Fake Review Detector

No tool can look at review text and certify that a review is fake — that is the first thing to understand. What the detector estimates is risk: how closely the language, timing, and reviewer patterns resemble known manipulation tactics. A high risk score means 'treat this rating with caution,' not 'these reviews are lies.'

That distinction matters because acting on certainty you do not have leads to bad decisions — dismissing a good product over a false alarm, or trusting a manipulated rating because the score felt reassuring. Use the risk score as one weighted input alongside the critical reviews and your own reading. The disclaimer page lays out the full caveats.

Genuine reviewers describe the same product in wildly different words — that is what makes repetition suspicious. When a dozen reviews all praise the 'exceptional build quality and outstanding customer service' in nearly identical phrasing, the writers were likely working from a common script rather than independent experience.

The scan clusters near-duplicate phrasing across reviews and weighs how unlikely that overlap is by chance. A few shared adjectives are normal; whole sentences in lockstep are not. This pattern shows up most in incentivized-review campaigns, where participants are coached on what to say.

Real customers write like people: specific, uneven, occasionally grumpy about small things. Manipulated reviews often read like marketing copy — strings of superlatives with no concrete detail, phrases like 'highly recommend to everyone' attached to products the reviewer barely describes using. Excessive polish is itself a signal.

The scan also watches for generic praise that could apply to any product in the category, and for odd phrasing that suggests translated or templated text. None of these markers is decisive alone; the risk score rises when several appear together inside the same review cluster.

The strongest checks take thirty seconds and a click. Open a suspicious reviewer's profile: a real account usually shows a mix of ratings across different products over time. A profile with nothing but five-star reviews, or a burst of twenty reviews posted on a single day, deserves skepticism.

Also note how many of their reviews carry Verified Purchase status, and whether their reviews ever mention a downside — even happy customers usually notice something. Profiles can be deleted or made private, which limits this method, but when the data is visible it is often the most telling evidence you will find.

Treat the risk score as a dial on the rating, not a verdict on the product. High risk plus high stars means the rating is doing the selling and the evidence is thin — read the critical reviews first and look for variant-level detail. Low risk means the rating is more likely earned, so the complaint patterns become the deciding factor.

The confidence level tells you how much evidence the score rests on: more reviews, more certain. With only a handful of reviews, even a low score is provisional. The honest use of this tool is narrowing where you spend your attention, not outsourcing the decision.