What the StackAdapt AI Vendor Research Found
StackAdapt and B2B agency Ledger Bennett published new research on July 27, 2026 showing that 77% of B2B marketers believe AI-related questions are not asked rigorously enough during vendor evaluations. The study, run with research partner NewtonX, surveyed 426 B2B marketers across the US, UK, and Asia-Pacific. It shows most marketers know their AI vetting process is weak, even as AI claims flood their vendor pitches.
Key Numbers From the StackAdapt B2B RFP Study
The gap between AI hype and real evaluation runs through the whole report. Over 60% of marketers say their request for proposal process cannot separate meaningful AI capability from marketing hype. Only 23% of B2B marketers surveyed assess AI capability using defined evaluation criteria, meaning most buyers are judging AI vendor claims on gut feeling rather than a checklist.
The measurement side looks just as shaky. Sixty-three percent of marketers say they cannot effectively measure performance across marketing channels, despite having clear KPIs in place. StackAdapt’s VP of Enterprise said the industry has increasingly confused what is easy to measure with what actually matters, and that mindset shapes RFPs, media plans, and vendor scorecards alike.
Who Ran the StackAdapt AI Vendor Evaluation Research
StackAdapt, an AI-driven advertising and orchestration platform, and Ledger Bennett, a B2B agency, commissioned the research. NewtonX, a B2B research firm, ran the survey of 426 marketers across the US, UK, Australia, Singapore, Thailand, and Indonesia. Demand Gen Report published the findings on July 27, 2026.
Why AI Vendor Claims Are Slipping Through B2B RFPs
Vendors know AI sells right now, and B2B marketers are telling researchers they cannot reliably tell a real capability from a slide deck. Marketing buyers already apply this same skepticism to overhyped stats everywhere else in their jobs. Start Some Shift covered a study co-authored by Perplexity that praised Perplexity’s own AI agents, built on data the vendor controlled and comparisons the vendor chose. The CMO Council recently found only 1 in 4 marketing leaders call their martech stack highly advanced, with 34% describing fragmented, patchwork systems the report calls “Frankenstack” setups. And TrustRadius found that 63% of B2B buyers use AI to research purchases, but 94% of them fact-check what it tells them before trusting it. Buyers are learning to doubt AI claims everywhere except, it seems, inside their own RFP process.
Talking Shift: Your RFP Is the Weakest Link in AI Vetting
Marketing teams have gotten sharp about spotting AI hype in a vendor’s sales deck. Then the same team writes an RFP that never asks the vendor to prove the AI claim at all. Start Some Shift’s take: an RFP without defined AI evaluation criteria is not neutral, it is a pass for whichever vendor pitches the loudest AI story. Only 23% of marketers in this study grade AI claims against a real standard. The other 77% are buying on trust they would never extend to a press release. Build the AI questions into the RFP template itself, or the vendor with the best marketing wins the deal before the tool is ever tested.
What B2B Marketers Should Do When Vetting AI Vendors
- Add a required AI capability section to every RFP template, with specific questions instead of open-ended claims.
- Ask vendors for a live, unscripted demo of the AI feature on your own data before signing.
- Score every AI claim against defined criteria, not the vendor’s own case studies.
- Require vendors to disclose what their AI was trained on and what tasks it was tested against.
- Involve a technical reviewer in AI evaluation, not just the marketing or procurement lead.
- Ask for a reference client using the AI feature in a use case close to your own.
- Set a 90-day review after purchase to confirm the AI performs as scored, not just as pitched.
What to Watch Next in B2B AI Vendor Evaluation
Watch whether more agencies and platforms publish standardized AI evaluation criteria that marketers can borrow for their own RFPs. Also watch whether procurement teams start requiring proof-of-concept testing before AI vendor contracts close, rather than after.
