How to spot fake reviews (and what they do to a rating)

Review fraud is common enough to move a rating by half a star. Here are the patterns that give it away.

Fake reviews are not rare

Estimates vary, but every large review platform runs continuous removal programs, and they publish the numbers. The volume they catch is large enough that the volume they miss is not zero. On a business with a few hundred reviews, a modest cluster of fakes moves the visible average by a meaningful amount.

You do not need to identify every fake. You need to notice when a rating is being propped up.

The patterns worth knowing

Bursts. Genuine reviews trickle. Fake ones arrive in clusters, because they are bought in batches. A month with ten times the usual volume, nearly all five-star, is the clearest signal there is.

Vagueness at length. Real positive reviews are specific and often short: "the lamb was excellent". Fabricated ones are long, enthusiastic and strangely contentless, praising "the atmosphere and the wonderful staff" without naming a single thing.

Reviewer histories that make no sense. One review ever, posted the week the account was made. Or an account reviewing businesses in six cities in four days.

Symmetrical negatives. Fraud is not only inflationary. A sudden run of one-star reviews with no detail, often mentioning a competitor by name, is the same phenomenon pointed the other way.

Owner responses that do not match. When an owner replies to a glowing review with a generic thank-you but engages in detail with the critical ones, the critical ones are usually the real ones.

What this means for reading a rating

Once you know fakes cluster, the average becomes less interesting than the distribution over time. A place whose recent reviews look nothing like its older ones is telling you something, whether the cause is fraud or a genuine change in management.

How RevuIQ helps

RevuIQ reads the review text and the dates, not just the ratings, and its verdict quotes the reviews it draws on, so vague praise is in front of you rather than folded into an average. It does not label reviews as fake. If a rating looks propped up, ask it: a follow-up such as "do any of these reviews look suspicious?" is answered from the reviews themselves, dates included.