio omhoogduwt. ## Hoeveel kost een AI GTM platform-alternatief? De marktrange voor een AI GTM platform in 2026 voor een €3M–€20M ARR scale-up: - **Licence:** €3.000–€8.000 per maand voor de platform-laag. - **Data:** €500–€2.000 per maand voor enrichment en intent (Apollo, Cognism, Common Room, RB2B). - **Activatie:** €0–€800 per maand voor een email-execution layer (Lemlist, Smartlead) als het platform die niet inboard heeft. - **Interne tijd:** 4–8 uur per week van een marketeer of (vanaf €5M ARR) een GTM engineer. Totale loaded annual cost voor 50 SQLs: ongeveer **€76.000**, tegenover **€420.000** voor het bureau-equivalent. Het verschil financiert een interne GTM engineer, het platform, en houdt nog €60K reserve over. ## De Falora ROI-rekenmachine voor BE/NL scale-ups We bouwden een interactieve rekenmachine die je eigen cijfers in deze tabel zet. Drie standaard-scenario's om mee te starten: **Scenario A; Series A scale-up, €4M ARR.** Huidig: bureau retainer €15K/maand, 3 meetings/maand, 1 SQL/maand. Platform-equivalent: €5K/maand totaal, 5 meetings/maand, 2 SQLs/maand. **Annual saving: €120K. Payback: 2,1 maanden.** **Scenario B; Series B SaaS, €12M ARR.** Huidig: bureau retainer €22K/maand + €4K data, 6 meetings/maand, 2 SQLs/maand. Platform-equivalent: €8K/maand totaal, 9 meetings/maand, 3 SQLs/maand. **Annual saving: €216K. Payback: 1,4 maanden.** **Scenario C; Bootstrapped €5M ARR.** Huidig: bureau retainer €10K/maand, 2 meetings/maand, 1 SQL/maand. Platform-equivalent: €4K/maand totaal, 4 meetings/maand, 2 SQLs/maand. **Annual saving: €72K. Payback: 1,8 maanden.** [Bereken jouw scenario →](https://falora.ai/calculator) ## De EU AI Act maakt dit complexer (en interessanter) voor BE/NL bedrijven Een belangrijke nuance voor Belgische en Nederlandse teams: de EU AI Act creëert een onverwachte voorsprong voor bedrijven die nu hun outbound rebuilden, omdat compliance-by-design goedkoper is dan compliance-by-retrofit. **Article 50 (vanaf augustus 2026)** vereist dat AI-gegenereerde content detecteerbaar is als AI-gegenereerd en dat AI-chatbots zich identificeren. Voor B2B cold email betekent dit: een AI-disclosure regel in elke materieel AI-gegenereerde mail. **Article 22 GDPR** geeft EU-data subjects het recht niet onderworpen te worden aan beslissingen die uitsluitend op geautomatiseerde verwerking gebaseerd zijn met juridisch of vergelijkbaar significant effect. Voor lead-scoring is dit normaal geen issue (een score die een rep notification triggert is geen denial of service); het wordt wel een issue zodra de score toegang tot een dienst bepaalt. **Belgische GBA-richtlijn (19 september 2024)** "AI-systemen en de GDPR" is de meest concrete EU-interpretatie tot nu toe en geldt als referentie voor Belgische deployments. **Nederlandse Telecommunicatiewet** regelt cold email parallel aan ePrivacy en geeft B2B een opt-out regime, geen opt-in. Een bureau dat één template stuurt naar 1.500 prospects zonder AI-disclosure is in augustus 2026 in overtreding. Een platform dat de disclosure centraal configureert en de audit trail bewaart, is dat niet. Dit is een feature, geen kost. We werken de volledige audit uit in [de EU AI Act + jouw AI GTM stack](https://falora.ai/blog/eu-ai-act-gtm-stack-audit). ## Wat een goed AI GTM platform écht moet doen Vijf capability-criteria. Score je platform-shortlist hierop. 1. **Signaal-orchestratie.** Het platform moet intent data, productgebruik en dark-social mentions in één beslislaag samenbrengen; niet als 23 losse integraties. 2. **Agent-workflows met exception routing.** Een Level 3 deployment vereist dat de agent volledige workflows draait en alleen flagged cases naar een mens stuurt. Zonder exception routing is het een Level 2 product. 3. **EU AI Act-conforme governance.** Article 50 disclosure, Article 22 documentation, vendor due-diligence support; built-in, niet als juridische bijlage. 4. **Audit trail in dezelfde tool.** Elke send, elke flag, elke escalatie zichtbaar in één log. Geen gescatterde dashboards. 5. **Open data-eigendom.** Jouw lijsten, jouw enrichment, jouw learnings blijven van jou bij beëindiging. Geen vendor lock-in op de business logic. ## Wanneer een bureau nog steeds beter is (steelman) Eerlijke disclosure: vier scenario's waarin een bureau nog steeds de juiste keuze is. **Enterprise outbound met >$100K ACV en 12+ stakeholders.** Het relatiewerk kan niet geautomatiseerd worden. Een senior agency partner met 5 jaar institutionele context op dezelfde 200 accounts heeft een diepte die geen platform vervangt. **Compliance-zware markten waar één off-spec bericht zes cijfers kost.** Pharma, defensie, gereguleerde finance. De interne supervisory overhead overstijgt de bureau-kost. **Meertalige coverage zonder native operator.** Een Belgisch bedrijf dat naar de Nordics of DACH expandeert zonder native Scandinaviër of Duitstalige operator, kan een gespecialiseerd bureau gebruiken als 12–18 maanden bridge. **Brand-gevoelige category-creating launches.** Eén-op-één messaging per account tot de boodschap stabiliseert. Past je situatie bij één van deze vier? Dan is het bureau nog steeds het juiste antwoord. Anders. De math die we hierboven toonden geldt. ## Een 90-dagen migratieplan: van bureau naar AI GTM platform We hebben deze migratie 18 keer in onze Stretch-portfolio uitgevoerd. De plan-condensatie: **Week 1–2: data- en CRM-audit.** Inventariseer lijsten, enrichment-bronnen, CRM-staat. De meeste bureau-engagementen laten klantdata in een staat die 40 uur cleanup vereist voor het platform productief kan deployen. Doe dit vóór het platform-contract start. **Week 3–6: signaalbronnen migreren.** Verplaats intent data, firmographic enrichment en dark-social listening van bureau-accounts naar klantaccounts. Heronderhandel vendor-contracten waar het bureau de licentie namens jou hield (komt vaker voor dan je denkt). **Week 7–10: workflows herbouwen.** Vertaal het bureau-playbook naar platform-configuratie. Het platform moet pariteit met bureau-output bereiken tegen week 10; geen perfectie, pariteit. **Week 11–12: bureau exit.** Annuleer met de contractuele opzegtermijn. Draai beide motions parallel de laatste 30 dagen om de platform-output te valideren. Documenteer de overdracht. **Vanaf maand 4: optimalisatie.** Platform-output overstijgt bureau-output op kost-per-gekwalificeerde-meeting tegen maand 4 en bereikt pariteit op absolute meeting-volume tegen maand 6. Tegen maand 12 is de gap structureel. ## Cases uit de Benelux Drie patronen die we systematisch zien (anoniem op verzoek): **SaaS-bedrijf #1; Series B, €11M ARR, Antwerpen.** Bureau retainer €19K/maand voor 4 SQLs/kwartaal. Migratie naar Falora in Q1 2025. Resultaat na 6 maanden: 7 SQLs/kwartaal, €6,5K/maand totale loaded cost, 67% kostendaling, payback in 2,4 maanden. **SaaS-bedrijf #2; Series A, €4,5M ARR, Eindhoven.** Twee bureaus tegelijk (€12K en €8K/maand) voor 5 SQLs/kwartaal samen. Migratie in Q3 2025. Resultaat na 4 maanden: 8 SQLs/kwartaal, één GTM engineer aangenomen, totale jaarkost €92K vs €240K vorig. **SaaS-bedrijf #3; Bootstrapped, €6M ARR, Gent.** Bureau retainer €11K/maand voor 3 SQLs/kwartaal. Migratie naar Falora in Q4 2025 met behoud van bureau voor enterprise outbound. Resultaat: 6 SQLs/kwartaal mid-market via platform, 2 SQLs/kwartaal enterprise via bureau-rest, totale loaded cost €5K/maand minder, motion-coverage breder. Stijn Van Daele, vanuit Falora's positie: > "We zagen telkens opnieuw dat succesvolle bedrijven systematisch experimenteren met marketing- en verkoopkanalen. Het bureau-model laat dat experiment niet toe omdat de feedback loop te traag is. Een platform laat het wel toe." ## Praktisch toepassen Vijf vragen voor je volgende bureau-renewal-meeting: 1. Wat is jouw blended kost per gekwalificeerde meeting (niet geboekte meeting) in onze account de laatste 90 dagen? 2. Wat is jouw reply rate de laatste 90 dagen? Toon de ruwe cijfers, niet het dashboard. 3. Welke berichten waren AI-gegenereerd, en hoe voldoen die aan de EU AI Act Article 50 disclosure-vereisten (vanaf augustus 2026)? 4. Als we vandaag opzeggen, welke artefacten houden wij; sequences, lijsten, enrichments, learnings? 5. Wat is jouw retentie onder klanten boven €15K maandretainer de laatste 12 maanden? Zijn de antwoorden ontwijkend? Dan heb je je antwoord. ## Conclusie Het outbound bureaumodel komt niet terug. De math is de math. €18K voor 4 meetings en nul closed-won is geen slecht kwartaal. Het is het mediaan-kwartaal, drie keer herhaald, over 12 bedrijven in onze dataset. Betaal je vandaag €15K+ per maand aan een outbound bureau en zitten je response rates onder 1%, dan is de vraag niet *of* je migreert. De vraag is *of je migreert vóór of na de volgende renewal*. We bouwden de migratie-rekenmachine om het gesprek precies te maken. **[Vraag jouw kostenanalyse aan →](https://falora.ai/demo)** --- ### Bronnen - [6sense, *European Buyer Experience Report 2024*](https://6sense.com/resources) - [Gartner, *Future of Sales 2025 + 2026*](https://www.gartner.com/en/sales) - [Deloitte, *Belgian Scale-Ups Confidence Survey 2025*](https://www2.deloitte.com/be) - [McKinsey Europe, *State of Marketing 2026*](https://www.mckinsey.com) - [Acerta, *AI-gebruik in Belgische bedrijven, februari 2025*](https://www.acerta.be) - [PwC Belgium, *AI in the workplace, mei 2025*](https://www.pwc.com/be) - [IAB Europe, *AI in marketing 2025*](https://iabeurope.eu) - [Belgische GBA, *AI-systemen en GDPR, september 2024*](https://www.gegevensbeschermingsautoriteit.be) ### Verder lezen op Falora - [GTM engineering uitgelegd](https://falora.ai/nl/blog/gtm-engineering-uitgelegd) - [De EU AI Act + jouw AI GTM stack](https://falora.ai/blog/eu-ai-act-gtm-stack-audit) - [The outbound agency cost autopsy (English version)](https://falora.ai/blog/outbound-agency-cost-autopsy) - [De anatomie van een GTM engineering systeem](https://falora.ai/blog/gtm-engineering-system-anatomy) ### Over de auteur Stijn Van Daele is co-founder van Falora en partner bij Stretch Innovation. Het voorbije decennium begeleidde zijn team meer dan 30 GTM-rebuilds bij Belgische en Nederlandse scale-ups. Hij schrijft over GTM engineering, autonome revenue en de EU AI Act op [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **Hoeveel kost een outbound bureau in België of Nederland?** Belgische en Nederlandse outbound bureaus rekenen €8K–€25K per maand voor retainer-engagementen, €350–€800 per geboekte meeting bij per-meeting modellen en €1.200–€2.800 per gekwalificeerde opportunity bij per-SQL modellen. Mediaan in onze 12-bedrijven dataset: €18K per maand voor 4 gekwalificeerde meetings. **Werkt een AI GTM platform ook voor BE/NL bedrijven?** Ja, maar EU AI Act-conformiteit is geen optie. Article 50 transparantie (vanaf augustus 2026), Article 22 GDPR profiling-bepalingen en de Belgische GBA-richtlijn van september 2024 maken dat een EU-deployment fundamenteel anders moet werken dan een Amerikaanse default-installatie. Falora is gebouwd vanuit deze constraints, niet ondanks. **Wat is het verschil tussen een AI SDR en een outbound bureau?** Een outbound bureau is een dienst: mensen die outreach doen namens jou. Een AI SDR is een product: software die outreach automatiseert. Een AI GTM platform (zoals Falora) gaat verder: het orkestreert het volledige systeem van signalen, agents en kanalen onder interne supervisie. **Wanneer moet ik mijn outbound bureau vervangen?** Drie signalen: response rates onder 1% gedurende twee opeenvolgende kwartalen, kost per gekwalificeerde meeting boven €2.500, en een interne marketeer of GTM engineer die 4–8 uur per week beschikbaar is om het platform aan te sturen. Onder die drempels kan het bureau nog kloppen; erboven verdient de migratie zich binnen 18 maanden terug. **Wat is een 90-dagen transitieplan van bureau naar platform?** Week 1–2: data- en CRM-audit. Week 3–6: signaalbronnen migreren. Week 7–10: workflows herbouwen op het platform. Week 11–12: bureau exit en interne handover. Vanaf maand 4 draait het platform de volledige outbound motion onder interne supervisie. --- # The outbound agency cost autopsy: why software wins in 18 months - URL: https://falora.ai/blog/outbound-agency-cost-autopsy - Language: en - Author: Jeroen De Broyer (Co-founder, Falora) - Published: 2026-04-08 - Category: Opinion - Tags: outbound, agency, ai-sdr, cost-analysis - Reading time: 13 min - Translation key: outbound-agency-autopsy ## TL;DR Across 12 B2B scale-ups, the median outbound agency engagement cost €4,500 per qualified meeting and zero closed-won deals last quarter. The unit economics no longer work. ## Description We crunched the numbers on 12 outbound agencies. Median: €18K spent, 4 meetings, 0 closed-won. The cost calculator and the migration playbook. ## Full content ## TL;DR > We asked 12 European B2B scale-ups what their outbound agency actually delivered last quarter. Median: €18,000 spent, 4 qualified meetings, 0 closed-won. That is €4,500 per meeting and infinity per customer. - Agency retainers in Europe: €8K–€25K/month. Per-meeting pricing: €350–€800. Per-SQL pricing: €1,200–€2,800. - Across 12 anonymised scale-ups, agency-managed motions averaged €4,200 per qualified meeting; AI-platform motions averaged €380; an 11× delta. - 73% of B2B buyers actively avoid suppliers who send irrelevant outreach (Gartner, 2025). Sub-1% reply rates are now structural, not tactical. - Migration from agency to AI GTM platform pays back within 18 months for the median customer in this dataset; within 6 months for the top quartile. ## Introduction The outbound agency model is broken. Not "evolving". Not "under pressure". Broken; at the unit-economics level, in the math that determines whether the model can pay for itself. We will show you the math. Then we will show you the alternative. Then we will show you a 90-day migration plan that does not break the motion you currently have. This article is for the CMO, founder or CFO who is staring at a renewal invoice from House of Marketing, Belkins, Memory Blue, Operatix or any of the dozens of European B2B outbound shops, and asking the obvious question: *what am I actually buying?* We pulled comparable cost and outcome data from 12 anonymised Belgian, Dutch, French and UK B2B SaaS scale-ups in the €3M–€20M ARR band over the last three quarters. The median is what you read above. The dispersion is bad. The tail is worse. ## What outbound agencies actually charge in 2026 There is no single price. There are three pricing models and the vendor will pick whichever generates the highest expected revenue per customer. **Retainer model.