TL;DR

Retailers have spent two years building AI recommendation engines. British shoppers have not come with them: just 10.6% place any trust in an AI suggestion at the point of purchase, and 53% distrust it outright. Two-thirds report the technology getting their needs wrong at some point. The research comes from ThoughtSpot, via YouGov, covering 2,216 UK adults.

Where the objection actually lies

It is not automation as such. Nearly a third back AI personalisation when it surfaces a better price, and 19.4% pick time saved as their priority. Shoppers accept the tools when the purpose is legible and the payoff arrives now.

The line gets drawn at tracking. When behaviour is followed onto other websites and apps, 62% call it too intrusive — and the identical proportion objects to real-time location tracking. Some form of retail personalisation strikes 94% as intrusive. Three-quarters, 76%, think retailers gather too much personal data, against 4.1% who disagree.

What shoppers ask for instead

The remedies named are unglamorous. Data transparency was chosen by 35.2%, straightforward opt-out controls by 34.8%, and plain explanations of why something was recommended by 31%. None of that requires better models. It requires telling people what is happening.

There is commercial force behind the preference. Some 63% said they would probably stop shopping with a retailer whose AI kept pushing products beyond their price range — a rate close to the 59% who would walk after a badly late delivery. Shoppers are judging recommendation quality by the same standard as fulfilment, which is a harsher bar than most retail technology roadmaps assume.

Looking forward

A neutral bloc of 35.2% told the researchers that transparency would raise their trust — a bloc larger than the committed enthusiasts, and reachable without building anything new.

This is the second UK consumer study in a week pointing the same direction. Last Thursday we covered research finding shoppers would let an agent spend £149 on their behalf and no more. Together they sketch a market that will delegate small, well-understood purchases and nothing else. As ThoughtSpot’s Mayank Sinha put it, a reliable customer-facing agent cannot be built “on fragmented inventory or opaque tracking” — the constraint on agentic retail in Britain is not model capability but the back-end and the consent behind it.