AI Is Talking Shoppers Out of Buying: The Solution Starts Upstream of the Chatbot
Summary:
A Semrush survey of 2,338 U.S. adults, reported by MarTech on September 18, found that 57.5% of AI users have decided against a purchase based on chatbot information. About the same share have bought something a chatbot recommended. A separate survey of 1,035 U.S. consumers by Envision Horizons, reported by eMarketer in July, found 50.9% of AI shoppers have decided not to buy because an AI assistant raised concerns. Mixed or negative reviews make most shoppers less likely to buy, paid placement can't control the answer, and the likely growth in AI shopping research makes this a lasting issue. The most practical response starts upstream: fix the operational problems that feed negative answers, publish honest information, and track what chatbots say.
Why it matters:
A chatbot answer can shape a buying decision before a shopper ever reaches the brand's website.
The influence runs both ways. AI recommendations can bring shoppers to a brand or send them to a competitor.
Among AI users, more than 85% are at least somewhat less likely to buy when a chatbot surfaces mixed or negative reviews.
Paid placement earns a spot in the conversation, not a say in the verdict.
Only 37% of brands audit how AI assistants describe them monthly, according to a Publicis Sapient report.
In one agency review, most negative chatbot commentary traced to operational issues, which brands can fix.
Tracking what chatbots say about a brand can be tied to pipeline and closed revenue.
The surveys behind the headline
A new Semrush survey of 2,338 U.S. adults, fielded in July and reported by MarTech on September 18, adds a wrinkle to the AI visibility conversation. Being mentioned in a chatbot answer solves only part of the visibility problem. The same answer can also send a shopper elsewhere.
Among AI users, 57.5% have decided against a purchase based on information from a chatbot. About the same share, 57.5%, have bought something a chatbot recommended.
A separate survey points the same way. Envision Horizons surveyed 1,035 U.S. consumers in February 2026. As eMarketer summarized the results in July, 50.9% of AI shoppers have decided not to buy a product because an AI assistant raised concerns, and 12.1% say it happens often. Both surveys are self-reported, so treat them as an early signal.
These conversations are private. A brand can test the questions shoppers ask, but it never sees the conversation that ended a sale.
The tool that recommends also rejects
Among consumers who named chatbots as one of their sources when buying online, the walk-away figure rises to nearly 81%. That's a narrower segment with a wider margin of error. It's also the group for whom chatbots are already part of the buying routine. Among weekly AI users, almost 70% have decided against a purchase based on AI advice.
Trust helps explain why. About 71% of AI users believe chatbots can recommend the best brands and products, and 59% have discovered a new brand through one. A survey of more than 1,800 consumers across four countries, run by performance marketing firm RTB House, found a similar figure: 59% of U.S. respondents credit AI platforms with highlighting brands they didn't know. BCG's global research found more than 60% of consumers express high trust in GenAI results.
The influence reaches loyal customers too. Envision Horizons found half of AI shoppers (50.1%) have switched from a brand they normally buy because of an AI recommendation at least sometimes. Those are customers a brand already had.
One answer can bring a brand into a shopper's view. The next can send the shopper somewhere else.
A channel that shoppers trust enough to act on will move sales up and down.
Mixed reviews make shoppers hesitate
More than 74% of all consumers said they'd be at least somewhat less likely to buy if a chatbot flagged mixed or negative reviews. Among AI users, that climbs above 85%.
Envision Horizons found 37.7% of AI shoppers use AI to summarize reviews. A mixed review profile used to sit on a product page that a shopper might never scroll to. Now a chatbot can summarize it in a sentence at the exact moment someone is deciding.
Now a chatbot can summarize it in a sentence at the exact moment someone is deciding.
Shoppers ask chatbots about the seller
Nearly half of all consumers (47.5%) at least occasionally ask a chatbot about a seller before buying. Among AI users who research companies this way, 87% would at least occasionally consult a chatbot before hiring a local business. Of those who have done it, nearly half did so within the past week.
That makes the question of what a chatbot says about a brand relevant well beyond national retailers.
Search hasn't gone anywhere
Only 29% of all consumers use chatbots for product research. Among weekly AI users, it's 55%, and for that group chatbots already outrank friends and family, social media, retail sites, and YouTube. Search engines (77%) and reviews (67%) still rank higher.
