Fake Reviews, Imported Reviews & GMC: What Google Actually Detects

Reviews build trust. Fake reviews destroy it — and get your Merchant Center account suspended. Google has become extremely sophisticated at detecting review fraud, and the signals they look for go far beyond just reading the text.

What Google Detects

1. AliExpress Review Import Plugins

Google scans your HTML source code for known review import app scripts. AliReviews, Loox (in import mode), DSers Reviews, and similar plugins leave identifiable JavaScript snippets in your page source. If detected, it's flagged as CRITIQUE severity — because the only reason to import reviews from AliExpress is if your products come from AliExpress.

2. Temporal Clustering

50 reviews appearing within 2 hours is not organic growth. Google checks the timestamp distribution of your reviews. Natural review patterns show gradual accumulation over weeks and months, with occasional spikes after promotions. Bulk import patterns show a sudden wall of reviews on one date.

3. Statistically Impossible Ratings

A store with 200 reviews and 0% negative ratings is statistically impossible with real customers. Google flags stores where less than 3% of reviews are negative on 20+ total reviews. Real businesses always have some dissatisfied customers — that's actually a trust signal.

4. Cross-Product Identical Reviews

"Great product, fast shipping, love it!" appearing word-for-word on 15 different products is a copy-paste import pattern. Google checks review similarity across your product catalog.

5. Schema.org Review Count Inflation

Some stores declare 1,000 reviews in their JSON-LD AggregateRating schema, but the page only shows 10 real reviews. This discrepancy between declared and visible reviews is flagged as CRITIQUE severity. It's explicit misrepresentation of social proof.

6. Review Volume vs Domain Age

1,000 reviews on a 3-month-old domain? Google cross-references your review count against your domain registration date. Unrealistic review velocity for a young domain is a strong fraud signal.

What's Actually Safe

Real customer reviews: Even if they're mixed (4-star average is more trustworthy than 5-star)
Photo reviews: Customer photos are extremely hard to fake at scale
Judge.me / Stamped (organic mode): These collect real post-purchase reviews — no importing
Trustpilot: Third-party platform with independent verification
Google Customer Reviews: Google trusts its own review system above all others

How Many Reviews Do You Really Need?

For GMC approval: 10-20 real reviews with a natural rating distribution (3.5-4.5 average) is far more effective than 500 imported 5-star reviews. Quality and authenticity beat quantity every time.