Start with patterns, not one suspicious sentence
Fake reviews rarely wave a flag. They usually look ordinary. Look for a cluster instead: vague praise, copied wording, a sudden posting burst, a thin account history, or product details that do not fit. One clue proves very little. Several independent clues should slow you down.
Before you buy, read the newest, middle-rated, and most critical posts, then open a few reviewer profiles. That small sample often tells you more than the star average.
For U.S. shoppers, platform policy and federal consumer-protection law do different jobs. A marketplace decides whether a post violates its rules; the FTC's rule addresses certain deceptive conduct by businesses and others involved in reviews. The FTC rule took effect on October 21, 2024, but it does not certify any particular review as truthful.
Turns out, you do not need to prove a stranger lied. Your job is simpler: decide whether the evidence is good enough to trust this listing, seller, or business.
Red flags that deserve a closer look
Use the table to decide what to check next. It is not a fraud score.
| Signal | What you may see | How to test it |
|---|---|---|
| Vague, interchangeable praise | "Amazing product" or "Highly recommend" with no useful detail | Search the phrase elsewhere and compare the review with the product's actual features |
| Sudden timing burst | Many reviews appear within a narrow period | Sort by date and see whether the pattern continues naturally |
| Thin reviewer history | A new account posts once, or reviews only one seller | Open the profile and look for varied, dated activity |
| Repeated wording | Different accounts use the same sentence structure or unusual phrase | Search a distinctive sentence in quotation marks |
| Details that do not fit | The post names the wrong model, location, size, or service | Compare the claim with the listing, receipt details, photos, or business information |
| Incentives or relationships | A free item, payment, employee connection, or family connection is not disclosed | Look for a clear disclosure and check whether the review sounds conditioned on praise |
| Generic or reused photos | Several accounts show the same image or a stock-looking customer photo | Run a reverse-image search and compare the image with the claimed experience |
| Extreme emotional pressure | "This changed my life" or "This business is a scam" without supporting facts | Look for specific events, dates, defects, or service details |
An unusual phrase can be coincidence. Copied wording and synchronized dates are harder to dismiss.
When AI and review farms leave clues
AI can produce polished nonsense. Look for repeated sentence shapes, sweeping claims, and reviews that never explain what happened during use. An AI detector may flag a pattern, but it cannot establish who wrote the post or whether the experience happened.
Review farms leave a wider footprint: several accounts may post in a narrow window, praise the same features, and seem focused on one seller or business, so the profile pages may look unrelated until you compare the dates and vocabulary side by side. The pattern is not always neat.
Thing is, coordination matters more than one magic phrase. Real customers can sound repetitive. AI-written text can sound natural.
What changes by platform
Amazon review warning signs
On Amazon, a Verified Purchase label adds context, not a truth seal. Amazon says it uses machine-learned models for ratings and reviews, and its rules prohibit creating, editing, or removing reviews in exchange for compensation. Read Amazon's guidance on customer reviews and ratings for the platform's stated standards.
Sort by newest if Amazon offers that view. Then compare highly rated, middle-rated, and critical posts. Pay attention when many reviews repeat the product title, seller language, or the same unusual compliment, especially in a tight burst of dates.
Yelp review warning signs
Yelp posts are about local experiences. Useful ones usually give some setting: when the visit happened, what service was requested, who helped, or what problem got fixed.
One-line praise and one-line rants are not automatically fake. They merit a closer look when several accounts use the same wording.
Check the person's history across local businesses. New profiles can be real; a group of thin profiles posting matching complaints or compliments is more concerning. One angry review can still be genuine.
Google review warning signs
Google reviews may be copied across listings or onto another platform. Search an unusual sentence in quotation marks, then check the business name, location, and service details against the claimed experience.
Google's reporting process is for policy violations, not opinions you simply dislike. Use the available report control and select the reason that best matches what you can show.