Some 65% of AI users say chatbots have replaced at least part of their product-related Google searches. Yet nearly half say their overall Google usage for purchase research has held steady over the past year, and 23% say they use it much more.
In practice, a single purchase can pass through search, a review site, the brand's own pages, and a chatbot. Each of those can move the decision.
Ads can't write the answer
Paid placement earns a brand a spot in the conversation. It doesn't decide what the chatbot says once a shopper starts asking pointed questions.
There's a trust cost too. About 42% of consumers dislike chatbot ads, 20% like them, and 38% are neutral. Among those who dislike them, 66% say ads make them doubt the integrity of the AI's answers. Semrush also notes that ChatGPT serves ads only on its Free and Go plans, so buyers on paid plans may never see them.
Why this is likely to keep growing
The surveys are snapshots, so growth is an inference. Several sources support it.
BCG found shopping-related GenAI use grew 35% between February and November 2025. Other BCG research found that among daily GenAI users, the tools were the most influential touchpoint in the purchase journey.
In the Semrush survey, more than 60% of consumers expect to use AI for product research more often within two years if the technology keeps advancing, though that's a self-reported expectation. The heaviest users are also the most willing to act on what they hear, with more than 80% of people who use AI several times a week having bought an AI-recommended product.
There's a counterweight. A Gartner survey of 322 U.S. consumers found willingness to let AI make purchase decisions topped out at 11% in lower-stakes categories. Openness was higher for narrowing choices, at 31% for household supplies and 28% for personal electronics. Gartner's analyst summed up the mood as consumers wanting help with research and comparison while keeping the final decision. RTB House found 44% trust AI tools, compared with 59% who trust friends and family.
Shoppers want to stay in control, and they still use a chatbot summary to shape what they consider. The bigger reason is structural. A chatbot now summarizes a brand's reputation at the moment someone decides. That's a new layer between a company and its buyers, and it's hard to see it going away.
Dissuasion also isn't automatically the problem. If a product has real issues and a chatbot says so, the tool is doing its job. The brand problem is an answer that's wrong, outdated, or built on thin or lopsided sources. That points to two jobs: fix what's true and correct what isn't.
What brands can control
Chatbots build their answers from public sources, and brands can work on those sources.
BCG notes that the models pull brand information from a company's own site, social channels, and third-party publications. That makes the work look like reputation management with better measurement.
Jordan Brannon of Coalition Technologies tested six chatbots with a simple "tell me about" prompt. His initial review covered 17 client brands, and a later analysis expanded the set. Most of the negative commentary was operational, covering turnaround, availability, returns, shipping, and price. It traced back to sources like Reddit, Yelp, Quora, BBB profiles, and Google Business Profile reviews.
Positive commentary often came from the brand's own website. The platforms also differed noticeably in how critical they were. The sample was small and the piece is an opinion column, so treat it as a signal. It still points in a consistent direction.
Audit more than one platform
Try the prompts a shopper would type. What's the best option in this category? What are the alternatives to this brand? What do customers complain about? Record the answers on a schedule, because answers can vary by tool and over time.
Few brands do this regularly. In a Publicis Sapient report on consumer-product companies, only 37% audited how AI assistants describe them monthly, and 25% did so once a year.
Fix what's true
If reviews keep citing the same problem, whether slow shipping, unclear billing, or weak onboarding, it can end up in a chatbot's answer.
No content tactic beats resolving the cause.
Marketing can pass these patterns to operations, product, and customer success as revenue data.
Publish honest answers to the hard questions
Comparison pages, transparent pricing, "best fit and not a fit" content, and pages that address known objections with specifics give chatbots accurate material to work from. Policy language deserves a second look too, since phrases about fees, refunds, and risk can be misread when the wording is ambiguous.
Keep the review record current
A steady flow of recent, detailed reviews and substantive replies gives a chatbot more balanced material to summarize.
Correct what's wrong
Look for outdated pricing on third-party listings, discontinued products still described as current, and resolved issues still being cited. Fix them at the source by updating listings, requesting corrections from publishers, and refreshing the brand's own pages.
