How to Spot Fake Reviews: 12 Red Flags Shoppers Should Know

You spot fake reviews by reading the pattern, not by judging one review on its own. The strongest red flags are reviews that arrive in a sudden burst, a rating page with almost nothing between one and five stars, the same phrases repeated by different people and reviewer accounts that exist only to praise or attack. One flag is a reason to look closer. Three together are a reason to shop somewhere else.

This guide expands the fake-review section of our complete guide to trusting online reviews into a checklist you can run in about 10 minutes. It's a detection guide only. We moderate reviews on TheUserOpinion ourselves, so parts of it come straight from the report reasons in our review guidelines, which spell out what our moderators remove and why.

Who This Guide Is For

It's for shoppers checking a product, an app, a hotel or a local service before they pay, and for anyone picking software from review pages, including the tools in our small business software stack guide. Business owners will find it useful too, because it's roughly what a moderator looks at when a rival's page seems too good.

It isn't for anyone hoping to learn how to write fakes that slip through. We don't cover that. In a growing list of countries it's a legal problem now, not a marketing tactic.

Why You Can't Spot a Fake Review by Reading It

Start with an uncomfortable finding. In a 2011 study, Cornell researchers paid 400 people $1 each to write a fake positive review of one of 20 Chicago hotels, then mixed those with 400 truthful five-star reviews of the same hotels. Human judges did little better than chance at telling them apart. A trained computer model got 89.8% right (Ott, Choi, Cardie and Hancock, ACL 2011).

Generative AI has made this harder. Clumsy grammar used to give a cheap fake away, and now a fluent one costs almost nothing. The FTC's 2024 rule explicitly covers fake reviews generated by AI.

So stop asking whether a review sounds fake. Ask whether the page behaves like real customers. Real customers arrive at an uneven pace, disagree with each other and complain about specific things. Campaigns don't.

Red Flags on the Rating Page

These four show up before you read a single word. They're the fastest checks and usually the strongest.

1. A Burst of Reviews in a Short Window

Canada's Competition Bureau puts "sudden spike or dip" first in its Five-star fake out consumer alert, and asks one useful question: are the reviews spread over a long time, or compressed into a short one? Sort by newest and scroll back a few months.

Real spikes happen. A launch, a big sale or a viral post brings a wave of genuine buyers. The tell is a spike with no reason behind it, followed by silence. If you can't find the launch or the news story, treat the batch as one data point, not 40.

2. A Sudden Wave of One-Star Attacks

Fake negatives exist, and they're easy to forget. The same Canadian alert warns that very negative reviews can come from a competitor trying to steer you toward another product. Check whether the one-stars describe a real incident with dates and details, or just say "scam" and stop.

Review bombing is a cousin of this. After a policy change or a news story, hundreds of angry people can pile onto a page in days. They may be sincere, but many never used the product. Read for first-hand use. "I paid for the annual plan and the export broke" counts. "I heard they're terrible" doesn't.

3. Only Five Stars and Nothing in the Middle

"Every company, no matter how good, is likely to have a few negative reviews," the Competition Bureau notes. Its advice is our favorite line on the subject: the meat is in the middle. Someone writing a fake is more likely to give one or five stars, so balanced criticism tends to live in the two-, three- and four-star reviews.

Watch the sample size before you judge, though. A café with six reviews can honestly have six five-stars. That's why our TrustScore puts 20% of its weight on volume and only gives full credit at 100 reviews. A perfect score from a handful of people is a small sample, not a red flag on its own.

4. Negative Reviews That Seem to Have Vanished

Some pages don't add fake praise. They quietly hide the criticism. The UK's Digital Markets, Competition and Consumers Act names this directly: removing negative reviews while publishing positive ones, or giving positive reviews more prominence, counts as publishing reviews in a misleading way. It's been banned in the UK since April 6, 2025.