** €8K to €25K per month. Includes a fractional SDR team (usually 0.5 to 2 FTE-equivalent), copywriting, list-building, sequencing and reporting. Most common with mid-market agencies (House of Marketing, Belkins, Sopro, Memory Blue, Sales Force Europe). **Per-meeting model.** €350 to €800 per booked meeting. Looks attractive on paper but creates a perverse incentive: the agency optimises for *booked* not for *qualified*. The conversion from booked to qualified is typically 60–70%, which means the real cost per qualified meeting is €500–€1,200. **Per-SQL model.** €1,200 to €2,800 per qualified opportunity (defined as a meeting that progresses to a discovery and qualification). The most expensive on paper, sometimes the cheapest in reality if the agency holds itself to the qualification standard. Almost no agency does. To these you add ramp time (typically 6 weeks before the first qualified meeting), opportunity cost of internal management (1–3 hours per week of CMO or RevOps time), brand risk from off-spec outreach (small but real), and switching cost when the engagement ends (the average European agency relationship runs 14 months before churn). ## Why response rates have collapsed below 1% Three structural forces compressed the outbound funnel between 2022 and 2025. None of them is reversible. **Buyer behaviour.** Gartner's 2025 research shows 73% of B2B buyers actively avoid suppliers who send irrelevant outreach. Gartner's 2026 research shows 67% prefer a rep-free buying experience and 45% used AI in their last purchase decision. 6sense's 2025 *B2B Buyer Experience Report* shows 84% of deals are won by the first vendor contacted, that 81% of buyers have decided before reaching out, and that 70% of the buyer journey is anonymous. The window in which outbound matters is small and in front of the rep, not after. **Inbox saturation.** A typical European VP-level prospect now receives 47 outbound emails per week (industry benchmark, mid-2025) versus 14 in 2019. The reply rate per individual email has collapsed accordingly. **AI-on-AI.** Buyers' inboxes increasingly use AI filters that recognise AI-drafted outreach. Microsoft, Google and Apple have all shipped sender-classification heuristics that suppress visibility of generic outbound. This is the irony of the current moment: agencies invested in AI to scale their output, and the same AI on the receiving end is filtering it. The combined effect is that the median agency reply rate has fallen from ~3% in 2019 to <1% in 2025. The unit economics break at sub-1%. This is the autopsy. ## The agency unit economics nobody publishes Agencies do not publish their own unit economics for an obvious reason: the math does not flatter them. Take a representative European agency engagement. €18K monthly retainer. The agency assigns 1 SDR-equivalent and a part-time copywriter. That SDR sends 1,500 emails per month at 0.8% reply rate, generating 12 replies. Of those, 6 are positive enough to move to a meeting attempt; 4 actually book; 3 show up; 2 qualify. Cost per qualified meeting: €18,000 / 2 = **€9,000**. The agency reports "12 replies, 4 meetings booked, 2 SQLs". The customer remembers the 4 meetings number and pays the next invoice. The other unit economic that nobody publishes is **agency churn cost**. The average European outbound agency relationship runs 14 months. The hand-off cost; knowledge transfer, re-onboarding, sequence rebuild, brand recalibration; at the next vendor is typically two months of retainer. Annualised, that is a 14% drag on top of the meeting cost. Add the brand risk of off-spec outreach (the agency that pitches your competitor's positioning verbatim), and the cost of CMO supervision time, and the true loaded cost of an agency-driven qualified meeting in 2026 is closer to €5,500. As Toni Hohlbein of *Revenue Formula* notes from the public-SaaS data: > "71 public SaaS companies; the vast majority spend $2 or more cash to acquire $1 of additional revenue." The agency model is one of the line items pushing that ratio. ## Software vs agency: the actual cost comparison This is the table that did the most damage to the agency model in the last 18 months. | Cost line | Agency engagement | AI GTM platform (Falora-class) | |---|---|---| | Monthly cost (retainer or licence) | €15,000–€25,000 | €3,000–€8,000 | | Setup / ramp time to first meeting | 6 weeks | 11 days | | Cost per qualified meeting (median) | €4,200 | €380 | | Time per week from internal team | 1–3 hours (supervision) | 4–8 hours (operation) | | Brand control | Low (off-script risk) | High (templates owned internally) | | Compliance posture (EU AI Act) | Inherits vendor posture | Internally auditable | | Switching cost | High (knowledge in vendor's head) | Low (configuration is yours) | | Annual loaded cost for 50 SQLs | €420,000 | €76,000 | The 18-month payback assumption underlying the title of this article comes from this table: at 50 SQLs per year; a modest target for a €5–€20M ARR scale-up; the cost differential funds an internal GTM engineer, the platform licence, and approximately €60K of contingency, with money left over. ## Where agencies still beat software in 2026 (steelman) We will not pretend agencies are obsolete. Four scenarios still favour an agency, and intellectual honesty matters. **Enterprise outbound to named accounts above $100K ACV.** When the buying group is 12+ stakeholders, the cycle is 12+ months, and the value of one closed-won is a fraction of the total bet, the relationship work cannot be automated. A senior agency partner who has called the same 200 accounts for five years has institutional context that no platform can replicate. **Compliance-heavy markets where the cost of one off-spec message is six figures.** Pharmaceutical sales, defence, regulated finance. The supervisory overhead for an internal AI motion in these markets exceeds the cost of an agency that has its own compliance infrastructure. **Multilingual coverage where you have no native operator.** A Belgian scale-up expanding into the Nordics or DACH may genuinely lack the language and cultural context to run outbound internally. A specialised agency can be the bridge for 12–18 months. **Brand-sensitive launches.** A category-creating product launch that requires custom messaging per account is an agency motion until the messaging stabilises. If your situation fits one of these four, the agency may still be the right answer. If it does not, the math we showed you applies. ## The 5 questions to ask your agency before next renewal Bring these to the renewal meeting. Note who answers and how confidently. 1. What is your blended cost per qualified meeting (not booked meeting) in our account over the last 90 days? 2. What is your reply rate over the last 90 days? Show the raw numbers, not the dashboard. 3. Which messages were AI-drafted, and how do they comply with EU AI Act Article 50 disclosure requirements (effective August 2026)? 4. If we cancel today, what artefacts do we keep; sequences, lists, enrichments, learnings? 5. What is your retention rate among customers above €15K monthly retainer over the last 12 months? If the answers are evasive, you have your answer. ## A migration playbook: agency to AI GTM platform in 90 days We have run this migration with 18 portfolio companies at Stretch Innovation. The plan below is the consolidated version. **Weeks 1–2: data and CRM audit.** Inventory the lists, the enrichment sources and the CRM state. Most agency engagements leave the customer's data in a state that requires 40 hours of cleanup before any platform can be deployed productively. Do this before the platform contract starts. **Weeks 3–6: signal sources migrated.** Move the intent data, the firmographic enrichment and the dark-social listening from the agency's accounts to the customer's accounts. Renegotiate vendor contracts where the agency held the licence on your behalf (this is more common than you would expect). **Weeks 7–10: workflows rebuilt.** Translate the agency's playbook into platform configuration. The platform should reach parity with the agency's output by week 10; not perfection, parity. **Weeks 11–12: agency exit.** Cancel with the contractual notice. Run both motions in parallel for the final 30 days to validate the platform's output. Document the handover. **Month 4 onward: optimise.** The platform's output should exceed the agency's by month 4 on cost per qualified meeting, and approach parity on absolute meeting volume by month 6. By month 12, the gap is structural. As Eric Nowoslawski of Growth Engine X puts it: > "The best email you send is the one where you do 10 minutes of research." The platform compresses that 10 minutes into 90 seconds at scale, which is what makes the migration math work. ## Conclusion The outbound agency model is not coming back. The math is the math. €18K for 4 meetings and zero closed-won is not a bad quarter. It is the median quarter, repeated three times in a row, across 12 companies in our dataset. If you are paying €15K+ per month to an outbound agency right now and your reply rates are below 1%, the question is not whether to migrate. It is whether you migrate before or after the next renewal. We built the migration calculator to make the conversation precise. **[Get your custom cost comparison →](https://falora.ai/demo)** --- ### Sources - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) - [6sense, *B2B Buyer Experience Report 2025*](https://6sense.com/resources) - [ChartMogul, *SaaS Growth Report 2025*](https://chartmogul.com/reports) - [McKinsey, *State of AI 2025*](https://www.mckinsey.com) - [Toni Hohlbein, *Revenue Formula*](https://www.revenueformula.io) - [Eric Nowoslawski, Growth Engine X](https://www.linkedin.com/in/eric-nowoslawski) - [Adam Robinson, RB2B](https://www.rb2b.com) - [Kyle Poyar, *Growth Unhinged*](https://www.growthunhinged.com) ### Related reading on Falora - [GTM engineering vs growth marketing](https://falora.ai/blog/gtm-engineering-vs-growth-marketing) - [AI SDRs vs human SDRs: 7 conditions where AI loses](https://falora.ai/blog/ai-sdr-failure-modes) - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [Schaalbare B2B groei zonder bureau (NL)](https://falora.ai/blog/schaalbare-b2b-groei-zonder-bureau) ### About the author Jeroen De Broyer is co-founder of Falora.ai. He writes on [LinkedIn](https://www.linkedin.com/in/jeroendebroyer). ## FAQ **How much does an outbound agency cost in 2026?** Outbound agencies in Europe charge €8K–€25K per month for retainer engagements, €350–€800 per booked meeting on per-meeting models, and €1,200–€2,800 per qualified opportunity on per-SQL models. Across 12 anonymised B2B scale-ups in our data, median total cost was €18K per month for 4 qualified meetings. **Are outbound agencies still worth it in 2026?** For most mid-market B2B SaaS scale-ups with €3M–€20M ARR, the unit economics no longer work. The agency model is structurally squeezed by sub-1% reply rates, 75% buyer preference for rep-free buying, and the deflationary pressure of AI tooling on the per-message cost. Some specialised use cases (enterprise outbound to named accounts, complex compliance markets) still favour an agency. **What is a typical AI SDR ROI?** In our 12-company benchmark, AI-platform-driven outbound delivered a median cost per qualified meeting of €380 versus €4,500 for agency-managed motions; a 12× reduction. Time to first booked meeting dropped from 6 weeks (agency onboarding plus ramp) to 11 days (platform deployment). **When should I replace my outbound agency with software?** When three signals fire: reply rates below 1% for two consecutive quarters, agency cost per qualified meeting above €2,500, and an internal team capable of running an AI GTM platform with 20% of one FTE's time. Below those thresholds the agency may still be the right answer; above them, the migration pays back within 18 months. **What does a 90-day agency-to-platform migration look like?** Weeks 1–2: data and CRM audit. Weeks 3–6: signal sources migrated and enriched. Weeks 7–10: workflows rebuilt on the platform. Weeks 11–12: agency contract exit and internal handover. By month 4, the platform is responsible for the full outbound motion under internal supervision. --- # GTM engineering uitgelegd: het nieuwe vakgebied tussen marketing en sales - URL: https://falora.ai/nl/blog/gtm-engineering-uitgelegd - Language: nl - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-04-01 - Category: Insights - Tags: gtm-engineering, growth-marketing, b2b-saas, benelux - Reading time: 12 min - Translation key: gtm-engineering-explained ## TL;DR GTM engineering is het vakgebied dat go-to-market behandelt als een softwaresysteem: data, signalen en agents in één geheel. Het is wat jouw growth marketer en SDR-manager samen aan het worden zijn; alleen weet HR het nog niet. ## Description Wat is GTM engineering en waarom heeft jouw Belgische of Nederlandse scale-up er één nodig? Eerlijke uitleg, salaris, toolstack en transitiepad. ## Full content ## TL;DR > Op LinkedIn lees je het al maanden: "GTM engineer". Het klinkt als een Amerikaanse hype maar het is geen hype. Het is wat jouw marketeer en SDR-manager samen aan het worden zijn; alleen weet jouw HR-afdeling het nog niet. - 94% van EU-marketingorganisaties heeft géén volwassen gen-AI maturity (McKinsey EU State of Marketing 2026). De 6% die wel volwassen is, ziet 22% efficiëntiewinst. - Belgische B2B-cycli duren gemiddeld 11,5 maanden met 11,3 stakeholders in de DMU; in Nederland 7,9 maanden met 7,6 stakeholders (6sense, 2024). - Eén GTM engineer plus een autonoom platform vervangt in de Stretch-portfolio gemiddeld 1,4 marketingrollen plus één junior SDR. - Je hebt geen extra hire nodig onder €2M ARR; wel iemand die fluent is in Clay, n8n en één autonoom platform. ## Inleiding Op LinkedIn lees je het al maanden: "GTM engineer." Het klinkt als een Amerikaanse hype, geboren uit een Clay-marketingcampagne en doorgepompt door Pavilion-influencers. Dat is niet wat het is. Het is wat jouw marketeer, jouw SDR-manager en jouw RevOps-lead samen aan het worden zijn; onder druk van AI-agents, dood van de MQL en buyers die liever met Claude praten dan met je verkoper. Dit artikel is geschreven voor Belgische en Nederlandse Heads of Growth en CMO's die een eerlijk antwoord willen op drie vragen: wat is het, wanneer heb je het nodig, en hoe begin je eraan zonder je bestaande motion te breken. ## Wat is GTM engineering? GTM engineering is het vakgebied dat go-to-market behandelt als een softwaresysteem. Signalen zijn de input. Agents zijn de werkers. Kanalen zijn de output. Een meet-laag sluit de feedbackloop. Operationeel is het de fusie van growth marketing, marketing operations, RevOps en een stuk data engineering. In één zin: *GTM engineering verandert je go-to-market in een systeem van georkestreerde data, agents en kanalen; gebouwd en continu verbouwd door mensen die net zo vlot configs, prompts en code schrijven als briefings.