Most personalization programs are being graded on the wrong scoreboard. They are winning on click-through rate and conversion lift while quietly losing the one asset that makes either number durable: whether the customer trusts what the brand is doing with their data. I think that trade is a mistake, and a new consumer survey gives the clearest evidence yet that the industry is making it at scale.
The Data Marketers Would Rather Not Read
Thoughtspot’s Retail AI Trust Report, published this month from a survey of more than 4,800 consumers across the US and UK, is not subtle. Ninety-one percent of respondents said at least one common personalization tactic feels intrusive. Tracking behavior across other sites and apps topped the list, followed closely by being recommended the same products too often and having real-time location used to target a message. Seventy-four percent said retailers collect too much personal data outright. And when it comes to the AI-driven recommendations that most personalization budgets now flow toward, only 13 percent of consumers say they trust them, while 47 percent actively distrust them and 66 percent say an AI system has flat-out misunderstood what they wanted.
None of that is a minor image problem. Fifty-nine percent of consumers say they are unlikely to keep shopping with a retailer whose AI repeatedly suggests the wrong price or product. That is a retention metric, not a sentiment score, and it sits directly downstream of the personalization stack most marketing teams spent the last two years building.
The Counterargument, and Why It Does Not Hold
The obvious defense is that personalization measurably works: aggregate conversion and average order value do rise when messages are tailored, and some friction is simply the price of relevance. That defense was true for a long time. But the same report undercuts it in a specific way: stacking more data dimensions into a personalization model does not produce a proportional lift in perceived relevance, it produces a disproportionate rise in perceived creepiness, with content built on ten data points rated as far more unsettling than content built on one. Marketing teams are still being rewarded internally for adding the eleventh signal, when the report’s own numbers say the return on that signal, in trust terms, is already negative.
Put plainly, the industry is optimizing personalization for the metric that is easiest to report to a CMO in a weekly dashboard, and ignoring the one that determines whether the customer relationship survives the next data breach headline or regulatory inquiry.
What Would Actually Move the Trust Number
The report is unusually specific about what would change consumer behavior, which is more useful than most surveys in this category. Roughly a third of respondents pointed to data transparency, a similar share pointed to real opt-out controls, and a similar share again pointed to explainability, being told why a recommendation was made. As retail technology analyst Miya Knights put it in the report, “The undecided third of consumers is the commercial prize, and it will be won by retailers that treat transparency and control as product features rather than policy pages.”
That framing matters because it reclassifies transparency and opt-out tooling from a legal compliance cost, the way most organizations currently treat them, into a growth lever with its own measurable audience. Marketers already know how to build and market a product feature. Very few are currently building “why am I seeing this” or “turn this off” as one, even though the consumer research says that is where the next real gain sits, not in a twelfth data input to the recommendation model. This is not an unrelated concern to the trust collapse this publication has already tracked in how consumers are losing confidence in AI search results and in how little marketers themselves trust the data setting their own budgets. The pattern across all three is the same: the industry keeps adding automated decisioning layers faster than it builds the transparency to make people, customers and marketers alike, comfortable trusting them.
The Actual Fix
Run a trust audit alongside the next personalization audit, and put it on the same dashboard as conversion lift. Track the rate of intrusive-signal complaints, unsubscribe and opt-out requests tied to specific personalization tactics, not just aggregate email churn. Then take the budget that would have funded one more third-party data signal and spend it instead on a visible opt-out control and a plain-language explanation of why a customer is seeing what they are seeing. The report suggests that is the trade a meaningful share of the market is actively waiting for a retailer to make first.
Source: Thoughtspot