Connect it to revenue
Add "Did you use an AI tool while researching us?" to intake forms and sales calls. Review lost deals for objections that match what chatbots are saying. A sales team that knows the objection is coming can address it before the prospect raises it.
Reputation has always influenced sales. Chatbots now put a summary of it in front of the shopper at the point of decision, which makes it a marketing responsibility with a measurable outcome.
A note on the data
Exploding Topics, a Semrush-owned company, ran the Semrush survey in July 2026. The margin of error is plus or minus 2 points for the full sample, and filtered segments like the 81% figure carry wider margins. The results are self-reported, so they show behavior people describe rather than behavior measured. MarTech discloses that Semrush is its parent company, and the original report closes by promoting Semrush's own AI visibility tools.
Envision Horizons is an ecommerce growth agency, and its report includes a checklist for brands. The full report is gated, so the figures here come from eMarketer's summary. The Publicis Sapient figures come from a consultancy's report on consumer-product companies, so treat them as directional.
One gap stands out. No study cited here has tested whether the answers that steer shoppers away from a purchase are accurate. The nearest evidence is a late-2025 Gartner survey of 846 U.S. consumers, in which 54% of recent AI shoppers said they had to double-check the accuracy of everything the tools told them. Gartner's analyst framed accuracy as a brand issue. That captures shoppers' experience, not measured error rates. Until someone tests the answers themselves, the numbers measure influence. They can't say how often the answers are right.
Frequently Asked Questions
How many shoppers has AI talked out of buying something?
Among AI users surveyed by Semrush, 57.5% have decided against a purchase based on chatbot information. The figure rises to nearly 81% among consumers who name chatbots as one of their sources when shopping online. A separate Envision Horizons survey found 50.9% of AI shoppers have walked away from a purchase because an AI assistant raised concerns.
Why do AI chatbots talk people out of buying?
Mostly because they surface mixed or negative reviews, flag operational complaints (shipping, returns, pricing), or answer direct questions about a seller's reliability. More than 85% of AI users say they're at least somewhat less likely to buy once a chatbot points to negative feedback.
Can paying for chatbot ads fix the problem?
No. Paid placement gets a brand into the conversation but doesn't control what the chatbot says when a shopper asks a follow-up question. About 42% of consumers also say they dislike chatbot ads, and 66% of that group say ads make them question the AI's answers.
Do AI shopping habits actually replace Google search?
Not yet. Only 29% of all consumers use chatbots for product research, and search engines and reviews still outrank AI tools even among weekly AI users. Most shoppers use chatbots alongside search rather than instead of it.
How can a brand find out what AI chatbots are saying about it?
Run the prompts a shopper would type (best option in a category, alternatives to a brand, common complaints) across multiple chatbots on a regular schedule. Publicis Sapient found only 37% of consumer-product companies do this monthly, so a routine audit is still a differentiator.
Is the AI shopping trend going to keep growing?
Likely yes. BCG found shopping-related AI use grew 35% in under a year, and heavier AI users are also the most willing to act on what a chatbot tells them. Consumer surveys still show people want to keep the final purchase decision themselves, but the chatbot's role in shaping that decision looks like it's expanding, not shrinking.
Sources:
AI chatbots talked 57.5% of AI users out of buying (Carlos Silva, Semrush, September 7, 2026)
AI is telling consumers not to buy your product (Constantine von Hoffman, MarTech, September 18, 2026)
Brands risk losing shoppers when AI questions product credibility (Grace Harmon, eMarketer, July 15, 2026)
2026 AI Shopping Report (Envision Horizons, survey fielded February 2026)
When Bots Shop: Why Brands Should Stop Ignoring ChatGPT Commerce and Start Designing for it (Simon James and Helen Merriott, Publicis Sapient, Retail TouchPoints, February 3, 2026)
Consumers Trust AI to Buy Better. Brands Need to Move Quickly. (BCG, January 2026)
Consumers warm up to agentic AI purchases (Retail Dive, August 13, 2026)
Gartner Survey Finds Consumers Want AI Shopping Help, But Not AI Purchase Decisions (Gartner, May 27, 2026)
AI Models Surface Negative Brand Commentary: Lessons For Brands (Jordan Brannon, Forbes Business Council, July 1, 2026)
Comments