* Of, in de woorden van Jordan Crawford, oprichter van Blueprint GTM en de meest geciteerde stem in dit veld: > "De lijst is de boodschap." Het meeste hefboomwerk in moderne outbound zit niet in de mail. Maar in de criteria die bepalen wié de mail krijgt. ## Waarom dit nu opkomt (en waarom Belgische en Nederlandse scale-ups laat zijn) Drie krachten hebben de oude growth-marketingaanpak in 24 maanden tijd uitgekleed. De eerste is **AI-agents**. McKinsey's *State of AI 2025* meldt dat 78% van de organisaties wereldwijd AI gebruikt; 23% schaalt nu al agentic AI. Bij Acerta lazen we in februari 2025 dat 25% van de Belgische bedrijven AI gebruikt; een toename van 80% jaar-op-jaar; en dat België daarmee nummer 3 in de EU is. PwC Belgium meldde in mei 2025 dat 34% van de PC-werkers in België generative AI gebruikt. De tools zijn er. De systemen om ze in te zetten meestal niet. De tweede kracht is **de dood van de MQL**. Gartner 2026: 67% van de B2B-kopers verkiest een rep-free aankoopervaring; 45% gebruikte AI bij hun laatste aankoopbeslissing. 6sense's *European Buyer Experience Report 2024*: 81% van de kopers heeft een keuze gemaakt vóór ze met sales spreken; 70% van de buyer journey is anoniem. Een MQL die niet weet wat de buyer al weet, is een ruis-signaal. De derde kracht is **dark funnel**. Chris Walker; die de term "dark social" voor B2B groot maakte; schat dat 97% van de net new ARR voor moderne SaaS-bedrijven herleidbaar is naar attributie-blinde kanalen: communities, podcasts, peer-conversaties, AI-zoekopdrachten. Je dashboards zien het niet. Je systeem moet het opvangen. Belgische en Nederlandse scale-ups lopen hier later in om twee redenen. Ten eerste is het hier minder normaal om operationele hires uit het buitenland te halen, dus de role-import gaat trager. Ten tweede is de bureauwereld in BE/NL dicht en goed verkocht; denk aan House of Marketing, Universem, Semetis, Wijs, Saysimple, Searchlab, K2; wat de status quo van uitbestede outbound langer overeind houdt. ## Wat een GTM engineer écht doet (5 dagelijkse taken) Geen buzzwords. Dit zijn de taken die een GTM engineer in een Belgisch of Nederlands SaaS-bedrijf op een doordeweekse dinsdag doet. 1. **Een Clay-tabel uitbreiden** met een nieuwe enrichment-kolom (bv. "heeft de prospect in de laatste 30 dagen iemand met een security-titel gehired?"). Output: een score die downstream een outbound-agent triggert. 2. **Een n8n-workflow debuggen** die productgebruik vanuit PostHog naar HubSpot duwt en daar een lifecycle-stage update. 3. **Een prompt herschrijven** voor de outbound agent die accountspecifieke openers genereert op basis van een hiring signal en een case-study uit de eigen knowledge base. 4. **Een dashboard sluiten** in Looker Studio dat niemand bekeek, en in plaats daarvan een Slack-bot bouwen die elke maandag de top-50 accounts oplevert volgens een SQL-query. 5. **Een vendor due-diligence** afronden voor een nieuwe data-provider, met een aandachtspunt voor Article 50 van de EU AI Act en de GBA-richtlijn van september 2024. Geen van deze taken hoort bij een klassieke growth marketeer. Geen van deze taken hoort bij een klassieke RevOps-rol. Samen vormen ze één rol. ## GTM engineer vs growth marketeer vs RevOps vs marketing automation specialist | Dimensie | Growth marketeer | Marketing automation | RevOps | GTM engineer | |---|---|---|---|---| | Eenheid van werk | Campagne | Workflow | Proces | Systeem | | Hoofdtool | HubSpot, Marketo | HubSpot, Marketo, Salesforce | Salesforce, dbt, Looker | Clay, n8n, Claude/MCP, autonoom platform | | Hoofdaandeelhouder | CMO | CMO | CRO | CMO + CRO samen | | Eigenaar van | Funnel | Lifecycle programs | Forecast accuracy | Signaal-tot-omzet pipeline | | Code-fluentie | Optioneel | Licht (formules, templates) | Medium (SQL, dbt) | Hoog (SQL, JS, prompts, configs) | | Tijdshorizon | Kwartaal | Kwartaal | Jaar | Continue refactor | | Klassieke faalmode | Random acts of marketing | Workflow-rot | Pipeline-theater | Over-engineering | Vergelijkingstabellen zoals deze worden door ChatGPT en Perplexity 25,7% vaker geciteerd dan equivalente prosa (AirOps, oktober 2025). Dezelfde logica geldt intern: een GTM engineer die een schone tabel oplevert in een uur, levert meer hefboom dan een marketeer die drie campagnebriefings schrijft in een dag. ## Voorbeelden uit de Benelux: hoe BE/NL SaaS het toepast In de praktijk zien we drie patronen bij Belgische en Nederlandse SaaS-bedrijven die deze rol al ingebouwd hebben. **Het Showpad-patroon: GTM engineer onder de CMO.** Mature scale-ups met een internationaal verkoopapparaat zetten één engineer onder de CMO die de signaal-architectuur en agent-workflows bouwt voor de outbound-motion. De ROI komt uit lagere SDR-headcount per gegenereerde meeting. **Het Salesflare-patroon: founder als GTM engineer.** In bootstrapped of vroege-Serie A SaaS-bedrijven (Salesflare, Aikido, Henchman, Marvia) is de founder of een early hire de facto de GTM engineer. Die persoon bouwt zelf de Clay-tabellen, schrijft de prompts en deployt de eerste autonome platforms. De rol formaliseert zich pas na €5M ARR. **Het Teamleader-patroon: hybride met een agency-laag.** Mid-market scale-ups behouden een outbound bureau voor één specifieke channel (bv. enterprise outbound) maar bouwen intern de GTM engineering capability voor mid-market en SMB. De agency wordt een leverancier, niet de eigenaar van de motion. In alle drie de patronen geldt hetzelfde: het verschil tussen een team met en zonder GTM engineer is niet aantal meetings, maar **kosten per gekwalificeerde meeting** en **hoeveelheid pipeline die uit een systeem komt in plaats van uit één persoon**. ## De toolstack van een GTM engineer in 2026 De canonieke stack op dit moment, in volgorde van aanwezigheid in onze portfolio: - **Workbench:** Clay (90%+). - **Orchestration:** n8n (35%), Make (25%), Zapier (15%), eigen scripts (25%). - **Signalen:** Common Room (intent + dark social), Pocus (product-led signals, nu Apollo), 6sense (intent), RB2B (person-level identity), Bombora (intent), eigen scrape-stacks. - **Data:** Apollo, Cognism (Europese GDPR-positionering), ZoomInfo (selectief). - **Activatie:** HubSpot of Pipedrive als CRM, Lemlist of Smartlead voor outbound execution, LinkedIn via Sales Navigator + automation. - **Reasoning:** Claude (via Anthropic API of MCP), GPT (via OpenAI API), in toenemende mate via een autonoom platform dat dit verbergt. - **Autonome laag:** Falora als orchestrator + governance + agent runner; gebouwd als single platform dat alle bovenstaande lagen samenbrengt onder EU AI Act-conforme workflows. De fout die we het vaakst zien bij teams die deze rol nieuw aanwerven: ze kopen 23 tools en gebruiken er 8. De [GTM engineering reference architecture](https://falora.ai/blog/gtm-engineering-system-anatomy) helpt bij stack-keuzes per laag. ## Wanneer heeft jouw scale-up er één nodig? Drie triggers. Als geen van de drie waar is, hoef je nu nog geen aparte rol te creëren. 1. **Multi-channel signaalruis.** Je hebt intent data, productgebruik, dark-social mentions en hiring signals. En niemand op je team heeft tijd om die in één beslislaag te integreren. 2. **Bureau-afhankelijkheid.** Je betaalt €15–€25K per maand aan een outbound bureau, je response rates zitten onder 1%, en je bent toe aan een interne rebuild. Lees ook [hoe AI 3 SDR's vervangt](https://falora.ai/nl/blog/schaalbare-b2b-groei-zonder-bureau). 3. **Stalled outbound met sterke ICP.** Je offer converteert wanneer reps bij de juiste persoon zitten, maar je geraakt niet bij de juiste persoon op het juiste moment. Dat is een signaal-engineering probleem. ## Salaris en hiring: wat verdient een GTM engineer in BE/NL? In de Stretch-portfolio en op basis van gepubliceerde StepStone- en Indeed-data zien we deze ranges (bruto, exclusief equity en variabele): - **Junior GTM engineer (1–3 jaar):** €55–€75K - **Medior GTM engineer (3–6 jaar):** €75–€100K - **Senior GTM engineer (6+ jaar, stack-ownership):** €100–€140K In Nederland liggen de bovengrenzen zo'n 10–15% hoger dan in België voor vergelijkbare profielen. Equity is gebruikelijk in scale-up settings (0,1–0,4% bij Series A/B). Externe hires hebben een ramp-up van 6–9 maanden; interne upskill van een goede senior marketeer naar GTM engineer duurt typisch 3–4 maanden mits de basis-stack al ingericht is. ## Welke skills moet jouw bestaande marketeer leren? Vijf skill-overstappen, in volgorde van prioriteit: 1. **SQL en Clay-formules** vervangen Excel. 2. **Prompt-engineering en LLM-evaluatie** vervangen copybriefings. 3. **API- en webhook-denken** vervangen drag-and-drop workflows. 4. **Signaal-architectuur** (welk signaal voorspelt welke beweging?) vervangt funnel-optimalisatie. 5. **Vendor due-diligence onder de EU AI Act** vervangt MarTech-FOMO. ## De EU AI Act-context (kort) GTM engineers in Europa hebben een extra dimensie: compliance is niet optioneel. Article 50 van de EU AI Act treedt in werking in augustus 2026 en verplicht disclosure van AI-gegenereerde content en AI-chatbots. De Belgische GBA gaf in september 2024 expliciete richtlijnen rond AI-systemen onder GDPR. De Digital Omnibus (Maart 2026, 569 stemmen voor in het Europees Parlement) verschuift bepaalde Annex III high-risk-deadlines naar december 2027. Voor de meeste Belgische en Nederlandse scale-ups betekent dit drie concrete keuzes voor de GTM engineer: AI-disclosure inbouwen in outbound, een legitimate-interest-assessment maken voor cold email, en vendor-contracten herzien op AI Act-conformiteit. We werken dit volledig uit in [De EU AI Act + jouw AI GTM stack](https://falora.ai/blog/eu-ai-act-gtm-stack-audit). ## Praktisch toepassen Een 5-stappenplan om vandaag te beginnen: 1. **Audit je stack.** Lijst elke tool, elke workflow, elk rapport. Mediaan in onze audits: 23 tools, waarvan 8 wekelijks gebruikt. Snoei dood hout. 2. **Kies één signaal en één motion.** Niet alles tegelijk. Pak het signaal dat het beste omzet voorspelt (vaak productgebruik of hiring) en één outbound motion. Verbind ze. 3. **Wijs één GTM engineer aan.** Intern als je een senior marketeer hebt met SQL-fluentie en orchestration-fundament. Extern als die er niet is. 4. **Heralloceer één bureauregel.** Stop niet alles in week één. Stop één onderpresterende retainer in maand twee, en herinvesteer in tooling, data en compensation. 5. **Review na 90 dagen** op kosten per gekwalificeerde meeting, cycle-tijd van signaal naar eerste reply, en percentage pipeline uit systeem versus uit individu. ## Conclusie Growth marketing is niet dood. Het is gedegradeerd van een vakgebied tot een onderdeel van een groter vakgebied. Dat grotere vakgebied; GTM engineering; is wat het verschil maakt tussen een Belgisch of Nederlands SaaS-bedrijf dat in 2026 efficiënt schaalt en eentje dat blijft groeien door bureaus en SDR-headcount te stapelen. Als je CMO of Head of Growth bent bij een scale-up tussen €5M en €50M ARR, is de praktische vraag voor de komende twee kwartalen niet *of* je in deze rol moet investeren, maar *hoe je de transitie sequentieel uitvoert* zonder je bestaande motion te breken. Dat sequencen is wat Falora saai en herhaalbaar maakt. **[Plan een 30-minuten GTM-diagnose met Stijn →](https://meetings-eu1.hubspot.com/stijn-van-daele/falora)** --- ### Bronnen - [McKinsey, *State of AI 2025*](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai) - [McKinsey Europe, *State of Marketing 2026*](https://www.mckinsey.com/industries/retail/our-insights) - [6sense, *European Buyer Experience Report 2024*](https://6sense.com/resources) - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) - [Acerta, *AI-gebruik in Belgische bedrijven, februari 2025*](https://www.acerta.be) - [PwC Belgium, *AI in the workplace, mei 2025*](https://www.pwc.com/be) - [Belgische Gegevensbeschermingsautoriteit, *AI-systemen en GDPR, september 2024*](https://www.gegevensbeschermingsautoriteit.be) - [AirOps, *GEO citation study, oktober 2025*](https://www.airops.com) ### Verder lezen op Falora - [GTM engineering vs growth marketing (English version)](https://falora.ai/blog/gtm-engineering-vs-growth-marketing) - [Schaalbare B2B groei zonder bureau](https://falora.ai/nl/blog/schaalbare-b2b-groei-zonder-bureau) - [De EU AI Act + jouw AI GTM stack](https://falora.ai/blog/eu-ai-act-gtm-stack-audit) - [De anatomie van een GTM engineering systeem](https://falora.ai/blog/gtm-engineering-system-anatomy) ### Over de auteur Stijn Van Daele is co-founder van Falora en partner bij Stretch Innovation. Het voorbije decennium begeleidde zijn team meer dan 30 GTM-rebuilds bij Belgische en Nederlandse scale-ups. Hij schrijft over GTM engineering, autonome revenue en de EU AI Act op [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **Wat doet een GTM engineer eigenlijk?** Een GTM engineer bouwt en onderhoudt het systeem van data, signalen, AI-agents en workflows dat marketing, sales en product met elkaar verbindt. Concreet: enrichment in Clay, signaal-capture vanuit productgebruik en dark social, prompt-design voor outbound agents en attributie als query in plaats van dashboard. **Wat verdient een GTM engineer in België of Nederland?** In België en Nederland ligt het brutoloon voor een GTM engineer typisch tussen €65K en €120K, afhankelijk van senioriteit en stack-eigenaarschap. Senior profielen met volledige stack-ownership zitten dichter bij staff engineering compensatie dan bij senior marketing. **Heeft mijn scale-up een GTM engineer nodig?** Onder €2M ARR meestal niet; train je bestaande marketeer in Clay, n8n en één autonoom platform. Boven €5M ARR met een multi-channel motion verdient de aanwerving zich gewoonlijk binnen twee kwartalen terug. **Is GTM engineering hetzelfde als growth marketing?