When it comes to consumers using AI for shopping, it all boils down to a matter of trust.
That’s according to a consumer survey from ThoughtSpot, the AI analytics company, titled “Trust, Tested: What consumers really think about retail personalization in the AI era.”
The research was based on a survey of more than 4,800 consumers in the U.S. and U.K. and conducted in partnership with YouGov. The findings were clear-cut. “Consumers are not rejecting AI outright; they are rejecting AI they cannot see, question or turn off,” the report’s authors said.
“That distinction runs through the whole study,” they noted. “Seventy-four percent of consumers believe retailers already collect too much personal data before a single AI feature is switched on. Consequently, every recommendation engine and ‘just for you’ banner faces a skeptical audience from the start.”
The findings showed that just 13 percent of shoppers polled trust AI to tell them what to buy, while 47 percent said they distrust it outright. “Furthermore, 91 percent can point to a specific form of personalization—such as cross-app tracking, location data or an overly perceptive suggestion—that feels less like service and more like surveillance,” the report stated.
The survey showed that 74 percent of those polled said retailers collect too much personal data. “Every AI feature is judged against this baseline, not in a vacuum,” the report’s authors said, adding that 91 percent of respondents find AI intrusive. “Consumers name at least one personalization tactic that feels intrusive, led by cross-app tracking (57.6 percent) and real-time location tracking (48.8 percent). Features retailers count as innovation often register as overreach.”
The report also found that an AI retail model is only as trustworthy as the data it is built upon. When a bad recommendation is generated, or an incorrect price, shoppers see it as a broken commitment instead of a technical bug. The researchers said this explains the 34 percent of consumers who are undecided about AI.
“While the data does not explicitly state why they are on the fence, it highlights what would move them: data transparency, clear explanations for recommendations and easy opt-out controls,” the report stated. “Consumers are not demanding a smarter algorithm; they are demanding an accountable one.”
ThoughtSpot said it can address this gap via three core capabilities. This includes search-driven analytics that allow retail teams to query verified, first-party data in natural language. This grounds recommendations in actual customer behavior rather than model guesswork. The second is unified customer, inventory and supply chain data that can “ensure recommendations are tied directly to the retailer’s ability to deliver.” And lastly, full data lineage that enables retailers and brands to explain why a recommendation was made. This fulfills a key demand for nearly 30 percent of consumers.
“The AI era in retail will come down to a simple test: not how much data a retailer collects, but whether they can explain in plain language how it was used and whether they kept their word,” the report concluded.