Google says reviews that violate its content policies can be removed after evaluation. A rapidly growing review count, by itself, does not prove the listing is fake. Its Business Profile review guidance covers the reporting and appeal process.
What does not prove a review is fake
Several popular shortcuts mislead shoppers. A missing Verified Purchase badge does not prove fraud, and a new account is not automatically fake. Detailed posts can be dishonest. Short posts can come from real customers.
Photos are useful, but they can be copied. No photo may reflect privacy or simple disinterest. To be honest, perfect grammar may mean careful editing, while messy grammar may come from a genuine customer. Even a review cluster can follow a legitimate product launch or service disruption.
Judge the pattern, not the vibe.
The U.S. rule behind review disclosures
The federal rule targets conduct behind reviews, not every opinion a shopper dislikes. In the United States, the FTC's Consumer Reviews and Testimonials Rule prohibits buying or selling fake reviews and covers certain incentives, undisclosed insider testimonials, and review suppression practices. The rule text appears in 16 CFR Part 465.
An honest review written after an incentive is not automatically fake. Still, a business cannot make the benefit depend on positive or negative sentiment. A material connection may also require a clear disclosure that ordinary consumers can understand.
The FTC says the rule creates no private right of action. A report may help a platform or agency spot a pattern, but it does not promise a refund or automatic removal. The FTC's questions and answers about the rule set out those limits.
A practical review check before you buy
Use this sequence when a listing feels unusually polished or strangely uniform:
- Sample the spread. Read the newest and most critical posts, then compare them with a few middle-rated reviews. One star rating hides a lot of detail.
- Open reviewer profiles. Check account history and dates. Look for people who review a range of products or businesses, rather than treating account age as a verdict.
- Compare the wording. Search an unusual sentence in quotation marks. Save the links if the same text appears across accounts or platforms.
- Test the details. Match claims against product specifications, the service location, photos, model number, or dates shown in the listing.
- Look for incentives and connections. Check for free-product disclosures, employee relationships, or language suggesting the reviewer was asked to praise or attack someone.
- Check independent evidence. Review the seller's identity, return terms, warranty, delivery promise, and reliable product testing. Suspicious reviews do not by themselves prove a defect or create a refund right.
- Pause or proceed deliberately. If several clues converge, compare another seller or product before paying.
How to report suspicious reviews
Report a specific policy concern. Do not report a review merely because it is negative. Save what you saw first.
- The direct review and profile links
- Screenshots showing the wording, rating, date, and any disclosure
- Related reviews showing copied language or coordinated timing
- A short explanation connecting the evidence to the platform's policy
Use the platform's report control and choose the closest reason. Keep the explanation factual; do not claim more than the evidence shows.
On Amazon, use the report option attached to the review or Amazon's customer-support route. On Google, use the review menu or the Business Profile reporting tools described in Google's guidance. A platform can leave a post online when it finds no policy violation, even if you think the post is unfair.
Yelp and other sites have their own processes; the FTC's consumer guidance on reporting suspicious online reviews lists routes for popular platforms. You can also report suspected business-driven deception to the FTC, but that is separate from asking a platform to remove one review or seeking a refund.
Questions worth checking
Can a Verified Purchase review still be misleading?
Yes. The label may show a purchase connection, but it does not prove every claim is accurate, complete, or unbiased.
Compare it with other firsthand details.
Are AI-sounding reviews automatically fake?
No. AI detectors make mistakes, and people can write formulaic reviews. Use unusual wording as one clue among several.
Should every suspicious review be reported?
Report a likely policy problem, such as deceptive content, an undisclosed relationship, or coordinated manipulation. Unusual writing, a negative opinion, or fast posting alone may not be enough.
Make the buying decision separately
Reviews change your confidence, but they do not replace checking the seller, return terms, warranty, or product specifications. Those details matter if the review evidence falls apart.
Before entering payment details, sample the newest, middle-rated, and critical reviews and check at least a few profiles. If the timing, wording, and account activity still look coordinated, pause and compare another seller or product.