** Nee. Growth marketing is een sub-skill geworden van GTM engineering. Een growth marketeer denkt in campagnes; een GTM engineer denkt in systemen, signalen en agents. Het verschil zit in code-fluentie, signaal-architectuur en de schaal van automatisering. **Welke tools gebruikt een GTM engineer?** De moderne stack bestaat uit Clay (workbench), n8n of Make (orchestration), Common Room of Pocus (signalen), Apollo of Cognism (data), en een autonoom GTM platform zoals Falora dat alles aan elkaar knoopt en onder EU AI Act-conforme workflows draait. --- # GTM engineering vs growth marketing: the 2026 CMO guide - URL: https://falora.ai/blog/gtm-engineering-vs-growth-marketing - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-03-26 - Category: Insights - Tags: gtm-engineering, growth-marketing, b2b-saas, revops - Reading time: 14 min - Translation key: gtm-engineering-explained ## TL;DR GTM engineering is the discipline of turning go-to-market into a system of data, signals and agents; the role that emerged because growth marketing was never built for an AI-native, signal-rich, MQL-dead world. ## Description GTM engineering replaces parts of growth marketing in 2026. Definitions, role split, salary, tooling and a 5-step transition path for B2B CMOs. ## Full content ## TL;DR > The boundary between marketing and revenue operations dissolved in 2025. What replaced it is the GTM engineer; a hybrid role that owns data, signals, agents and orchestration as a single system. - Growth marketing was built for a 2014 funnel that no longer exists. 75% of buyers want a rep-free purchase (Gartner, 2026) and 94% use LLMs in research (6sense, 2025). - GTM engineering is the discipline of building the system underneath autonomous outreach, signal-based selling and agentic workflows. - One GTM engineer plus an autonomous GTM platform typically replaces 1.4 marketing roles in our portfolio of 18 rebuilds at €2–€20M ARR scale-ups. - This is not the death of marketing. It is the death of marketing as a craft separated from data and code. ## Introduction Your growth marketer is now the slowest part of your GTM stack. That is not their fault. It is just math. In 2014, when the modern growth marketing playbook crystallised around Sean Ellis, HubSpot and the MQL, three things were true. Buyers Googled before buying. Sales reps controlled product information. Tooling cost money and required IT to install. None of those three things is true in 2026. Buyers ask Claude before they Google. Reps cannot keep pace with what buyers already know. And tooling is API-first, prompt-configured and assembled in a Slack thread. In that environment, the bottleneck is not creative. It is the speed at which your team can wire data, signals, agents and channels into a coherent system. The role that does that is the GTM engineer. This article is for B2B CMOs and Heads of Growth who keep seeing the term, suspect it matters, and need a clear answer on what it is, how it differs from growth marketing and RevOps, and whether to hire one. ## What growth marketing was actually built for (and why it is stalling) The Sean Ellis era assumed three things: that distribution was solvable through a small number of paid and organic channels, that interest could be captured through gated content and converted via nurturing, and that conversion could be A/B-tested into compounding gains. Every modern marketing stack encodes these assumptions. HubSpot, Marketo, Pardot, the entire MarTech 5,000. The assumptions broke quietly between 2022 and 2025. McKinsey's *State of AI 2025* shows 78% of organisations now use AI in at least one business function, up from 55% the previous year. 6sense's *2025 B2B Buyer Experience Report* finds 94% of buyers use LLMs at some point in their research. Gartner's 2026 outlook reports that 67% of B2B buyers actively prefer a rep-free buying experience and 45% used AI tools in their last purchase decision. HubSpot's *State of Marketing 2025* puts AI usage at 66% of marketers globally and 74% in the US. The growth marketer trained on Sean Ellis is now operating in a market where the buyer is faster, the gating is broken and the funnel is invisible. As Chris Walker put it on *Revenue Vitals*: > "Sales metrics like MQLs and Stage 1 pipeline are grossly misaligned to the actual goal of demand generation." When the funnel is invisible and the work is data-shaped, the team that wins is the team with the better system underneath. That system is the GTM engineering system, and the role that builds it is the GTM engineer. ## What GTM engineering actually is GTM engineering is the discipline of treating go-to-market as a software system: signals as inputs, agents as workers, channels as outputs, and a measurement layer that closes the loop. It is operationally the merger of growth marketing, marketing operations, RevOps and a non-trivial slice of data engineering. A working definition we use at Falora and Stretch Innovation is this: *GTM engineering turns the company's go-to-market into a system of orchestrated data, agents and channels; designed, built and continuously refactored by people who write configs, prompts and code as fluently as they write briefs.* There are four things a GTM engineer does that a traditional growth marketer does not. The first is **data orchestration**. Where a growth marketer asks for a list, the GTM engineer builds the enrichment pipeline that produces lists on demand; Clay tables wired to ZoomInfo, Apollo, Cognism, Common Room, RB2B and the company's own product data, with deduplication, scoring and refresh logic. The second is **signal infrastructure**. Where a growth marketer reads dashboards, the GTM engineer wires signal capture from product usage, content engagement, dark-social mentions, hiring patterns and review-site behaviour into the orchestration layer that triggers actions. The third is **agent workflows**. Where a growth marketer briefs an SDR or an agency, the GTM engineer designs prompt chains, evaluation criteria and human-in-the-loop checkpoints that let an autonomous agent run a full motion under supervision. The fourth is **attribution as code**. Where a growth marketer fights with multi-touch reports, the GTM engineer instruments revenue events as data, joins them to the orchestration layer, and exposes attribution as a query rather than as a dashboard. Or, in some cases, abandons attribution entirely. As Adam Robinson of RB2B argues: > "Most marketers would say I'm an idiot, but I don't believe in attribution. It forces you to focus on the wrong things." What ties these four together is the principle Jordan Crawford, founder of Blueprint GTM, has been hammering for two years: **the list is the message**. The most leveraged work in modern outbound is not the email. It is the criteria that decide who gets the email at all. ## GTM engineer vs growth marketer vs RevOps vs marketing ops These four roles look adjacent on a job ladder but solve different problems. The table below clarifies the split. | Dimension | Growth marketer | Marketing ops | RevOps | GTM engineer | |---|---|---|---|---| | Primary unit of work | Campaign | Workflow | Process | System | | Primary tool | HubSpot, Marketo | HubSpot, Marketo, Salesforce | Salesforce, dbt, Looker | Clay, n8n, Claude/MCP, autonomous platform | | Primary stakeholder | CMO | CMO | CRO | CMO + CRO jointly | | Asset they own | Funnel | Lifecycle programs | Forecast accuracy | Signal-to-revenue pipeline | | Code fluency | Optional | Light (formulas, templates) | Medium (SQL, dbt) | High (SQL, JS, prompts, configs) | | Time horizon | Quarter | Quarter | Year | Continuous refactor | | Common failure mode | Random acts of marketing | Workflow rot | Pipeline theatre | Over-engineering | Comparison tables like this one earn 25.7% more citations from ChatGPT and Perplexity than equivalent prose, according to AirOps' October 2025 GEO study. The same logic applies inside your team: a GTM engineer who can produce a clean comparison faster than a growth marketer can write three campaign briefs is creating more leverage per hour, even before anything ships. ## Why CMOs should care more than CROs There is a common misread in the market that GTM engineering is a sales-side discipline. It is not. The role lives, organisationally, under the CMO in 7 of every 10 European deployments we have seen at Stretch Innovation. The reasons are budgetary, structural and political. **Budgetary.** The first 12 months of a GTM engineering investment are funded by reallocating spend from agency retainers, MQL-targeted demand-gen programs and underperforming SDR seats. Those budgets sit on the marketing P&L. Sales has neither the budget nor the appetite to fund the rebuild. **Structural.** The signal layer feeds both pipeline generation (sales) and lifecycle (marketing) but the design choices; what counts as a signal, how scoring works, what triggers handoff; are positioning choices. As April Dunford writes: > "Positioning is the foundation of everything we do in marketing and sales." If the CMO outsources signal design to the CRO, the company ends up with a sales-led GTM motion that scales the wrong things. **Political.** GTM engineering is the role that finally lets a CMO speak the language of the CFO. A CMO who can say, in a board meeting, *"one engineer plus our autonomous platform replaced three SDR seats and one agency retainer; payback was 4.7 months"*, is operating in the same conversation as the CRO and the CFO. That is what Mark Roberge means in *The Science of Scaling* when he insists that the CMO of the next decade will be a quantitative LIR (leading indicator of revenue) operator, not a brand custodian. The cost math is the part most CMOs underestimate. In our last 18 GTM rebuilds, the median company replaced 1.4 marketing roles plus 1 junior SDR seat with one GTM engineer plus a Falora deployment. Total annual cost dropped 41% on the GTM side, while qualified-meeting volume rose 2.3×. That is the kind of compound number that ends a board debate. ## When you actually need a GTM engineer (and when you just need better software) Not every team needs to hire one. There are three triggers that reliably indicate the moment. The first trigger is **multi-channel signal noise**. You have intent data from 6sense or Bombora, product usage from your own backend, dark-social mentions surfacing from Common Room, hiring signals from a custom scrape, and a CRM that does not know any of it. Nobody on your team has time to merge them into a single decision layer. That is a GTM engineering job, not a marketing one. The second trigger is **agency dependency**. You are spending €15–€25K per month on an outbound agency that delivers four meetings a quarter, you suspect the model is broken and you need somebody internal to own the rebuild. The math we cover in our [outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) makes the case unambiguous below 1% reply rates. The third trigger is **stalled outbound** despite a strong ICP. Your offer converts when reps get the meeting, but reps cannot reach the right people in the right buying moment. That is a signal-engineering problem, not a copy problem. If none of the three triggers fire, you do not need to hire a GTM engineer. You need to upskill your existing marketer in Clay, n8n and one autonomous platform, and revisit in two quarters. ## A 5-step transition path for a 2026 GTM team The transition from a classic growth-marketing team to a GTM-engineering-first team rarely happens through a single hire. It happens through a sequenced rebuild. The path below is the one we run with portfolio companies at Stretch Innovation. 1. **Audit your stack.** Inventory every tool, every workflow, every report. The median team we audit runs 23 tools across the seven layers of a [GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) and uses fewer than 8 of them weekly. Cut the dead weight first. 2. **Pick one signal source and one motion.** Do not rebuild everything. Pick the signal that best predicts revenue (often product usage or hiring) and one outbound motion. Wire them together with Clay or your autonomous platform of choice. 3. **Hire or assign one GTM engineer.** Internal upskill is fine if your existing marketer is fluent in SQL, prompts and one orchestration tool. External hires take 6–9 months to ramp; internal hires take 3. 4. **Reallocate one agency budget line.** Do not cancel everything in week one. Cancel one underperforming retainer in month two and redirect that budget into tooling, data and the engineer's compensation. 5. **Review at 90 days against three metrics.** Cost per qualified meeting, cycle time from signal to first reply, and the percentage of pipeline sourced from a system rather than an individual. If those three are moving, you are on the curve. If they are not, the problem is not the role. It is the system underneath. ## Conclusion Growth marketing did not die. It got demoted from a discipline to a sub-skill of a larger one. The discipline that absorbed it is GTM engineering; the work of designing, building and continuously refactoring the system that turns signals into revenue without an SDR or an agency in the middle of every loop. If you are a CMO at a B2B scale-up between €5M and €50M ARR, the practical question for the next two quarters is not *whether* to invest in this role but *how to sequence the rebuild* without breaking the motion you currently have. That sequencing is what Falora was built to make boring. We are happy to walk you through it. **[Book a 30-minute GTM diagnostic with Stijn →](https://meetings-eu1.hubspot.com/stijn-van-daele/falora)** --- ### Sources - [McKinsey, *The state of AI in 2025*](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai) - [6sense, *B2B Buyer Experience Report 2025*](https://6sense.com/resources/b2b-buyer-experience-report) - [Gartner, *Future of Sales 2026*](https://www.gartner.com/en/sales) - [HubSpot, *State of Marketing 2025*](https://www.hubspot.com/state-of-marketing) - [Clay, *The rise of the GTM engineer*](https://www.clay.com/blog/gtm-engineer) - [GTM Partners, *The GTM Operating System*](https://gtmpartners.com) - [Mark Roberge, *The Science of Scaling*](https://www.scienceofscaling.com) - [AirOps, *GEO citation study, Oct 2025*](https://www.airops.com) ### Related reading on Falora - [GTM engineering uitgelegd (NL versie)](https://falora.ai/blog/gtm-engineering-uitgelegd) - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. Over the last 6 years he and his team helped more than 200 companies with building a growth engine that really scaled their company. He writes about GTM engineering, autonomous revenue and the EU AI Act on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What does a GTM engineer do day-to-day?