Fake Review

Also known as review fraud, spam review

The four kinds need four different responses, which is why treating them as one problem goes nowhere. Bought reviews come from a broker, a marketplace listing, or a private group that trades them, and they are a decision somebody made. Insider reviews come from staff, family, or a founder's second account, and are often written by people who did not think of it as fraud at all. Undisclosed incentives are the subtlest: a genuine customer with a genuine opinion who was paid, discounted, or sent free product and never said so. And competitor sabotage is the one most Shopify merchants meet first, because it is the only one that happens to you rather than by you.

That asymmetry matters. The first three you can end this afternoon by changing a policy and telling your team. The fourth you cannot end at all, only detect, document, and report, and the platform decides. Anyone selling you certainty about the fourth is selling a heuristic.

Verified buyer badges are not the answer people assume. A verified badge means the app matched the reviewer against an order the store actually processed. It says nothing about a review submitted by someone who never ordered, if the app accepts unverified submissions at all, and it does not touch an insider who placed a real order first. Turning off unverified submissions closes that door completely, at the cost of losing genuine reviews from customers who bought the same product through a stockist or a marketplace. That is a real trade-off rather than a free win, and it is worth making deliberately rather than by default.

Consider a Shopify store selling a folding treadmill that receives eleven one-star reviews across four days. They are all about the same length, all describe a motor fault in slightly different words, and none of the email addresses match an order in the store's records. The instinct is to reply to each one publicly. What actually worked was quieter: export the eleven, check every address against orders, gather what the app's log showed about submission times, report them together under the platform's policy with the order evidence attached, and switch off unverified submissions. The one public reply the store did write that week was to a genuine one-star about how loud the belt was, which was the review a shopper was going to read anyway.

There is a legal layer, and it changed recently. The United States Federal Trade Commission adopted a rule specifically on consumer reviews and testimonials which treats buying, selling, or knowingly disseminating fake reviews as a violation, with civil penalties assessed per violation and adjusted annually. The scope covers more than fabrication, including insider reviews without disclosure and, on some readings, the suppression of genuine negative reviews. Read the rule text on ftc.gov rather than a summary, including this one, because the boundaries are argued over and the penalty figure moves each year. Other markets have their own instruments and the UK and EU regimes are not identical. The practical rule is the same everywhere: if a review would read differently to a shopper who knew how it was obtained, disclose it or do not publish it.

The last honest point is about detection. No review app can verify a purchase it did not process, so what a vendor calls fake review detection is pattern matching on text, timing, addresses, and account behaviour. Pattern matching has a false positive rate. Ask any vendor, this one included, what their filter does when it is wrong, whether a wrongly flagged review is deleted or held for a human, and whether the customer is told. A filter that quietly discards real customers to keep a chart clean is a slower version of the problem it claims to solve.