You can't see a deleted review, but you can see the gaps. A seller's own website widget that shows only five-stars, a review count that doesn't match what you can scroll through or complaints on a forum that never appear on the seller's page are all signs. Prefer independent review sites over a seller's own testimonials, and on any site, check whether businesses can remove reviews. On ours, they can't, at any price.

Red Flags in the Review Text

Text signals are weaker than timing signals, because honest people write badly and dishonest people write well. They still add up.

5. Big Feelings, No Details

"It's the best ever!!!" gets its own warning in the Canadian alert, especially when the review doesn't say why. Real reviews mention the plan, the version, the delivery date, the thing that broke and what support did about it. Our guidelines ask reviewers to "say what happened and when" for exactly this reason. Specifics help the next buyer, and they're harder to invent at scale.

6. The Same Phrases From Different People

Some companies hand reviewers scripts with suggested phrases and keywords, the Competition Bureau says. When three strangers all call a product "a must-have for every busy team", that isn't a coincidence. Copy an odd phrase from a glowing review and search the page for it with Ctrl+F (Cmd+F on a Mac).

AI-written batches leave a softer version of the same print: the same rhythm, the same tidy three pros and one mild con, the same upbeat closing line. One polished review means nothing. Twenty that could swap places without anyone noticing is a pattern.

7. More Scene-Setting Than Product

The Cornell team found that truthful hotel reviews used more concrete words about the place, like "bathroom", "check-in" and "price", while fakes leaned on things that set the scene, like "vacation", "business trip" and "my husband" (Cornell Chronicle, July 2011). It's hard to tell a specific lie about a place you've never been.

The same logic carries over to software, with a caveat: the study only validated the method on Chicago hotel reviews. Here's our own example, not from the study. A real review of a project management tool mentions the board view, the import that failed and the price per seat. A thin one talks about how the team's whole energy changed.

8. Praise for a Different Product

This one fools careful shoppers because the reviews themselves are real. They just belong to something else. In 2023 the FTC finalized an order requiring The Bountiful Company to pay $600,000 for review hijacking: it used Amazon's variation feature to group new products with older, well-reviewed ones, so ratings and badges appeared on products that hadn't earned them (FTC analysis in the Federal Register, February 22, 2023).

The tell is in the text. If reviews mention a different flavor, size, model or a feature the listing doesn't have, the rating isn't about what you're buying. Filter by the exact variant when the site lets you.

Red Flags on the Reviewer

Click the name. Many review sites show a review count or a join date, and that small profile says more than the review does.

9. A Brand-New Profile With One Review

The Competition Bureau flags reviewers who created a profile recently and left positive feedback on a few products over a short period, noting they may work for the company. Be fair here. Plenty of real people review once in their lives, usually when something went very well or very badly. A one-review account is a weak signal alone and a strong one in a batch. Five of them in the same week is a different story.

10. A History of Extremes or Unrelated Products

Look at what else the account has reviewed. Only five-stars, or only one-stars, with nothing in between is unusual for a real person. So is a burst of reviews for unrelated products from unrelated brands in a few days. A normal shopping life is slower and messier than that.

11. A Reward Mentioned in Almost Every Review

"I received this product free in exchange for my honest review" isn't a fake review. It's a disclosed incentive, and disclosure is the honest version. The problem is when nearly every positive review carries that line and the paying customers sound far less happy. The incentive changed who wrote the reviews.

The law draws the line at sentiment and secrecy. The FTC rule bars businesses from offering incentives conditioned on a positive or negative review (16 CFR 465), and the UK act bans reviews that conceal they were incentivized. Canada's Competition Bureau has also warned that employees must disclose their connection when they review their employer or its competitors.

12. A One-Star Review That Recommends a Rival

A negative review that ends with "I switched to [competitor] and never looked back" deserves a second look, especially if several say the same thing. The Canadian alert describes this exact move: discourage you from one product while suggesting another. Real switchers do write this way sometimes. A cluster of them naming the same rival is how a campaign reads.