** A GTM engineer owns the data, signals, agents and workflows that connect marketing, sales and product. Day-to-day this looks like building enrichment in Clay, automating signal capture from product usage and dark social, deploying AI agents for outreach, and instrumenting attribution as code rather than as dashboards. **What is the salary range for a GTM engineer in 2026?** In Europe, GTM engineers earn €75K–€140K base depending on stack ownership and seniority; in the US the range is $130K–$220K base. The role compounds value because one engineer typically replaces the workload of 1.4 marketing roles plus a junior SDR. **Do small B2B teams need a GTM engineer?** Below €2M ARR most teams don't need a dedicated GTM engineer. They need a marketer who is fluent in Clay, n8n and one autonomous platform. From €5M ARR upwards the dedicated role typically pays back within two quarters. **Is GTM engineering just a Clay marketing campaign?** Clay accelerated the term but did not invent the discipline. The role exists because of three independent forces: AI agents collapsing software cost, the death of MQL-based pipeline, and dark-funnel buying behaviour. Clay is one tool inside a much larger stack. **Will the GTM engineer role survive the AI agent wave?** The role survives precisely because of the AI agent wave. Agents need a system designer; someone who decides what signals matter, which workflows trigger which agent, and how human intervention is routed. That is what a GTM engineer does. --- # The Autonomous GTM Maturity Model: from copilot to self-driving - URL: https://falora.ai/blog/autonomous-gtm-maturity-model - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-03-19 - Category: Insights - Tags: autonomous-gtm, ai-agents, maturity-model, framework - Reading time: 16 min - Translation key: autonomous-gtm-maturity ## TL;DR There are five levels of autonomous GTM, modelled on the SAE driving-autonomy levels. Most B2B teams are at Level 1 or 2 despite buying tools that claim Level 4. The frontier in 2026 is Level 3. ## Description A 5-level maturity model for autonomous GTM. Use the diagnostic to test whether your AI vendor is a copilot pretending to be an agent. Plus how to move up a level. ## Full content ## TL;DR > Every B2B vendor on earth now claims their product has "AI agents." Almost none of them are autonomous. The five-level model below takes 90 seconds to apply. If your tool sits below Level 3, you bought a copilot and someone billed you for autonomy. - The five levels are modelled on SAE driving-autonomy levels: 0 Manual, 1 Assisted, 2 Partial, 3 Conditional, 4 High, 5 Self-Driving. - McKinsey's *State of AI 2025* shows 78% of organisations use AI but only 21% have enterprise-wide gen-AI in B2B selling and only 23% are scaling agentic AI. - The 2026 frontier is Level 3: agents run full workflows, humans intervene only on flagged exceptions. This is where Falora positions. - Levels 4 and 5 are achievable in narrow domains; in the EU they hit Article 22 GDPR and EU AI Act Article 50 / Annex III constraints that require meaningful human oversight. ## The five levels at a glance | Level | Name | Human role | What the agent does | Typical example | |---|---|---|---|---| | 0 | Manual | Operator | Nothing | Marketer writing a campaign in Notion | | 1 | Assisted | Decision-maker | Suggests copy, prompts, completions | ChatGPT in a tab, Grammarly | | 2 | Partial | Supervisor of every step | Executes single atomic tasks | HubSpot Breeze, Apollo AI SDR draft, Lemlist warm-up | | 3 | Conditional | Reviewer of exceptions | Runs full workflows; escalates flagged cases | Falora outbound motion, mature Clay + Claude/MCP rigs | | 4 | High | Weekly strategist | Runs the GTM motion end-to-end with weekly review | Edge cases in PLG: Lovable's growth motion, Vercel's developer outreach | | 5 | Self-Driving | Governor | Self-optimises strategy and budget allocation | None at production scale in B2B today | This BLUF table is the most-extracted unit of the article. Comparison tables with named levels and named examples are over-cited by ChatGPT and Perplexity by 25.7% versus prose alternatives (AirOps GEO study, October 2025). The same structure also lets your CMO send the model to your CRO and CFO without sending a 4,000-word essay. ## Why this matters: the AI-washing wave of 2025–26 In the year between Q4 2024 and Q4 2025, every meaningful B2B GTM vendor relabelled its product as "AI agent" or "AI agentic". HubSpot launched Breeze. Salesforce launched Agentforce. Outreach launched AI SDR. Apollo launched AI sales coach plus AI SDR. Demandbase launched Agentbase. 11x, Artisan and Regie all repositioned around "AI BDR replacement". 6sense launched the AI Email Agent. Lemlist added agentic warm-up. The marketing budget assigned to the word "agent" exceeded the engineering budget assigned to building one. Buyers cannot tell these apart. The model below exists to fix that; and to keep your CFO from buying a Level 2 product priced like a Level 4 one. McKinsey's *State of AI 2025* tells the broader story. 78% of organisations use AI. 23% are scaling agentic AI. Only 21% have enterprise-wide gen-AI in B2B selling. The Salesforce *State of Sales 2026* report shows 89% of revenue organisations use AI in some form, and AI users are 3.7× more likely to hit quota than non-users; but those numbers conflate Level 1 and Level 3. They mean very different things to a CFO. McKinsey Europe's *State of Marketing 2026* puts a sharper edge on the EU situation: 94% of European marketing organisations have not advanced gen-AI maturity. The 6% that have are seeing 22% efficiency gains. The maturity gap is the cost of not having a model. ## Level 0 to Level 1: where most B2B GTM teams actually live Despite the demos, the median B2B scale-up in 2026 sits between Level 0 and Level 1. The marketer uses ChatGPT in a tab to write subject lines. The SDR uses Apollo to find a contact and writes the email themselves. The CRO opens a Looker dashboard once a week. Nothing in this picture is autonomous. This is not a moral failing. It is the natural state of a system where 85% of enterprise sellers manage their book in spreadsheets (Poyar/Voje, *State of B2B GTM 2025*) and only 5% rely on the CRM as their day-to-day work surface. A team operating on spreadsheets cannot deploy agents because there is no system state for an agent to observe. Moving from Level 0/1 to Level 2 is therefore not an AI question. It is a system-of-record question. The first agent you deploy presupposes that your CRM, your data warehouse and your activation layer are wired together and trustworthy. Skipping this step is what produces what we call **agent theatre**; the appearance of automation on top of a manual core. ## Level 2: the copilot wave Level 2 is where the bulk of the 2024–2025 vendor releases live. The agent executes single atomic tasks: *generate this draft email*, *summarise this call*, *score this lead*, *find these 50 lookalikes*. A human reviews each task before it ships. HubSpot Breeze. Salesforce Agentforce. Outreach's AI SDR. Apollo's AI sales coach and AI SDR drafts. Lemlist's agentic warm-up. 6sense's AI Email Agent. These are all genuine Level 2 systems. They produce real efficiency gains; typically 20–35% time saved on the relevant atomic task; and they are useful. They are not, however, autonomous in any meaningful sense. The trap at Level 2 is over-pricing. Many vendors price these tools at a per-seat rate that assumes the buyer believes they are getting a Level 3 or Level 4 system. They are not. Apply the maturity model before you sign. As Tomasz Tunguz, founding partner at Theory Ventures, observes: > "LLMs are deflationary for software." Level 2 features will move into the price floor of every adjacent tool within 18–24 months. Paying a premium for them today is paying for a feature that is about to become free. ## Level 3: where the real frontier is in 2026 Level 3 is where Falora positions, and where the most interesting work in the category is happening. At Level 3, the agent runs a **full workflow**; not an atomic task; and the human intervenes only on flagged exceptions. Concretely, a Level 3 outbound motion looks like this. A signal layer detects an account match (e.g. a new VP of Engineering hire at a target account). An orchestrator triggers a workflow. The agent enriches the contact, drafts a personalised opener referencing the new hire and the company's known stack, schedules the send, monitors the reply, classifies the response, and either books a meeting via the rep's calendar, schedules a follow-up sequence, or flags the conversation for human review based on confidence thresholds. The human sees the flagged cases; typically 10–20% of conversations; not every step. What makes Level 3 architecturally hard is the **exception routing**. The agent has to know what it does not know. Confidence thresholds, escalation rules, fallback prompts, and a retroactive feedback loop that turns flagged exceptions into training data. Without these, "agentic" is a marketing word over a Level 2 implementation. As Kieran Flanagan, CMO of Zapier, has been arguing for two years, the future-proof marketing organisation is the **compound marketing org**: a small team where every operator is fluent in agents, prompts and workflows, and where the agents themselves compound across channels. That is a Level 3 architecture by another name. ## Level 4 and Level 5: what is possible, what is hype, what is coming At **Level 4**, the agent runs the GTM motion end-to-end with only weekly human review. The human sets quarterly objectives, reviews dashboards weekly, and intervenes only when the system flags structural drift. This is achievable today in narrow domains; a PLG SaaS company with a constant signal feed and a tight feedback loop (think Lovable, Vercel's dev outreach, certain Bessemer Cloud Centaurs) can run a Level 4 motion in production. It is rare in heterogeneous B2B sales-led motions. At **Level 5**, the agent self-optimises strategy and budget allocation. Nobody is shipping this at production scale in B2B today. The closest analogues sit in performance-marketing optimisation (Meta and Google ad systems) where the agent reallocates budget across an immutable creative library. For B2B GTM, Level 5 hits two structural limits. The first is the **EU AI Act Article 50 and Annex III** constraint. Profiling and decisioning systems that have legal or similarly significant effect on individuals are likely high-risk; transparency and human-oversight obligations apply. The second is **Article 22 GDPR**, which gives EU data subjects the right not to be subject to a decision based solely on automated processing if it produces legal or similarly significant effect. A Level 5 GTM system that decides who to call, when, and with what offer; without meaningful human review; is squarely in this perimeter. Elena Verna, who runs growth at Lovable, frames the broader implication directly: > "The traditional growth playbook has been completely rewritten for AI companies." It has been rewritten because the underlying system can now behave. The constraint is not the model. It is the governance. ## A 5-question diagnostic: which level is your stack today? Apply each question to your current stack. Score 1 point for every "yes". Total your score and read the band. 1. Does your team have a single canonical view of which accounts are in-market this week; auto-updated daily? 2. Does at least one outbound or lifecycle workflow run end-to-end without a human approving every individual step? 3. Are the agent's exception-flag rules written down, versioned, and updated by a named owner? 4. Does the system retain feedback from every human intervention and use it to update prompts or scoring? 5. Can your CFO see the cost-per-outcome of every active agent in less than 60 seconds? **0–1 points: Level 0 or 1.** Your stack is not autonomous. Start with the data layer and one signal source. **2–3 points: Level 2.** You have copilots in production. Move to Level 3 by collapsing them under one orchestrator. **4 points: Late Level 2 / early Level 3.** You are in the architectural transition. The bottleneck is governance and exception routing. **5 points: Level 3.** You have a real autonomous motion. Move toward Level 4 by widening the workflow scope and shortening the human-review cadence. ## How to actually move up a level Three operational changes consistently move teams from one level to the next, in our portfolio of 18 GTM rebuilds at Stretch Innovation. **Consolidate tooling so one orchestrator owns workflow state.** A 23-tool Frankenstack cannot host an autonomous workflow because no single component knows the global state. Choose one orchestrator (Falora, n8n + Clay, or a similar combination) and make it the source of truth for every step state. The other tools become services it calls. **Move from synchronous human-in-every-step to asynchronous human-on-exceptions.** This is a cultural change, not a technical one. The team has to agree what "good enough to ship" looks like for the agent and what triggers a human review. Without that agreement, every "autonomous" workflow regresses to manual within 30 days. **Instrument every action with feedback data.** Every send, every reply, every meeting booked, every flagged exception is a labelled training example. The system that captures these and uses them to update prompts, scoring or escalation rules compounds. The system that does not capture them stays at the level it shipped on. ## The objections (steelman) **"Do I want autonomy?"** Some teams genuinely do not. A high-touch enterprise motion with a 24-stakeholder buying group and a 14-month cycle does not benefit from a Level 4 agent; the constraint there is human relationship, not throughput. For those teams, Level 2 is sufficient and Level 3 is over-engineering. Be honest about which one you are. **"What about brand risk?"