Go-to-market

Semrush asked 2,338 US consumers about AI shopping, and 57.5 percent of AI users have decided not to buy something because of a chatbot

September 7, 2026 at 4:20 AM PT

Chart from the Semrush survey on AI chatbots and buying recommendations

Image: Semrush

Why it mattersChatbots now filter purchase decisions before buyers reach a website, so what answer engines say about a product carries direct revenue consequences alongside search rankings.

Semrush published its largest consumer survey to date on 7 September: 2,338 US adults, surveyed in July 2026 under the company's Exploding Topics research brand, on how AI chatbots figure in their buying decisions. The finding Semrush leads with: 57.5 percent of AI users have decided not to buy something based on information a chatbot gave them.

The buying side is already established. Semrush reports that 41 percent of all consumers have made a purchase based explicitly on a chatbot recommendation. Among people who use AI at least occasionally the figure is 57.51 percent, and among those who use it multiple times per week it reaches 80.45 percent, with 54.95 percent of that group buying on AI advice more than once.

The survey's more useful contribution is measuring the blocking side. Beyond the 57.5 percent headline figure, Semrush finds that 80.93 percent of consumers who name chatbots as one of their purchase-research sources have decided against a purchase on AI guidance. Review signals carry weight here: 74.15 percent of all consumers say they would be less likely to buy a product if a chatbot flagged mixed or negative reviews, and among AI users that rises to 85.31 percent.

The search shift behind all of this has a number too: 65.01 percent of AI users say they have at least partially replaced Google searches with chatbots for product research.

Ads inside chatbots test badly

Semrush also asked about paid placement. 41.63 percent of consumers dislike ads in AI chatbots, against 20.28 percent who like them and 38.09 percent who are neutral. Among the people who dislike them, 66.05 percent say an ad makes them doubt the integrity of the AI's whole response, and that concern leads every age group.

All of these figures are Semrush's own. The company states a margin of error of plus or minus 2 percentage points at a 95 percent confidence level for full-sample statistics, with wider margins on filtered segments, and it publishes its segment definitions. Semrush also sells AI visibility tooling, and the report closes with a pitch for it, so the survey supports the company's own product line. No independent party has checked the data.

For a team that ships software or runs a commercial site, the survey puts numbers on where the answer layer now sits: between the product and the buyer, recommending in one direction and rejecting in the other. Monitoring what ChatGPT, Perplexity and Google's AI surfaces say about a product, and which reviews they summarise, is revenue work of the same kind as watching search rankings. The ad figures matter for budget decisions too: two thirds of the people who dislike chatbot ads say an ad makes them doubt the whole response, which favours earning organic citations over buying placement.

Source

Primary source: AI chatbots talked 57.5% of AI users out of buying, Semrush blog, by Carlos Silva, 7 September 2026.

Source: Semrush

This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.

More from AI News

95% of ChatGPT users still use Google, but the same households search less and click less

Matt Southern collects three datasets in Search Engine Journal that all point the same way: Google keeps its audience, the same households run fewer queries after gaining access to ChatGPT Search, and forced AI Mode cuts external clicks by 18.8 percentage points.

Source: PressGo-to-market

SEJ column argues no content audit can catch what an AI model invents about your company

Duane Forrester argues in Search Engine Journal that when an AI model has no evidence about a company, it produces a confident description of the nearest well-documented neighbour, and no content audit run against your own pages can find it.

Source: PressGo-to-market

ChatGPT.com passed 1 billion U.S. monthly visits and Bing lost half its traffic, Semrush data shows

Search Engine Land's Andy Chadwick pulled Semrush Traffic Analytics data for the top U.S. websites in July 2026, and found ChatGPT.com up 48.38% year over year to 1.09 billion monthly visits (rank 9), Bing down 50.43%, and enough attribution loss inside GA4 that a publisher's own numbers may no longer explain themselves.

Source: PressGo-to-market