Nearly three out of four consumers—74%—think retailers collect too much personal information, according to a new report from ThoughtSpot, a vendor that sells data analytics software to retailers and other businesses.

The survey, fielded in partnership with YouGov, polled 4,833 adults in the US and UK. It asked respondents to choose up to two entities they distrust most with their personal data. Retailers ranked third at 30.8%, behind tech companies (56.4%) and governmental bodies (41.8%). Retailers fared worse than banks and financial services (19.9%), healthcare providers (16.7%) and small or local businesses (14%).

For marketers, the report's sharpest finding concerns where consumers draw the line on personalization. "Personalization built on explicit information customers have knowingly provided feels like service," Miya Knights, a ThoughtSpot retail technology analyst, wrote in the report's foreword. But "personalization inferred from implicit data, such as tracked behavior, location, or prediction, feels like surveillance."

The numbers back her up. Only 27.8% of respondents objected to retailers using information they had knowingly provided. That is less than half the 57.6% who objected to retailers using behavior tracked from other websites or apps.

The consent gap is the story. Shoppers will accept personalization when they understand its inputs; they reject it when the inputs are invisible. First-party data collected with clear permission, in other words, carries a trust premium that third-party or inferred data does not.

AI recommendations face an even steeper trust deficit. Only 13% of respondents said they trust AI-driven recommendations. Another 47% said they distrust them, while 34% were neutral.

A June report from AI shopping platform Product.ai, based on a survey of 1,463 US online shoppers, points to how consumers actually behave despite that distrust. They use AI research and recommendations as one component of purchase decisions, not the sole source. Among shoppers who had used AI to research products, 86% said they checked other sources before buying, and 45% said they always do so.

Taken together, the two reports suggest AI can influence the path to purchase but cannot close it. Consumers treat algorithmic input as a starting point and verify it against sources they trust more.

For retailers and the marketers who serve them, the implication is direct: data practices built on explicit consent and transparent AI recommendation systems are no longer just compliance questions. They determine whether personalization reads as service or as surveillance—and that perception shapes whether shoppers engage at all.