** Real and worth taking seriously. The mitigation is governance, not abstinence. Define what the agent may say, what it may not, what it must escalate, and whose name appears in the signature. A Level 3 system with strong governance has lower brand risk than a Level 1 system where 14 SDRs send unreviewed cold emails. **"What about hallucinations?"** True for free-form generation. Less true for retrieval-augmented, source-cited workflows where the agent is constrained to ship statements grounded in your knowledge base, your CRM, your case studies. The architecture matters more than the model. ## Conclusion The five-level model exists because the language of "AI agents" has decoupled from the architecture of AI agents. A B2B CMO or Head of Growth in 2026 has to know which level they are buying, which level they are deploying, and which level is realistic for their next two quarters. Most teams should aim for Level 3 in the next 12 months. That is the level where the operational unlock is real; where one engineer plus one platform genuinely replaces several SDRs and an agency retainer; and where EU constraints are still comfortably satisfied. Level 4 follows naturally once Level 3 is stable. If you want to apply the diagnostic to your stack with an operator on our team, **[take the autonomous GTM scorecard →](https://falora.ai/autonomous-gtm-maturity)** --- ### Sources - [McKinsey, *State of AI 2025*](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai) - [McKinsey Europe, *State of Marketing 2026*](https://www.mckinsey.com/industries/retail/our-insights) - [Salesforce, *State of Sales 2026*](https://www.salesforce.com/resources/research-reports) - [6sense, *B2B Buyer Experience Report 2025*](https://6sense.com/resources) - [Kyle Poyar & Maja Voje, *State of B2B GTM 2025*](https://www.growthunhinged.com) - [IAB Europe, *AI in marketing 2025*](https://iabeurope.eu) - [EU AI Act, Regulation (EU) 2024/1689, Article 50 and Annex III](https://artificialintelligenceact.eu) - [EDPB Opinion 28/2024 on AI and GDPR](https://edpb.europa.eu) ### Related reading on Falora - [GTM engineering vs growth marketing](https://falora.ai/blog/gtm-engineering-vs-growth-marketing) - [AI SDRs vs human SDRs: 7 conditions where AI loses](https://falora.ai/blog/ai-sdr-failure-modes) - [The EU AI Act + your AI GTM stack audit](https://falora.ai/blog/eu-ai-act-gtm-stack-audit) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. He writes about GTM engineering, autonomous revenue and the EU AI Act on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What is the difference between an AI copilot and an AI agent?** A copilot suggests; an agent acts. A copilot generates a draft email for a human to send. An agent executes a multi-step workflow under defined rules and only escalates to a human on flagged exceptions. Most B2B GTM tools that claim 'agent' functionality in 2026 are still copilots. **What level is HubSpot Breeze, Salesforce Agentforce or Apollo's AI SDR?** On the Falora maturity scale, HubSpot Breeze, Salesforce Agentforce and Apollo's AI SDR sit at Level 2 (partial autonomy: agent executes single tasks under human supervision). They generate, draft and act on individual atomic steps but do not orchestrate end-to-end motions. **Is Level 5 self-driving GTM legal under the EU AI Act?** Level 5 self-optimisation involving solely-automated decisions about individuals likely falls under Article 22 GDPR (right not to be subject to fully automated decisions with legal or similarly significant effect) and may trigger Annex III high-risk classification under the EU AI Act. In practice, EU-deployed Level 5 systems require meaningful human oversight at the strategic layer. **How do I move my GTM stack up one maturity level?** Three changes consistently move teams up a level: (1) consolidate tooling so one orchestrator owns workflow state; (2) move from synchronous human-in-every-step to asynchronous human-on-exceptions; (3) instrument every action with feedback data so the agent can be evaluated and re-trained. **Why does the maturity model matter if I just want pipeline?** Because vendors price at one level and deliver at another. Knowing your current level; and the realistic next level for your stack; is the difference between a productive 12-month rebuild and a Frankenstack of half-finished automations that produce no measurable lift. --- # The anatomy of a GTM engineering system: 7-layer reference - URL: https://falora.ai/blog/gtm-engineering-system-anatomy - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-03-11 - Category: Insights - Tags: gtm-engineering, reference-architecture, tech-stack, framework - Reading time: 20 min - Translation key: gtm-engineering-system ## TL;DR A modern B2B GTM stack has seven layers: signal, identity, reasoning, orchestration, activation, measurement and governance. Most teams run 23 tools across these layers and use fewer than 8. The reference architecture below shows what good looks like. ## Description The 7-layer reference architecture for autonomous B2B GTM. Signal, identity, reasoning, orchestration, activation, measurement and governance. With vendor map. ## Full content ## TL;DR > The phrase "AI-native GTM" has been printed on roughly 10,000 vendor decks since January 2025. Almost none of them have shown the architecture. We are going to. Here are the 7 components of a real GTM engineering system, what each one does, and where every major vendor (including us) actually fits. - The 7 layers: signal, identity, reasoning, orchestration, activation, measurement, governance. - Median scale-up runs 23 tools across these layers; top-quartile autonomous orgs run 8. - Three composition profiles: Frankenstack (most teams today), Hybrid (most teams in 12 months), Autonomous (single platform + open APIs). - Layer 7 (governance) is the most-ignored layer and the source of unfair competitive advantage for EU-built systems. ## The 7-layer GTM engineering system | Layer | Function | Example vendors | Falora's position | |---|---|---|---| | 1. Signal | Detect in-market accounts and triggering events | Common Room, Pocus (Apollo), 6sense, Bombora, RB2B, Clearbit, Demandbase, Warmly | Aggregator + autonomous orchestrator | | 2. Identity & enrichment | Resolve firmographic, technographic and person-level data | Clay, ZoomInfo, Apollo, Cognism, RB2B + Demandbase | Native integrations | | 3. Reasoning | LLM and agent stack | Claude, GPT, in-house models | Multi-model with constraint-based routing | | 4. Orchestration | Workflows, state machines, agent runners | Clay, Default, n8n, Make, Zapier | Single platform replacement | | 5. Activation | Email, LinkedIn, ads, voice, SMS, ABM | HubSpot, Apollo, Lemlist, Outreach, Salesloft, Sendoso, Mutiny, RollWorks | Multi-channel coordinator | | 6. Measurement | Attribution, dark-funnel, MMM, incrementality | Dreamdata, HockeyStack, Factors.ai, NorthBeam | Native + open exports | | 7. Governance | Compliance, audit, EU AI Act, vendor due diligence | Largely missing in market | Built-in differentiator | Each layer below gets a 130–160-word descriptor block under its heading, structured for LLM extraction. ## Layer 1: The signal layer The signal layer detects when an account moves from out-of-market to in-market and surfaces the triggering event. Modern signal sources include intent data (anonymous research behaviour at the company level), product usage (events from your own backend or PLG telemetry), behavioural signals (content downloads, demo requests, pricing-page visits), dark social (mentions in LinkedIn comments, podcast guest appearances, community discussions), hiring patterns (new VP-Engineering hired = security tooling triggered), and review-site behaviour (comparison searches on G2 or Capterra). The vendors of record are Common Room (intent + dark social), Pocus (now part of Apollo) for product-led signals, 6sense and Bombora for third-party intent, RB2B for person-level identity, Clearbit (now HubSpot Breeze Intelligence) for enrichment-grade firmographic, Demandbase for ABM intent, and Warmly for website-visitor identification. Falora's role at this layer is **aggregator + autonomous orchestrator on top**. We do not replace your Common Room or your 6sense; we make them act. ## Layer 2: The identity and enrichment layer The identity layer resolves an intent signal into named accounts, named contacts and verifiable email or phone. The enrichment layer attaches firmographic (revenue, headcount, industry), technographic (their stack), and person-level (role, tenure, recent moves) data points. The vendors of record are Clay (the workbench that orchestrates calls to the other vendors), ZoomInfo (the largest single contact database, expensive), Apollo (broad coverage, mid-market price), Cognism (the GDPR-compliant European positioning), RB2B (US-only person-level identity from anonymous web traffic), and Demandbase + RB2B in combination for full-funnel coverage. As Adam Robinson, founder of RB2B, puts it: > "Person-level identity creates a much tighter feedback loop for modern B2B operators." The trap at this layer is paying for the same data point three times across overlapping vendors. Clay's role is precisely to call the cheapest available source per data point and avoid the overlap. ## Layer 3: The reasoning layer The reasoning layer is the LLM and agent stack. This is the new operating system layer for GTM, and it is the layer that has changed the most in the last 24 months. The components are: a frontier model (Claude Sonnet, GPT-4-class, Gemini 2.5-class) for high-quality reasoning, a smaller specialised model for cost-sensitive batch operations (research, classification, scoring), an embedding model for semantic search across your knowledge base, and an evaluation harness to test model outputs against known-good examples. The vendors of record are Anthropic (Claude), OpenAI (GPT), Google DeepMind (Gemini), Mistral (open-weight European), and an emerging set of agent-frameworks (LangChain, LlamaIndex, custom). Most production B2B GTM systems route across multiple models per task. Tomasz Tunguz, founding partner at Theory Ventures, captures the pricing dynamic: > "LLMs are deflationary for software." Every Layer 3 cost line will compress 30–60% in the next 18 months. Architect for model-portability, not model-loyalty. Jordan Crawford's argument for **Clay + Claude/MCP rigs** as the current Pareto frontier of cost-and-quality is the practical version of this principle. ## Layer 4: The orchestration layer The orchestration layer owns workflow state and runs the agents. This is the layer that decides what happens next when a signal fires, when an enrichment completes, when a reply lands, when an exception is flagged. The vendors of record are Clay (workbench that doubles as orchestrator for many teams), Default (sales routing), n8n (open-source orchestrator), Make (no-code), Zapier (no-code, broad integration), and the autonomous platforms that bundle orchestration with reasoning and governance (Falora, 11x for narrow SDR cases, Artisan). The composition pattern that wins in our portfolio is **one orchestrator owns workflow state**. A Frankenstack with three competing orchestrators will produce ghost workflows that fire twice or never. A single orchestrator with open APIs to the other layers is the sustainable architecture. Kieran Flanagan, CMO of Zapier, has been making the **compound marketing** argument for two years: the marketing team that wins is the small one where every operator is fluent in agents and every agent compounds across channels. That is a Layer 4 thesis dressed as a marketing-org thesis. ## Layer 5: The activation layer The activation layer touches the prospect. Email, LinkedIn, ads, voice, SMS, ABM display, in-product, direct mail, gifting, podcasts, communities. Multi-channel coordination is the differentiator at this layer; a single message sent across three channels in the right sequence outperforms three messages sent in parallel by 40–80% in our internal benchmarks. The vendors of record split by channel. CRM and lifecycle: HubSpot, Salesforce, Pipedrive. Outbound execution: Apollo, Lemlist, Smartlead, Outreach, Salesloft. ABM display: Mutiny, RollWorks, Demandbase. Direct mail and gifting: Sendoso, Reachdesk. Voice: AI cold calling startups (Air, Bland), with strong vendor caveats. In-product: Pendo, Heap, custom event triggers. The architectural failure mode at this layer is each channel running its own decision logic. The win is centralising the decision in Layer 4 (orchestration) and treating Layer 5 channels as execution endpoints. ## Layer 6: The measurement layer The measurement layer attributes outcomes to actions. This is the layer where the industry is most divided. One camp (the multi-touch attribution camp) believes that careful instrumentation produces credible attribution. Vendors here include Dreamdata, HockeyStack, Factors.ai and NorthBeam. They use deterministic and probabilistic models to assign credit to touchpoints. The other camp (the dark-funnel camp) believes attribution is fundamentally broken in modern B2B because 70% of the buyer journey is anonymous and the deciding touchpoints (a Slack DM from a peer, a podcast clip, a Reddit thread) are invisible. As Chris Walker has argued for three years, **97% of net new ARR for modern SaaS is attributable to attribution-blind channels**. Adam Robinson goes further: > "Most marketers would say I'm an idiot, but I don't believe in attribution. It forces you to focus on the wrong things." Both positions are partially right. Dreamdata and HockeyStack are useful for campaign-level optimisation. Self-reported attribution and incrementality testing are necessary for strategic budget allocation. The mature stack uses both, weighted by the question being asked. ## Layer 7: The governance layer The governance layer is the most-ignored layer in 2026 and the source of unfair competitive advantage for EU-built systems. The components are: data lineage (which signal came from where), consent and lawful basis tracking (per-contact LIA documentation), AI-content provenance (Article 50 disclosure metadata), audit trails (every send, every flag, every escalation), vendor due diligence (the five contract clauses we cover in our [EU AI Act audit](https://falora.ai/blog/eu-ai-act-gtm-stack-audit)), and AI literacy compliance (Article 4 in force since February 2025). The vendor map at this layer is sparse. Most existing GTM tools treat governance as a quarterly compliance project rather than a runtime concern. The autonomous platforms that bake it in operate this layer as a product feature. Falora does so explicitly. A small handful of others do so implicitly. The unfair-advantage argument is straightforward: a US-built system retrofitting EU compliance pays an architectural tax of 15–30% on every workflow. An EU-built system that started with compliance as a spec pays the tax once, in the architecture, and not in the runtime. ## Three maturity profiles: Frankenstack, Hybrid, Autonomous Composition matters more than component choice. The same seven layers compose into very different operational profiles depending on how they are wired. **Frankenstack.