A five-star rating is useless if someone holds the scale.

In March 2026, the UK Competition and Markets Authority opened consumer-law investigations into Autotrader, Dignity, Feefo, Just Eat and Pasta Evangelists. The regulator said it would examine concerns ranging from omitted one-star reviews and inflated ratings to staff-written reviews and undisclosed incentives. These are investigations, not findings of wrongdoing.

The useful point is structural. A rating is a piece of purchase infrastructure. It decides which restaurant, service or seller is even considered. If platforms choose which evidence survives, then moderation policy is no longer back-office housekeeping—it is part of the price comparison.

Trust needs provenance

Consumers should be able to tell whether a review was verified, incentivised, removed or folded into a rating for another product. A score without that context looks objective while quietly concealing the decisions that produced it.

The best review systems keep the messy evidence visible: low scores, clear labels, moderation reasons and a separate trail for paid endorsements. Trust does not come from a clean star average. It comes from knowing what has not been cleaned away.