How Strong Each Red Flag Is

No single flag proves anything. This table is our judgment of how much weight each one carries alone, and what to check next before you decide.

A 10-Minute Check Before You Buy

You need the product or business page, one other independent review site and about 10 minutes. Run the steps in order. The early ones catch most problems.

1. Write down the total review count, the average and the shape of the star breakdown.

2. Sort by newest and scroll back three months, looking for bursts and silences.

3. Read five reviews rated two, three or four stars and note the complaints that repeat.

4. Copy one odd phrase from a glowing review and search the page for it.

5. Open the profiles of three five-star reviewers and check their history and join dates.

6. Check that the reviews describe the exact model, plan or variant you're buying.

7. Count how many positive reviews mention a free product or a reward.

8. Open the same product on a second independent site and compare the themes, not just the stars.

9. Decide: with zero or one flag, trust the repeated themes; with two, trust only the middle ratings; with three or more, walk away or buy only where returns are easy.

Our recommendation: step 3 does the most work for the least effort. If the middle reviews describe a worse product than the five-stars do, believe the middle.

If It Doesn't Work

• The site hides reviewer profiles. Skip step 5 and lean harder on timing and repeated phrases.

• You can't sort by date. Look for a stated date of experience on each review, or check the newest page of reviews on a second site instead.

• There are too few reviews to see a pattern. Most of these flags need dozens of reviews to mean anything. Ask the business for references, read its refund terms and buy somewhere with an easy return.

• You want a tool to do it for you. Detector tools can help as a second opinion, but they aren't infallible, and Fakespot, one of the best known, shut down on July 1, 2025. Your own 10 minutes beat a score from a tool you can't inspect.

What to Do When You Find Fake Reviews

Report it to the platform first, because the platform can remove the review and fix the rating. On TheUserOpinion, sign in, use the flag on the review and pick the closest reason: "Spam or fake review", "Not a genuine customer experience" or "Conflict of interest", which covers reviews by the business, its staff, a competitor or someone who was paid. You can add up to 500 characters of detail, and what you noticed (the dates, the repeated phrase) is the most useful thing to put there.

Reporting doesn't take a review down by itself. A moderator, a person rather than an algorithm, reads it against our guidelines and keeps it, holds it while we check or removes it. A removed review stops counting toward the rating and the TrustScore. No business can pay to change that outcome.

If the problem looks organized, tell a regulator as well. In the US, the FTC takes reports at ReportFraud.ftc.gov, and its fake-review rule has let it seek civil penalties against knowing violators since October 21, 2024. In Canada, the Competition Bureau asks people to report fake reviews to it directly. In the UK, the CMA's fake reviews guidance explains what businesses and review sites now owe you.

Our Take

Trust the shape of the page over any single review. A product with 200 mixed, specific reviews and a few public owner replies is a safer bet than one with 40 perfect ones. We'd rather buy the 4.3 with complaints we can live with than the 4.9 with nothing to read.

The exception is anything new or small. A local plumber or a software tool that launched last month will have few reviews, and that's not suspicious. For those, the flags above matter less than references, a free trial and a fair refund policy. When you're comparing established tools, compare several at once, for example by reading user reviews of project management tools side by side, so one inflated page stands out against its neighbors.

The Bottom Line

How to spot fake reviews: check the timing, the star breakdown, the wording and the reviewer, and weigh the flags together. A burst with no reason, repeated phrases, vanished negatives and reviews for the wrong product are the strongest signs. A new account or a gushing line on its own proves very little.

For quick purchases, the 10-minute check is enough. For big ones, add a second site and read the middle ratings first. For tiny businesses and brand-new products, rely on references and refund terms instead.

The main warning: verified doesn't mean honest, and AI has made fake reviews fluent. Patterns are still harder to fake than words.

Next step: run the check on the next thing you're about to buy. And if you've used software listed here, find it and write a review that says what happened and when. That's the kind fakers can't copy.