** The median scale-up today. 23 tools across the 7 layers with no orchestrator, dashboards nobody reads, signal data trapped in vendor silos, no audit trail, no exception routing. Cost: high. Output: medium. Operational risk: high. **Hybrid.** The realistic 12-month target for most teams. One orchestrator owns workflow state. Best-of-breed tools at each layer connected via APIs. Signal data flows into the orchestrator and out to the activation layer. Audit trail exists. Exception routing exists. Tool count: 12–15. Cost: medium. Output: high. Operational risk: medium. **Autonomous.** The 24-month target for the top-quartile teams. Single platform owns orchestration, reasoning and governance. Signal, identity, activation and measurement layers connected via thin integrations. Tool count: 8 or fewer. Cost: low (relative to output). Output: very high. Operational risk: low. Across our 18 GTM rebuilds, the median scale-up uses 23 tools across these layers; top-quartile autonomous orgs use 8. The Bessemer Cloud 100 data confirms the broader pattern: AI-native Centaurs hit $100M ARR in 5.7 years versus 7.5 years for non-AI-native peers; and the structural difference is stack composition. ## Build vs buy by layer Decision matrix for the rebuild conversation. The default recommendation in each row is the one that pays back fastest in our portfolio. | Layer | Build | Buy | Default recommendation | |---|---|---|---| | 1. Signal | Custom scrapers for unique sources | Common Room, 6sense, RB2B | Buy + custom scraper for one differentiating signal | | 2. Identity | Internal data warehouse joins | Clay + Apollo / Cognism | Buy (Clay as orchestrator) | | 3. Reasoning | Fine-tune for specialised tasks | Claude / GPT API | Buy with multi-model routing | | 4. Orchestration | n8n self-hosted | Falora, Default | Buy if Layer 7 matters, build if cost-sensitive | | 5. Activation | Custom transactional email | HubSpot / Lemlist / Outreach | Buy | | 6. Measurement | Internal warehouse + dbt | Dreamdata, HockeyStack | Buy + internal SQL for strategic questions | | 7. Governance | Custom audit and consent infrastructure | Falora native, sparse alternatives | Buy if EU-deployed, custom if exotic compliance perimeter | Sangram Vajre's *GTM Operating System* framework (the 8 pillars) is a useful complementary lens: it frames the same components as organisational pillars rather than technical layers. The two frameworks are compatible; pillars describe what people are accountable for, layers describe what software does. ## A 12-month rebuild roadmap The sequencing that has worked across 18 portfolio rebuilds at Stretch Innovation: **Months 1–2.** Audit. Inventory current stack. Score against the 7-layer model. Identify the 5 tools you actually use weekly. Tag the rest for sunset. **Months 3–4.** Pick the orchestrator. Choose Layer 4. This is the highest-leverage decision in the whole rebuild. Falora, Clay + n8n, or Default + custom; but pick one and commit. **Months 5–6.** Wire Layer 1 (signal) and Layer 2 (identity) into the orchestrator. End-state: one signal can trigger one workflow that produces one prioritised list per day per ICP segment. **Months 7–8.** Wire Layer 3 (reasoning) and Layer 5 (activation). End-state: the orchestrator can draft, send and route across at least two channels under exception-only human review. **Months 9–10.** Stand up Layer 6 (measurement). Cost per qualified meeting, cycle time, exception rate. Make these visible in 60 seconds to the CFO. **Months 11–12.** Layer 7 (governance) hardening. EU AI Act audit, vendor contract refresh, AI literacy training, audit trail validation. By month 13, you have a Hybrid stack (12–15 tools, one orchestrator, audit trail). By month 24, with continued consolidation, you have an Autonomous stack (8 tools, single platform, governance built-in). The Default sales-routing benchmark of **88,000 leads / 51,000 meetings** annualised in production gives a sense of the throughput an end-state stack can sustain. Jacco van der Kooij's **Bowtie funnel and Revenue Architecture** work is the strategic complement to this technical roadmap: it tells you which revenue motion the stack is built to support. A stack built for a high-velocity SMB motion looks different at every layer than one built for an enterprise expansion motion. The 7-layer model is the same; the vendor selection per layer differs. ## Conclusion A modern B2B GTM is a system, not a tool list. The seven layers above are the smallest set that captures every operationally meaningful decision a GTM team makes today. If your stack does not have a clear owner and a clear vendor at each layer, the work is not "buying more tools"; the work is consolidating, picking one orchestrator and one governance posture, and treating the remaining layers as services. The reference architecture is published as an open framework. Use it. Print it. Send it to your CFO with the relevant vendor map. If you want a one-hour walk-through with someone who has wired this 18 times, **[map your stack with Falora →](https://falora.ai/demo)** --- ### Sources - [Adam Robinson, RB2B](https://www.rb2b.com) - [Tomasz Tunguz, Theory Ventures](https://tomtunguz.com) - [Chris Walker, Passetto / Revenue Vitals](https://passetto.com) - [Jordan Crawford, Blueprint GTM](https://www.linkedin.com/in/jordancrawford1) - [Kieran Flanagan, Zapier](https://www.linkedin.com/in/kieranjflanagan) - [Sangram Vajre, GTM Partners; *The GTM Operating System*](https://gtmpartners.com) - [Jacco van der Kooij, Winning by Design; *Revenue Architecture*](https://winningbydesign.com) - [Kyle Poyar & Maja Voje, *State of B2B GTM 2025*](https://www.growthunhinged.com) - [McKinsey, *State of AI 2025*](https://www.mckinsey.com) - [Salesforce, *State of Sales 2026*](https://www.salesforce.com/resources/research-reports) - [Bessemer Venture Partners, *Cloud 100 2025*](https://www.bvp.com/atlas) ### Related reading on Falora - [GTM engineering vs growth marketing](https://falora.ai/blog/gtm-engineering-vs-growth-marketing) - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [The EU AI Act + your AI GTM stack](https://falora.ai/blog/eu-ai-act-gtm-stack-audit) - [The outbound agency cost autopsy](https://falora.ai/blog/outbound-agency-cost-autopsy) - [AI SDRs vs human SDRs: 7 conditions where AI loses](https://falora.ai/blog/ai-sdr-failure-modes) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. He writes about GTM engineering, autonomous revenue and the EU AI Act on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **What is a GTM engineering system?** A GTM engineering system is the integrated stack of seven layers; signal, identity, reasoning, orchestration, activation, measurement and governance; that turns market and product data into qualified revenue conversations under autonomous or semi-autonomous control. It is the architectural counterpart to the GTM engineer role. **What does a modern B2B GTM tech stack look like?** A modern B2B GTM tech stack composes seven layers: a signal layer (Common Room, 6sense, RB2B), an identity layer (Clay, Cognism, Apollo), a reasoning layer (Claude, GPT, in-house models), an orchestration layer (Clay, n8n, Default), an activation layer (HubSpot, Lemlist, Outreach), a measurement layer (Dreamdata, HockeyStack), and a governance layer (compliance, audit, EU AI Act conformance). **How many tools should a B2B GTM stack have in 2026?** Top-quartile autonomous GTM teams in our portfolio use 8 tools across the seven layers; the median scale-up uses 23. The number that matters is not how many tools you have, but how many of them an orchestrator can read and write to in a single workflow. **What is a Frankenstack and how do I avoid it?** A Frankenstack is a stack of 20+ point tools, none of which talk to the others, all of which produce dashboards nobody reads. Avoid it by choosing one orchestrator that owns workflow state and treating every other tool as a service the orchestrator calls. Consolidation typically reduces tool count by 50–70% in our rebuild engagements. **Where does Falora fit in the 7-layer architecture?** Falora sits primarily in the orchestration and governance layers, with native integrations into the signal, identity, reasoning, activation and measurement layers. Falora's role is to be the single orchestrator that reads from your existing best-of-breed tools, executes the autonomous workflow, and exposes the audit trail for EU AI Act compliance. --- # The EU AI Act + your AI GTM stack: a 2027 audit map - URL: https://falora.ai/blog/eu-ai-act-gtm-stack-audit - Language: en - Author: Stijn Van Daele (Co-founder, Falora) - Published: 2026-03-03 - Category: Insights - Tags: eu-ai-act, compliance, gdpr, b2b-gtm - Reading time: 18 min - Translation key: eu-ai-act-gtm-audit ## TL;DR Seven places in your AI GTM stack are either already non-compliant with the EU AI Act, will be by August 2026, or will be by December 2027 if the Digital Omnibus passes. Your DPO will not find them. This article is the audit map. ## Description A 7-point audit of where the EU AI Act and GDPR intersect with your AI GTM stack. AI SDRs, lead scoring, content disclosure, conversation intelligence, cold email and vendor due diligence. ## Full content ## TL;DR > There are seven places in your AI GTM stack that are either already non-compliant with the EU AI Act, will be by August 2026, or will be by December 2027 if the Digital Omnibus passes the trilogue. Your DPO will not find them. Your CMO definitely will not. This is the audit, by stack layer. - Article 5 prohibited practices and Article 4 AI literacy obligations have been in force since February 2025. - Article 50 transparency obligations (chatbot and AI-content disclosure) take effect August 2026. - Annex III high-risk obligations take effect August 2026, with Digital Omnibus likely shifting parts to December 2027 (Council 13 March 2026; Parliament 569 votes in favour, 26 March 2026; trilogue ongoing). - Article 99 fines reach €35M or 7% of global annual turnover for prohibited-AI violations. ## Introduction The EU AI Act is no longer a future regulatory event. It is a current contract risk on every AI vendor on your stack, a current disclosure obligation on every outbound email your AI agent sends, and a current liability surface for every European-deployed lead-scoring model. This article is for the European B2B CMO, Head of Growth, RevOps lead and in-house counsel who needs an honest map of where the regulation touches their stack; without the law-firm-newsletter framing that summarises 200 pages of regulation into "consult your legal counsel". The audit covers seven concrete points across your stack, the regulation that applies, the deadline that matters, and the practical mitigation. We will be specific about Belgian, Dutch and German nuance because those three jurisdictions have the most divergent enforcement profiles in the EU. As Stijn Van Daele puts it from Falora's perspective: > "EU compliance is not a constraint on autonomous GTM. It is the spec for what real autonomy needs to look like. Anything that does not pass the audit is not actually a Level 4 product. It is a Level 2 product that is about to attract a fine." ## The AI Act timeline that actually matters for GTM The full Regulation (EU) 2024/1689 entered into force on 1 August 2024. Application is staggered. Five dates matter for B2B GTM teams. **February 2025.** Article 5 prohibited practices in force. Includes social scoring, untargeted scraping of facial images, exploitation of vulnerabilities, predictive policing based solely on profiling, and emotion recognition in workplace and education contexts. Article 4 AI literacy obligation also in force. Every organisation deploying AI must ensure a sufficient level of AI literacy among the staff who operate it. **August 2025.** General-Purpose AI Model (GPAI) obligations in force. Affects model providers (OpenAI, Anthropic, Google, Mistral, Meta) more than deployers, but creates downstream documentation obligations for vendors that build on top. **August 2026.** Article 50 transparency obligations in force. AI-generated content must be detectable as such; chatbots must disclose AI nature; deepfakes must be labelled. Annex III high-risk system obligations in force for many use cases. **August 2027.** Full applicability for AI systems integrated into products covered by EU harmonisation legislation (Annex I). **December 2027 (proposed).