Fake reviews now make up an estimated 4–11% of all online reviews, depending on the category. They distort buying decisions, damage honest businesses, and erode the entire concept of social proof. Here's the short version of how we — and you — can catch them.
The five tells
- Burst patterns. Real review traffic decays smoothly. A flat zero for months followed by twelve 5-star reviews in a single day is almost always a paid burst.
- Identical phrasing. Templates copy-pasted across reviewer accounts. "Great service, would recommend, A+++" is the canonical example.
- Generic praise without specifics. Real reviewers mention the agent's name, the order number, the model of the product. Fakes stay vague.
- Reviewer history. Brand-new accounts with one review, or accounts that only review one company across years.
- Unnatural rating distribution. Real businesses sit on a J-curve. A perfect 5.0 across hundreds of reviews is statistically rare and worth a second look.
What Compasly does about it
We run every submitted review through a layered pipeline: rate-limit per IP and per account, language-model classifier for AI-written prose, burst-detection on the company timeline, and manual review for anything the model flags. Verified-purchase reviews carry a separate badge so you can filter to them when you need certainty.
You can flag any review you suspect is fake using the "Report" link below it. Our team reviews every report within 24 hours.