Fake reviews work because they look like all the others. They sit in the same list, wear the same stars, and ask you to trust them for the two seconds it takes to decide whether a restaurant, a plumber or a £400 gadget is worth your money.
The good news is that fabricated reviews are written under constraints real ones are not. The writer was never in the shop, never used the product, and usually needs to produce a lot of reviews quickly. Those constraints leave marks. Once you know the marks, you see them everywhere.
Here is what to look for, and, just as importantly, what each sign does and does not prove.
1. It never says anything specific
This is the most reliable sign there is. A real customer names things: the dish, the fitter who turned up, the day the delivery went missing, the exact feature that irritated them. They have details because they were there.
A fabricated review stays abstract. "Great service, highly recommend, will use again" could be pasted onto any business in the country without changing a word. Ask yourself whether the review could be moved to a competitor's page untouched. If it could, it is telling you nothing about this business, and it may not be about this business at all.
2. The sentiment has no middle
Genuine reviews are usually mixed, even the happy ones. People say the food was excellent but the table was cramped, or the tool does the job but the app is clunky. Real experiences have texture.
Fakes tend to sit at the extremes: flawless praise with no reservation at all, or total condemnation with nothing redeeming. Both are written to move a rating, not to describe an afternoon.
3. It reads like an advert
Watch for reviews that repeat the full business name, as in "I would recommend Anytown Family Dental Practice to anyone looking for a dentist in Anytown". Nobody talks about a place they visited that way. That phrasing exists because whoever wrote it was thinking about search terms.
The same goes for marketing vocabulary. Real customers say the room was clean. Paid ones say the establishment offers an unparalleled level of service.
4. The reviewer's history is thin, or impossible
Click the reviewer's name. Two profiles should worry you.
The first has exactly one review, posted the same week the account was created, and nothing else. The second has dozens, spread across restaurants, locksmiths and car dealerships in cities hundreds of miles apart, all glowing, all posted in the same fortnight. One person did not have that month.
5. Several arrive at once
A genuine business collects reviews at a fairly steady drip, because customers arrive at a fairly steady drip. A sudden cluster is unusual, and worth reading as a group rather than one at a time.
Look at the dates. Five five-star reviews in a single afternoon, from accounts with little history, following a run of angry one-star reviews, is not five customers. It is one purchase.
6. The wording is too clean
Real reviews are a bit of a mess. People type on phones, they miss apostrophes, they trail off, they use a word twice because they were annoyed.
Text that is uniformly polished, with sentences of similar length, correct punctuation throughout and not a single personal aside, is worth a second look. This is now the most common tell in AI-generated reviews. It is not proof, because plenty of genuine customers write neatly, and some now use AI to draft their reviews about experiences that really happened. It is a reason to keep reading, not a conclusion.
7. Different people, identical phrasing
Fake review campaigns are produced at volume, often from a template or a single prompt, so the output repeats itself. If three reviewers who have never met all describe the staff as "very professional and highly attentive", you are reading one author with three accounts.
8. It mentions a reward
A review that says the writer received a free sample, a discount, a gift card or an entry into a prize draw is an incentivised review. That is a different thing from an invented one, and it is not automatically against the rules, but it must be disclosed, and platforms including Google prohibit reviews exchanged for payment or discounts. An undisclosed incentive tells you the rating was bought, whatever the experience behind it.
9. The punctuation shouts
Rows of exclamation marks, capital letters used for emphasis, and strings of emoji are common in both spam and coordinated attacks. On their own they mean very little, because some people simply write that way. In combination with the signs above, they add weight.
No single sign is proof
This matters more than any item on the list. Every one of these signs has an innocent explanation. A one-review account might belong to someone who has never reviewed anything before. A short, glowing, unspecific review might come from a genuinely delighted customer in a hurry.
Fake reviews are identified by accumulation, not by a single tell. One sign is noise. Four or five stacked in the same review, or the same pattern repeated across a cluster of reviews posted on the same day, is a signal worth acting on.
And be honest about the ceiling: only the platform can see the account data, IP addresses and posting history that would settle the question. Everyone else, including every detection tool, is reading the text and the pattern.
Checking a review you are unsure about
If a particular review is bothering you, you can run the text through our free fake review checker. You paste in the review and it analyses the wording for the patterns above: how original and varied the language is, whether the sentiment sits at an extreme, whether the length and punctuation look typical of genuine reviews, and how much specific, first-person detail is actually there.
It reports the signals it finds and tells you when a signal is ambiguous, because most of them are. It will not tell you a review is definitely fake, and you should be sceptical of any tool that claims it can. What it does is make the pattern visible faster than reading by eye, so you can decide what the review is worth.
If the fake reviews are on your own listing
Everything above is written for a reader deciding whether to trust a business. If you are the business, the problem changes shape: a fabricated one-star review is not an inconvenience, it is lost revenue, and the fix is procedural rather than forensic.
Flagging is the route, and it works more often than most owners expect, provided you can point at the specific policy the review breaks. Our step-by-step guide on how to report and remove fake Google reviews covers the flagging workflow, what to do when the standard flag fails, and how to escalate an appeal.
The longer-term defence is volume. A single fake review distorts a rating built on twelve reviews and barely registers on one built on three hundred. Steady collection of genuine feedback is the only thing that makes you difficult to attack.