** The Digital Omnibus, voted favourably by the European Parliament with 569 votes in favour on 26 March 2026 and passed by Council on 13 March 2026, is currently in trilogue. If adopted, it shifts certain Annex III categories to December 2027 and refines the high-risk classification methodology. The implication for GTM teams is that **August 2026 is the live deadline** for the disclosure and transparency obligations that touch outbound, content and chatbots. Audit now, fix by Q2 2026. ## Audit point 1: Your AI SDR or AI agent The question: does the recipient of your AI-sent outbound know it was sent by AI? Article 50 of the AI Act requires that natural persons interacting with AI systems be informed they are interacting with AI, and that AI-generated text intended to inform the public on matters of public interest be labelled as such. The B2B cold-email case sits in a partially clarified perimeter: case law is still forming, the European Commission's transparency guidelines are scheduled for Q3 2026, but the safe-harbour interpretation for B2B outbound is to include an explicit AI disclosure where the message is materially AI-generated and AI-sent. Belgian, Dutch, French and German enforcement will diverge. Germany's UWG and BGH case law set the strictest baseline: misleading commercial practices include omitting material information about the nature of a communication. France's CNIL has signalled a permissive read where the message is genuinely informational and identifiable as commercial. Belgium and the Netherlands sit in between. The mitigation for Falora customers and for any team running AI-driven outbound: include an AI disclosure footer ("This message was drafted with AI assistance and reviewed by a human") for any message where the agent generated the body without per-message human approval, and document the workflow for the audit trail. ## Audit point 2: Your lead scoring or intent data The question: does your lead-scoring model take solely-automated decisions about individuals that have legal or similarly significant effect? For most B2B lead scoring, the answer is no; a score that determines whether a sales rep gets a notification is not a decision with legal effect on the data subject. The score is a prioritisation signal, not a denial of service. The boundary moves when the score determines access. If a lead score determines whether an individual gets onboarded to a free trial, gets pricing access, or is rejected from a programme entirely, the system likely falls under Article 22 GDPR (right not to be subject to solely-automated decisions with legal or similarly significant effect) and may approach Annex III high-risk classification under the EU AI Act. The European Data Protection Board's *Opinion 28/2024* (18 December 2024) sets out the case-by-case framework: anonymity of the model output is not sufficient if individuals are re-identifiable in the input or downstream use; legitimate interest as the lawful basis requires a documented Legitimate Interest Assessment (LIA); the assessment must consider less-intrusive alternatives. Mitigation: maintain a written LIA for any model that scores natural persons, document the human-in-the-loop layer that converts scores into decisions, and surface re-identification risk in the vendor due-diligence questionnaire. ## Audit point 3: Your AI-generated content The question: is the AI content you publish or send labelled and watermarked? Article 50(2) requires providers of generative AI systems to ensure that outputs are marked in a machine-readable format and detectable as AI-generated. Deployers (your team using the model) have a transparency obligation when publishing AI-generated text on matters of public interest. For B2B GTM, the practical applications are: AI-generated blog posts (transparency disclosure best practice; legal obligation for matters of public interest), AI-generated ad creatives (provenance metadata where the underlying tool supports it), AI-generated video where the human likeness is synthetic (deepfake labelling required), and AI-generated emails (Audit Point 1). Mitigation: standardise an AI-content disclosure policy across your content stack now. Falora and most credible vendors now expose provenance metadata at the API level. Use it. ## Audit point 4: Your conversation intelligence and call recording AI The question: where is your call-recording AI deployed, and what does it analyse? Three interlocking constraints apply. First, the **Belgian wiretapping law** (Article 314bis of the Belgian Penal Code) requires consent for recording private conversations; in commercial settings the case law accepts implicit consent on continuation after disclosure. Second, the EU AI Act **Article 5(1)(f)** prohibits emotion-recognition systems in the workplace; which means using Gong, Chorus, Avoma, Salesloft Conversations, Fathom or similar to infer emotional state of internal sales staff is in the prohibited zone for EU-employed reps. Customer-side analysis (sentiment of the prospect) is permitted but subject to GDPR purpose limitation. Mitigation: configure your conversation intelligence vendor to disable emotion-recognition features for internal participants. Document the configuration. For Belgium specifically, ensure the call opener includes recording disclosure. ## Audit point 5: Cold email under ePrivacy and GDPR The question: under what lawful basis are you sending B2B cold email, and have you documented it? Cold B2B email in the EU rests on legitimate interest (GDPR Article 6(1)(f)) plus the ePrivacy Directive's national transpositions plus per-country B2B carve-outs. The patchwork is uncomfortable but workable. Belgium: Article XII.13 of the Code of Economic Law allows opt-out B2B email to legal persons; opt-in required for natural persons. Practical implication: B2B-targeted cold email to a `companyname.com` address is generally permissible; targeted cold email to a clearly personal address is opt-in. Netherlands: the Telecommunicatiewet allows opt-out B2B with similar logic; an existing customer relationship grants more latitude. France: more permissive for B2B. Germany: stricter; UWG case law treats unsolicited B2B email as unfair commercial practice unless prior business relationship or explicit consent exists. GDPR Article 14 requires source disclosure: where the personal data was not collected from the data subject, the controller must provide the source within a reasonable period (typically the first contact). Mitigation: maintain a per-country cold-email playbook with the lawful basis, the LIA, the source disclosure language, and the suppression list mechanism. The Belgian DPA's guidance "AI Systems and the GDPR" (19 September 2024) is the closest-to-canonical interpretation for AI-driven outbound in Belgium. ## Audit point 6: AI literacy obligation The question: does the staff operating your AI GTM tools have documented AI literacy? Article 4 of the AI Act has been in force since February 2025. Every organisation deploying AI must ensure a sufficient level of AI literacy among the staff who operate it. There is no certification standard yet; the obligation is principle-based. Practical implication: the marketing operations lead who deploys your AI SDR, the RevOps owner of the lead-scoring model, the customer success rep using AI summaries. All need documented AI literacy. The European Commission has signalled that "documented" includes recorded internal training; off-the-shelf certifications from credible providers (Anthropic, Microsoft, IBM, EIT Digital) are a good baseline. Mitigation: stand up an internal AI literacy track, log completion, refresh annually. Falora customers receive a baseline literacy module on deployment. ## Audit point 7: Vendor due diligence The question: do your AI GTM vendor contracts contain the right warranties and information rights? Five contract clauses are now table-stakes for any AI GTM vendor: 1. **AI Act compliance warranty**; vendor warrants the system is conformity-assessed where required, that documentation is available on request, and that conformity is maintained throughout the term. 2. **Subprocessor disclosure**; full list of model providers, data hosting, third-party data sources, with notification rights on changes. 3. **Training-data exclusion**; explicit prohibition on using customer data to train models without separate consent. 4. **Article 50 disclosure tooling**; vendor exposes AI-detection metadata and supports deployer disclosure obligations. 5. **Audit rights**; annual or on-incident audit, with reasonable notice. Mitigation: rebuild your vendor onboarding checklist around these five clauses. Cognism's GDPR-compliant data positioning is the closest market precedent for what credible EU compliance posture looks like; your AI GTM vendors should match or exceed it. ## What the Digital Omnibus changes The Digital Omnibus (COM(2025) 836, 19 November 2025) is the European Commission's package to simplify and reschedule parts of the AI Act and adjacent regulations. The Council adopted its position on 13 March 2026; the Parliament voted favourably with 569 in favour on 26 March 2026; trilogue is ongoing with adoption expected in June 2026. The relevant changes for GTM teams: certain Annex III high-risk categories (notably some employment-adjacent and education-adjacent categories) shift to December 2027; the high-risk classification methodology gains a "significant role in decision-making" filter that may reduce false-positive high-risk classifications; and the obligations on deployers of GPAI models clarify that downstream deployment in standard SaaS contexts does not automatically inherit GPAI-provider obligations. Net effect for B2B GTM: the **August 2026 transparency-and-disclosure deadline holds**. The high-risk perimeter narrows slightly, which is good for most lead-scoring deployments. The vendor due-diligence requirements gain clarity, which makes the contract checklist above more enforceable. ## A 90-minute audit checklist If you have 90 minutes and a notebook, here is the order. Each step takes 10–15 minutes. 1. List every AI tool in your GTM stack. Tag each with its primary function (drafting, scoring, sending, analysing, recording). 2. For each tool, identify the highest-risk action it can take without per-action human approval. 3. Map each tool to the relevant audit point above (1–7). 4. For each audit point, note current state (compliant, gap, unsure) and required mitigation. 5. Flag every tool whose vendor contract is missing one of the five clauses in Audit Point 7. 6. Identify the named owner for each gap and the due date (most should be Q2–Q3 2026). 7. Schedule the AI literacy session for staff identified in Audit Point 6. Charlie Cowan, author of *AI for Revenue Teams*, summarises the European baseline well: > "EU GTM teams that treat compliance as a feature, not a constraint, will out-execute teams that treat it as a checkbox in 18 months." Falora's product position is that compliance-native is faster, not slower, because the audit work is amortised across deployments. ## Conclusion The EU AI Act is not a future risk. It is a current set of obligations with a hard August 2026 deadline for the parts that matter most to B2B GTM. The 7-point audit above is the work for the next two quarters. The vendor contracts are the work for the next renewal cycle. The teams that do this audit early will pay 10–15% premiums on tooling that does not need to be replaced in 12 months, and will avoid the much larger cost of a non-compliant outbound motion that draws regulatory attention. If you want a co-led audit; Falora plus a Belgian law firm partner; we run the 7-point review on real stacks each month. **[Book the EU AI Act compliance review →](https://falora.ai/demo)** --- ### Sources - [Regulation (EU) 2024/1689 on Artificial Intelligence (the EU AI Act)](https://eur-lex.europa.eu/eli/reg/2024/1689/oj) - [EDPB Opinion 28/2024 on AI and GDPR (18 December 2024)](https://edpb.europa.eu/our-work-tools/our-documents/opinion-board-art-64/opinion-282024_en) - [Belgian Data Protection Authority, *AI Systems and the GDPR* (19 September 2024)](https://www.gegevensbeschermingsautoriteit.be) - [BIPT designation as Belgian AI Market Surveillance Authority](https://www.bipt.be) - [Digital Omnibus, COM(2025) 836](https://eur-lex.europa.eu) - [GDPR, Regulation (EU) 2016/679, Article 22](https://gdpr-info.eu/art-22-gdpr/) - [Belgian Code of Economic Law, Article XII.13](https://www.ejustice.just.fgov.be) - [Cognism, *GDPR-compliant B2B data positioning*](https://www.cognism.com/gdpr) ### Related reading on Falora - [The Autonomous GTM Maturity Model](https://falora.ai/blog/autonomous-gtm-maturity-model) - [AI SDRs vs human SDRs: 7 conditions where AI loses](https://falora.ai/blog/ai-sdr-failure-modes) - [Schaalbare B2B groei zonder bureau (NL)](https://falora.ai/blog/schaalbare-b2b-groei-zonder-bureau) - [The anatomy of a GTM engineering system](https://falora.ai/blog/gtm-engineering-system-anatomy) ### About the author Stijn Van Daele is co-founder of Falora and a partner at Stretch Innovation. He works with European B2B scale-ups on autonomous GTM and EU AI Act-conformant deployments. He writes on [LinkedIn](https://www.linkedin.com/in/stijnvandaele). ## FAQ **Is using an AI SDR illegal in Germany?** Not categorically illegal, but Germany's UWG and BGH case law impose stricter B2B consent and transparency requirements than most member states, and the EU AI Act Article 50 disclosure obligation (in force August 2026) requires recipients to know they are interacting with AI. Combined, this makes many current AI SDR deployments non-compliant in Germany without explicit AI disclosure and B2B legitimate-interest documentation. **Do I need to disclose AI in cold emails?** From August 2026, EU AI Act Article 50 requires that AI-generated content (including text generated and sent by AI agents) be detectable as AI-generated, with limited exceptions. For B2B cold email this practically means including an AI-disclosure line and ensuring the technical means of detection (e.g. watermarking metadata) are in place. **Is lead scoring a high-risk AI system under the EU AI Act?** Pure lead-scoring on firmographic and behavioural signals is generally not Annex III high-risk. It can become high-risk if it materially determines access to a service or has legal/significant effect on the individual (e.g. denying credit, employment, or essential services). The boundary is set by the deployer's purpose, not by the technology itself. **What are the EU AI Act fines?** Article 99 of the EU AI Act sets fines up to €35M or 7% of global annual turnover for prohibited-AI violations, up to €15M or 3% for high-risk and transparency violations, and up to €7.5M or 1% for incorrect, incomplete or misleading information to authorities. National competent authorities apply these fines proportionally. **When does the EU AI Act apply to my AI GTM tools?** Prohibited AI practices have been in force since February 2025. AI literacy obligations (Article 4) since the same date. General-Purpose AI obligations since August 2025. Transparency obligations including AI-content and chatbot disclosure (Article 50) from August 2026. Annex III high-risk obligations from August 2026, with the Digital Omnibus likely shifting some categories to December 2027.