AI chatbots talk shoppers out of purchases nearly as often as they drive them: 57.5% of AI users have decided not to buy a product based on information from a chatbot, while 65.01% have at least partially replaced product-related Google searches with AI conversations. Those figures come from Exploding Topics' largest survey to date — 2,338 US adults surveyed in July 2026, with a margin of error of plus or minus 2 percentage points at a 95% confidence level for full-sample statistics.

The numbers sharpen considerably among heavier users. Among consumers who use AI at least weekly, 73.58% have bought a product based on an organic AI recommendation, 81.24% believe chatbots can recommend the best brands and products, and 69.7% have declined a purchase on the basis of AI advice. Among "power users" — those using chatbots multiple times per week — 80.45% have bought an AI-recommended product at least once, with 54.95% doing so multiple times. Of consumers who specifically named AI chatbots as a purchase research source, 80.93% have decided against a purchase on AI guidance, and 42.81% have done so multiple times.

Review signals carry particular weight. 74.15% of all consumers would be at least somewhat less likely to buy a product if a chatbot flagged mixed or negative reviews. Among current AI users, that figure climbs to 85.31%.

Chatbots as a discovery channel

The survey also documents AI's role as a brand-discovery engine. 59.27% of AI users have discovered a brand or product they were previously unaware of through a chatbot recommendation, and 25.79% report doing so "frequently." Among consumers actively using AI for product research, only 9.38% have never discovered a new brand from a chatbot. Across all consumers, 43.14% have discovered a new brand through AI — a figure Exploding Topics notes aligns with Semrush's March buyer journey study, which put the same number at 43%.

Consumers also run pre-purchase background checks through AI. 47.54% of all consumers at least occasionally consult a chatbot for seller information before buying; excluding non-users, 65.42% do so, and 32.68% of those do it frequently. Nearly half (48.97%) of consumers who have researched a company with AI did so within the last week, with another 20.3% within the past two weeks.

Local businesses face the same dynamic. 87.41% of people who use AI for company research would at least occasionally consult AI before hiring a local business, and 31.48% would always do so. Notably, 42.01% of these users think AI gives better advice on local businesses than on large companies, with a further 43.7% saying it performs equally well for both.

Adoption and the Google question

These are survey self-reports, not clickstream data — but they show broad adoption. 33.23% of respondents use chatbots like ChatGPT, Gemini, Claude, and Perplexity multiple times per week, and another 15.41% weekly. Non-usage clusters among the over-60s (60.05% never use chatbots); only 13.61% of consumers aged 18-60 report never using them. Among weekly AI users, chatbots rank as a product research source ahead of friends and family (49.28%), social media (48.38%), retail websites (44.42%), and YouTube (43.35%), with 55.13% using AI for purchase information — though search engines and reviews still lead.

On Google specifically: more than 1 in 5 AI users have completely replaced product-related Google searches with chatbots. The 45-60 cohort leads, with 39.04% reporting complete replacement, while 63.36% of AI users over 60 have replaced none. Overall Google usage has not declined, though — 47.6% of AI users maintained their Google level over the past 12 months and 23.01% used it more — suggesting chatbots complement search rather than replace it. Excluding non-users, 12.72% of consumers now start purchasing journeys with AI chatbots, third behind search engines (29.89%) and online reviews (28.13%). Among upper-middle-income AI users ($125,000-$199,999 household income), 25.12% start with chatbots, nearly matching reviews (24.13%).

Ads underperform organic mentions

Chatbot advertising, active in ChatGPT in the US since February with a no-minimum-spend self-serve beta launching in early May, faces measurable skepticism. 41.63% of consumers dislike chatbot ads, and 66.05% of those say ads make them doubt the integrity of AI responses — integrity concerns being the top complaint across all age groups, rising from 59.13% among 18-29-year-olds to 68.86% among those 60+. Among all AI users, 71.63% feel neutral or negative toward chatbot ads.

Brand impact tilts slightly negative: 23.12% of consumers would form a worse opinion of a product after seeing a chatbot ad, versus 20.36% a better one; 56.52% report neutral or content-dependent impact. Self-reported click preference favors Google ads overall (45.42% vs. 24.88%), though consumers who have actually seen chatbot ads flip: 48.94% say they would more likely click a ChatGPT ad, versus 30.64% for Google. A structural limitation remains — ChatGPT serves ads only on Free and Go plans, leaving Plus, Pro, and business accounts ad-free, which excludes much of the B2B audience entirely.

Sentiment has also moved quickly: an April Exploding Topics survey, run while chatbot ads were in limited testing, found 48.35% of consumers expected shopping features and ads to improve AI responses — optimism that has given way to skepticism after months of wider rollout.

The GEO takeaway

For search marketers, the survey's central argument is that Generative Engine Optimization is now defensive, not optional: brands that ignore their AI presence risk having purchases actively intercepted. More than 60% of respondents expect to use AI for product research more frequently within two years. What to monitor next: whether chatbot ad clickthrough converges with Google ads as exposure grows, whether trust erosion from paid placement deepens, and how share-of-voice in organic AI recommendations develops as discovery shifts away from classic SERPs.