Fake Review Detective

Spot fake reviews before you get burned

Import reviews from a URL or paste them manually. Computes real statistics, then AI scores each review individually and detects manipulation patterns.

Overview

Fake Review Detective uses a two-phase approach: first, JavaScript computes real statistics from your pasted reviews (star distribution, verified %, date clusters, language flags) — instant, no AI needed. Then AI scores each review individually for authenticity (0-100 with red/green flags) and analyzes cross-review patterns (manipulation detection, genuine consensus, purchase recommendation). Every number you see is computed, not hallucinated.

How to use it

1. Paste a product URL to auto-extract reviews, OR paste review text manually
2. Extracted reviews appear in the text area — edit them if needed
3. Select the product category for category-specific benchmarking (auto-detected from URLs)
4. Click 'Detect Fakes' — instant stats appear immediately
5. AI then scores each review individually (Step 1) and analyzes patterns (Step 2)
6. Review the Quick Verdict card for the overall trust score
7. Expand individual review cards to see per-review red/green flags
8. Check the Genuine Consensus section for what real reviews actually say
9. Use the Purchase Recommendation to inform your decision

Example

Scenario: You're considering wireless headphones with 4.5 stars but the reviews seem suspicious — lots of 5-star reviews posted on the same day with generic language, plus a few detailed reviews from verified buyers

What you do: Paste all the reviews, select 'Electronics' category, click Detect Fakes

Result: Instant stats show: 37% verified (red flag), date cluster of 3 reviews within 48 hours. AI scores the generic 5-star reviews at 15-25/100 (likely fake) and the detailed verified reviews at 80+/100 (likely genuine). Quick Verdict: Trust Score 42/100 — 'Approach with Caution.' Genuine consensus: decent sound quality, weak bass, comfortable for short sessions. Verdict: WAIT for more verified reviews.

Tips

• Include as many reviews as possible — pattern detection improves with volume (minimum 100 characters, 3+ reviews recommended)
• Copy reviews with their star ratings and dates for timeline analysis
• The instant stats panel gives useful data even before the AI runs
• Sort reviews 'Most suspicious first' to quickly find the fakes
• The Genuine Consensus section is the most actionable — it tells you what the product actually is based on reviews you can trust
• Try the example reviews to see the tool in action before pasting your own

Common pitfalls

• Don't paste just 1 review — pattern detection requires multiple reviews
• Not every 5-star review is fake — some products are genuinely great
• URL extraction may fail on sites that require JavaScript or block automated requests — paste reviews manually as a fallback
• AI analysis is guidance, not a guarantee — always use your own judgment
• Low trust score means the review SET is unreliable, not that the product is bad

Frequently asked questions

How can I tell if product reviews are fake?

The reliable tells are patterns, not individual reviews: bursts of five-star reviews in a short window, repeated phrasing across 'different' reviewers, reviews that describe the product category but not the specific product, and star distributions with no middle. Fake Review Detective analyzes the reviews you paste and scores the manipulation patterns it finds.

What percentage of online reviews are fake?

Estimates vary by platform and category, but researchers and platforms themselves have suggested that a meaningful share of reviews on major marketplaces are inauthentic — some analyses have put problem categories north of a quarter. The safer assumption is that any product with heavy incentives to fake (dominant categories, low margins) has some contamination — which is why reading the pattern matters more than trusting the average.

What are the red flags of a fake review?

Extreme sentiment with no specifics, marketing copy vocabulary ('game-changer'), timing clusters, reviewer histories full of same-category five-stars, and reviews that answer objections nobody raised. Real reviews complain about weird specific things — the absence of weird specifics is itself a flag.

Does it work for reviews from any website?

Yes — it analyzes the review text you paste rather than scraping a specific platform, so marketplace listings, app-store reviews, hotel and restaurant reviews all work. Paste a representative sample including some negatives; the pattern analysis improves with more text.

Why do fake reviews matter if the product is decent?

Because manipulated ratings redirect your money from honest products to whoever paid for the campaign — and sellers who buy reviews tend to cut corners elsewhere (warranty games, review-gating, quiet relistings). Manipulation detected in reviews is a vendor-trust signal, not just a product-quality one.