Published on 2026-07-25
Where do fake reviews actually come from?
Researchers traced how sellers buy five-star reviews through private groups, then measured what it did to sales, theirs and their honest competitors'. Here's the machinery behind it.
Picture the moment: you're about to buy a kettle, a supplement, a pair of headphones, and there it is, glowing at the top of the page, 4.8 stars from twelve hundred people. You relax. Somebody did the homework for you. But who, exactly? A stranger who bought the same kettle and genuinely loved it, or somebody who never touched it at all, paid a few dollars to say so?
That second option is not a hunch. Researchers Sherry He, Brett Hollenbeck and Davide Proserpio spent nine months tracking around 1,500 products whose sellers were soliciting fake reviews through private online groups, in a study later published in the journal Marketing Science. The mechanism they documented was almost boringly simple: a seller posts a product link in the group, a member buys it, leaves a glowing review, and gets paid back through the seller directly, off the platform, where nobody official is watching the transaction.
A marketplace hiding inside another marketplace
Those groups did not stay small or hidden for long. The UK's Competition and Markets Authority spent years pressing Facebook over exactly this problem, describing what it found as thriving communities built purely for trading reviews. Facebook's own cleanup, after that pressure, removed sixteen thousand such groups in a single push, a jump from barely a few hundred the previous year. Sixteen thousand storefronts for opinions that were never actually formed by using the product.
What makes the setup work is how ordinary it looks from the outside. Nobody involved has to lie about being a real customer, because they are one, technically. They did receive a parcel. They did open the box. The review they leave just isn't about whether the product deserved five stars, it's about collecting the refund waiting on the other end. Multiply that by dozens of members in a single group, all working through the same list of products on the same afternoon, and a brand new item can pick up a wall of glowing reviews before a single ordinary shopper has even unboxed theirs.
It also explains why the reviews so often read the same way. Someone writing quickly, about a product they were never invested in beyond the payout, tends to reach for the same safe words everyone else in the group reaches for: 'exceeded my expectations', 'great quality for the price', 'will definitely buy again'. None of those phrases are wrong on their own. It's the repetition, review after review, that gives the game away.
The tell is what happens after
Here is the part that should make anyone pause before trusting a shiny rating: He, Hollenbeck and Proserpio also tracked what buying fake reviews actually does to a listing. Sellers who bought them raised their prices and still saw unit sales climb by an average of 27.2 percent. Honest competitors selling the same kind of product saw their own sales fall by an average of 4.4 percent, and often responded by cutting their own prices just to compete with a rival whose advantage had been manufactured. The fake review doesn't just mislead the one shopper who reads it. It reshapes which seller wins the sale.
A rating built on paid enthusiasm doesn't hold its shape once the payments stop. Look at what's underneath it, not just what it adds up to.
Reading past the number
None of this means every glowing review is bought, or that a high average is automatically fake. Most people who leave a review genuinely used the thing. But the number at the top of the page cannot tell you which crowd you're looking at, so the useful habit is to look one layer down, at the reviewer rather than the review.
- Open a reviewer's profile: do they review one category obsessively, or a normal, messy mix of everyday purchases?
- Check how long the account has existed and whether this is its first review ever
- Read a few of their other reviews: do the words sound like a different person wrote each one, or like a template with the product name swapped in?
- Notice if praise arrives in a tight cluster of dates rather than spread naturally over months
A single-review account, created the week the product launched, gushing in language that could describe any object on the shelf, is not proof of anything on its own. But three or four of those in a row, sitting above the fold, are worth more suspicion than the star average itself ever earns you. The number was always the easy part to fake. The trail behind it is harder to fake convincingly, which is exactly why it's worth following.