Most US B2B buyers who use AI say it shaped their shortlist. What the survey can and can't tell you

What did Semrush find?

Semrush found that most US B2B buyers who use AI say it now plays a part in choosing vendors. Its survey of 622 US B2B professionals in March and April 2026, published on 8 July, found 519 who use AI tools at work. Of those, 66% regularly use AI to research vendors and 92% say AI has shaped their vendor shortlist.¹ In the same group, 75% said they fully or mostly trust AI vendor recommendations.¹

Semrush sells SEO and AI visibility software, so it has an interest in the finding. That does not make the numbers wrong, and the method is stated: 643 people surveyed, 21 removed in quality checks.¹ It does mean the questions deserve a careful read, starting with the base: every percentage is a share of the respondents who already use AI at work, not of all B2B buyers.

The detail is more useful than the headline. Asked what they do after an AI recommendation, 71% said they visit vendor websites, 63% run a Google search and 38% check G2 reviews.¹ ChatGPT was the most used tool for product research at 71%, followed by Google Gemini at 61%.¹

What the survey can't tell you

The survey cannot tell you how much AI changed any decision, because every figure is self-reported. "AI shaped my shortlist" can mean the shortlist came from ChatGPT, or that ChatGPT confirmed a list the buyer already had. Both answers count the same in a percentage.

People also describe their research by the tools they remember using, and AI is the new, memorable tool. A survey asked in 2026 about AI will find AI. That is a reason to read the figures as direction, not as a measure of influence on revenue.

Our view: treat surveys like this as evidence that buyers are in AI answers, and measure the effect on your own pipeline separately. Our guide to proving AI visibility in pipeline sets out the signals that do hold up, such as self-reported attribution fields and branded search lift.

How it squares with G2 and Gartner

Other buyer surveys point the same way, with their own caveats. G2, which runs a software review marketplace, surveyed 1,076 B2B decision makers in North America, EMEA and APAC in March 2026. It found 51% start software research in an AI chatbot more often than in Google.² In the same survey, 69% said they had chosen a different vendor than expected after information from an AI chatbot.³

Gartner adds the check that matters. In its survey of 645 B2B buyers from August to September 2025, data now close to a year old, 69% said they turn to sales reps to validate AI-generated insights.⁴ Buyers use AI to build the list, then check it with people and with vendors' own material.

All three are self-reported surveys by companies that sell to the vendors they describe.

Behaviour data tells a quieter story about clicks. Our note on ChatGPT mentions and visits covers evidence that brands shown in AI answers gain visits mostly through Google search afterwards. That fits Semrush's finding that buyers search after an AI recommendation.

Three pages a SaaS vendor should check first

If buyers go from an AI answer to your website, the pages they land on decide whether the shortlist holds. Three do most of the work.

1. Pricing. AI answers often state a price, right or wrong. Publish plans and prices as text, with what each includes and the date. If pricing is on request, say what drives it and give a typical range where you can.
2. Comparisons. Buyers ask "X vs Y" and "alternatives to X". A fair comparison page with named alternatives, written for the buyer rather than to win every row, gives engines and buyers a source to quote.
3. Integrations and security. Shortlists fail on "does it work with our stack" and "will security approve it". List integrations by name, and state certifications and data residency plainly.

Then check what engines say on the same topics. Run the prompts a buyer would use, in each country you sell to, on several days. At Sigzen AI, our AI answer monitor covers six engines (ChatGPT, Gemini, Perplexity, Claude, Copilot and Google AI), and audits run each prompt on three separate days.

Our view: the survey is a reason to fix those three pages this quarter, whatever you think of its percentages. They serve buyers who arrive from AI, from search and from a colleague's recommendation alike.