Forrester: two in three B2B providers will fail AI buyer checks

2 days ago

6 min read

TL;DR: Forrester's 2027 B2B predictions say agentic buying will play a role in 75% of B2B purchases, buyers will double their human conversations to validate AI recommendations, and two in three providers will fail some aspect of AI buyer verification. All three are forecasts. G2 and TrustRadius survey data points the same way: buyers let agents research but rarely buy, and 94% of those who use AI fact-check its answers at least some of the time. The practical response is to list the claims a buyer would most want to verify and make sure each can be confirmed on your site, somewhere you do not control, and by a person on your team.

A buyer who has been given your name by an AI tool still has to decide whether to believe it. Forrester's 2027 B2B go-to-market predictions, published on 21 October, suggest that decision is about to become harder for most providers to win.

Three of the five predictions concern the buyer side: how far agents will go, who buyers will check with, and what happens when claims do not hold up. All three are forecasts, not measurements, so I have set them beside survey data from G2 and TrustRadius to see how far current behaviour supports them. Forrester sells research, and G2 and TrustRadius sell review products, so each has an interest in buyers checking what they are told. Read the figures as direction rather than proof.

Agents will do the groundwork and stop before the order

In its predictions guide (free, registration required), Forrester expects agentic buying to play a role in 75% of B2B purchases in 2027, without saying how "play a role" is measured. The guide describes business buyers using AI agents, either through a procurement system or on their own, for specific, limited jobs: analysing contract renewals, analysing spend, drafting requests for information and proposals (RFIs and RFPs), finding suppliers and gathering information. Fully autonomous buying, it says, will stay concentrated in high-volume, low-risk, repeat purchases, and agents will usually stop just short of placing an order. Its advice to providers is to prepare for buyer agents, meaning AI agents working on a buyer's behalf, and to make sure your systems and sales team can answer them quickly and accurately.

G2's survey of more than 1,000 software buyers, published in July, points the same way. Only 9% would let an agent execute purchases within approved guardrails, and 2% would allow it without pre-approval. A further 47% would let an agent research and recommend while people keep the final decision.

An agent asking questions about your business reads the same information a human buyer reads. The first step in the preparation Forrester describes is making your pricing logic, scope and exclusions plain enough that either reader reaches the same answer. I cover the agent side in my guide to agent experience and the wider change in the B2B buyer journey.

Buyers will want more conversations, not fewer

Forrester's reasoning is that AI can identify and compare vendors in minutes but cannot fully judge fit, feasibility, trust or accountability. It predicts buyers will double their human interactions to validate AI recommendations, seeking conversations with provider teams and with trusted third parties such as analysts, customer references and partners. Its advice is to move from being visible to AI systems towards building a network of trusted experts who can validate claims.

TrustRadius's survey of 1,862 technology buyers, run in January and published in July, adds a complication. 74% of buyers used reviews to inform their purchase decisions, while only 13% used analyst reports, a fall of 63% since 2022. That is a different question from Forrester's, and TrustRadius sells reviews while Forrester sells analyst research. It does suggest that the third parties buyers turn to may be peers more often than analysts. That reading is mine, not a finding from either report.

Whoever the buyer checks with, they arrive at the conversation with questions, and that conversation is where weak proof gets exposed. A page that says "proven results" invites the next question: which results, for whom, and measured how? The answer has to be something your team can point to, not something it improvises. It is why visibility has never been enough on its own, which I argued in why most businesses are solving the wrong AI search problem, and why AI-assisted visitors behave differently on your website.

Forrester's call to build a network of references, partners and analysts goes beyond what a website can do. Those are relationships to build, and they extend your digital footprint beyond your own website. Your website's part in this is to make that network easy to find and check. Name your customers where you have permission and state the results they got, show your work so a prospect knows what to ask about, and give buyers an obvious way to reach a person who can answer follow-up questions.

Two in three providers will fail some aspect of verification

Forrester's Business Buyers' Insights Survey 2026 found that buyers using AI-assisted search and analysis most often use it to gather information about products and vendors. They synthesise provider claims and check facts, which exposes hollow or inconsistent statements. Forrester's 2025 data found that only 40% of B2B decision-makers who contribute to the content lifecycle said their teams were restructuring content so AI tools can interpret and summarise it accurately. It expects buyer fact-checking to grow faster than providers restructure, and concludes that two in three providers will "fail to pass some aspect of AI buyer verification". Its headline says the providers that fail will lose deals. Its detailed text says only that they will fail some aspect of the check, and does not say which aspects or how many deals that costs.

TrustRadius's data shows the checking is already common. 63% of buyers used AI during their purchase journey, and 94% of those fact-checked its answers at least some of the time.

To be included on buyer shortlists, Forrester says providers must share credible, consistent proof points that the AI tools buyers use can find and check. In practice, consistent means the same figure, scope and wording wherever a claim appears: your site, your profiles and your reviews. A buyer or a tool that finds two versions has less reason to trust either.

My earlier posts keep reaching this problem from different directions. Most websites are optimised for exploration rather than confirmation, and a page that cannot confirm its own headline gives a verifying buyer nothing to hold. The 5 second confirmation test is a quick way to check yours.

What to check first

List your main claims, the ones a buyer would most want to verify. For each one, note:

where a buyer can confirm it on your site
where a buyer can confirm it somewhere you do not control, such as reviews, profiles or customer references
who on your team can answer the follow-up question

A claim with a gap in any of those three is where Forrester's forecast would reach you first.

If you would like an outside view of which claims would not hold up, that is what the AI readiness and buyer journey audit covers. Get in touch.

Frequently asked questions

What do Forrester's 2027 B2B predictions say about AI buyers?

Forrester expects agentic buying to play a role in 75% of B2B purchases in 2027, buyers to double their human interactions to validate AI recommendations, and two in three providers to fail some aspect of AI buyer verification. All three are forecasts.

Will AI agents make B2B purchases on their own?

Forrester expects fully autonomous buying to stay limited to high-volume, low-risk, repeat purchases. G2 found 9% of software buyers comfortable letting an agent execute purchases within approved guardrails.

Do B2B buyers fact-check AI recommendations?

In TrustRadius's survey of technology buyers, 94% of those who used AI said they fact-check its answers at least some of the time.

What does failing AI buyer verification mean?

Forrester predicts two in three B2B providers will fail some part of the checks buyers run with AI tools, such as claims that look hollow or inconsistent. Its headline says those providers will lose deals. It does not say how many deals or which claims.

How can a B2B provider prepare for AI buyer checks?

List the claims a buyer would most want to verify. Check that each can be confirmed on your site, confirmed somewhere you do not control, and answered by someone on your team. Keep the same figures and wording everywhere a claim appears.