AI boosts discovery but reviews still seal the deal, new research shows
This holiday season, shoppers aren’t blindly trusting AI assistants to do their shopping on their behalf.
According to a new Feedviser survey, ChatGPT is where 16% of shoppers start looking for products. However, only 9% go there to actually buy something. And when an AI assistant recommends a product, only 5% of respondents say they buy it on the spot, with the rest checking other platforms like Google and Amazon first for better prices.
For experts, this customer behavior signals that AI recommendations are a starting point, not the final word. Shoppers still check reviews, compare prices, visit retailer and brand sites, and look at shipping and return policies, for example, before deciding. The challenge for brands is to ensure these touchpoints reflect an accurate image of their offerings and ultimately convert customers.
PartnerCentric’s latest survey-based findings show that, following an AI suggestion, 33% of shoppers go to check existing reviews, 22.3% compare prices, and 18.8% search Amazon or another retailer. Only 14.8% visit the brand’s website, while 7.6% look for creator content about the product. When it comes to brands’ holiday marketing reach, 33% of shoppers say shipping and return policies often or always change what they buy.
“AI is a starting point, not a final verdict,” said Stephanie Harris, founder and CEO of PartnerCentric. “The shoppers who trust it most are still checking the work.”
In PartnerCentric’s data, 42.2% of shoppers in the top two AI-trust categories still want reviews before acting, while 31.2% go to actually check the reviews right away. “So the recommendation has to survive the next question the shopper asks,” Harris said. “And that question usually gets answered in the reviews, on the retailer page, or in the shipping and return terms.”
Kieran Powell, executive vp at Channel V Media, which released its The State of AI Adoption in America report this month, agreed that today AI is winning the research step of the customer journey.
As such, Powell said retailers should think of AI answer engines as a replacement for the Google search bar and blue recommendation links. “That is how consumers are most comfortable with AI recommendations right now,” said Powell. “The rest of the shopper journey still exists.”
Powell pointed to a recent survey by Prosper Insights & Analytics showing that consumers are using AI to shortlist products, but not to make their purchase decisions.
For example, 12.1% would let an AI agent find and book a restaurant on their behalf, 11.0% to buy groceries, 10.0% to run the smart home and 9.8% to pay their bills. Meanwhile, only 7.3% would let it buy apparel. “If people won’t let AI buy a $40 shirt for them, they won’t let it tell them which shirt is good,” Powell said.
Brands are increasingly realizing that being mentioned by AI assistants isn’t enough to get those customers to click purchase.
Musely, which offers prescription beauty and health treatments through its online platform, is one brand noticing the need to meet increased AI search demand with accurate product information and real reviews to convert potential customers. Musely has seen traffic from AI assistants grow steadily since 2025, so it’s currently focused on its GEO strategy to boost brand visibility and provide accurate information.
Paris Roche, head of strategic partnerships at Musely, said AI is becoming an increasingly important part of how consumers research healthcare, especially when they’re trying to understand ingredients, compare treatment options, read customer reviews or determine whether a brand is credible.
“For Musely, trust is especially important,” she said. Given that the company offers prescription medications across beauty and health, Roche said, “It’s critical that prospective patients have access to information that is not only easy to find, but also accurate and responsible.”
AI tools are making it easier for people to educate themselves before making a decision, Roche said.
“For us, the goal isn’t simply to be mentioned by AI,” Roche said. “It’s to make sure that when someone asks about Musely’s ingredients or one of our prescription treatments, the information being surfaced is clear, useful and accurate.”
With the upcoming holiday shopping rush, Harris said brands will still have to earn the recommendation. That’s because the AI assistants are reading the same ecosystem shoppers use to verify a product’s info, including reviews, product pages, retailer listings, creator content, pricing, shipping and returns.
In turn, Harris said the brands that go into the peak season with consistent proof of strategy already in place will be easier for AI to recommend and easier for shoppers to actually buy. “When those signals conflict, AI does not smooth it over,” she said. “It surfaces the gap sooner.”