Amazon Starts Suing Its Own Sellers Over Fake Reviews

Image: Getty
Image: Getty

In its escalating fight against fake product reviews, Amazon has for the first time started suing its own sellers instead of just going after other websites that promise to write five-star recommendations for cash. It’s another sign that the internet giant is increasingly serious about eliminating incentives for fake reviews that erode trust, even if it means targeting the third-party retailers that hawk their wares through Amazon’s platform and who, unlike the fake-review sites, are a source of revenue.

Advertisement

The web giant has sued three of its sellers: a Chinese company called CCBetterDirect, Michael Abbara, and Kurt Bauer. For all three, fake reviews comprised up to 40 percent of the stores’ total reviews, so the violations were clearly not subtle.

The newest lawsuits continue the wave of litigation that Amazon started in early 2015. The company sued various websites (like the not-obvious-at-all buyamazonreviews.info) in April and October of last year and again two months ago. In total, Amazon sued over 1,000 people, and more are probably coming. The move away from targeting sites like paidbookreviews.org and toward people on the Amazon platform shows how the company wants to send a very strong warning—which makes sense given that without trustworthy reviews Amazon wouldn’t be a very useful marketplace.

The company has long tried to develop ways to combat fakes, from ranking trusted reviews to developing algorithms that sniff out the obviously fake ones to suspending accounts altogether. It’s also tried out “verified purchase” designation of people who’ve bought the product, though as Evan Schuman at ComputerWorld points out, it’s really not hard for fake reviewers to buy one item, get that designation, and then go on to fake a review.

For marketplace sites like Amazon, reviews are crucial to distinguish themselves from competitors, and the lagging trust that consumers have across the board only hurts the company. A study from the University of Colorado at Boulder that came out last month found that user reviews and actual product quality had very little correlation. It’s unsurprising then that a separate study from Mintel, a marketing company, found that almost 70 percent of Americans ask for other people’s opinions online before buying anything, and yet only 59 percent trust the actual recommendations.

All that in mind, it makes total sense that Amazon is cracking down on fake reviews. Fake reviews are bad and discourage people from buying things! And Amazon really, really wants to sell people lots of things.

[TechCrunch]

Advertisement

Angela Chen is the morning editor at Gizmodo.

DISCUSSION

I hope that Amazon truly is cracking down on fake reviews, but I have reason to doubt it. Here’s why:

I’m a top 500 reviewer there. I do mostly video reviews, I use things for days/weeks before reviewing them, I give detailed reviews... pretty much what I’d hope for from other reviewers, because as a buyer I want to be able to count on what I read (I know I can’t always do that...).

A few months ago, I got some headphones (Aonsen) to review (I review stuff I buy, but I also get offers to review other products). They were loud, and reasonably comfortable, but the sound quality sucked, so I gave them 3 stars and said they’d be good for people with hearing issues but others should look elsewhere.

Within days, I had almost 140 “unhelpful” votes on a headphone I doubt even 10 people actually bought, and so did all the other non-5-star reviews. My rank dropped from ~380 to ~780 in days. Side note: if someone who’s been reviewing for years and years can drop precipitously because of one review out of hundreds, something is wrong with their algorithm.

A top 50 shill (who “reviews” 5-10 things a day) bumped his review from 4 to 5 stars because he “decided they were really worth 5 stars after listening more.” He and all the other 5 star reviews (including one that basically said “sounds good”) had roughly 140 “helpful” votes. My theory? The seller used Amazon Turk to buy votes (oh, the irony).

So I wrote to Amazon. This was the clearest case of vote manipulation I’d ever seen: a very low selling off-brand headphone with absolutely polar vote counts centered exclusively on the stars given. Nearly identical numbers of votes, given at about the same time. I’d gotten about a third of the downvotes I had total over my >10 years and >500 reviews in mere days.

Their reply was that they’d looked into it and saw nothing wrong and that no action would be taken.

My brother is a top 200 reviewer there. He once gave a low-star review and then, shortly thereafter, a months-old tooth whitener strip review he’d done suddenly sprouted 100+ unhelpful votes. He wrote and got, nearly word for word, the same reply.

So, sure, they’re suing some people and sellers. I think it’s for the PR. But they have clear and obvious cases of seller vote manipulation as well as clear and obvious cases of shill reviewers (seriously: who can do 5-10 video reviews a day, day in and day out, and adequately review anything?) and they’re doing nothing about it.

I deleted the review after a month or so (when the seller had moved on) and re-posted it (“scrubbing”) which is the one and only time I’ve ever done that. I jumped hundreds of rank positions within days.

If Amazon truly wants people to start believing the reviews again, there are some extraordinarily simple steps they can take to do so besides suing some sellers:

(1) Check vote polarization around star ratings. Anything where helpful/unhelpful votes are nearly completely divided based on rating, it’s not unlikely something is amiss.

(2) Force a “purchased to discounted product” review ratio. Lots of top reviewers review almost exclusively items for which they paid less than 10% (often 0%) of the price. That’s fine, as long as they also review many things they bought at full price. This would help reduce the full-time shills - and believe me, there are full-time shills.

(3) Promote video reviews over non-video reviews for products the reviewer didn’t pay full price on Amazon to get. Related: since they have to review the videos when submitted anyway, make sure the review shows the product. This would get rid of a lot of the “awesome, dude” 5-star reviews.

(4) Weigh reviews from people who paid full price on Amazon for a product over discounted/got elsewhere reviews.

(5) Reduce the impact of one or a few “unhelpful” reviews as well as the impact of one or a few “super helpful” ones. Base ranks on, say, the middle 90% of reviews (so something like that).

(6) Certify reviewers (unsure how this would work). Similar to the Vine program but broader, where Amazon picks some people and basically vets them.

There really, truly, definitely is a shill issue on Amazon. As someone who’s done a lot of reviews of discounted products, I have seen the underbelly of that issue. Feel free to ask about it. It’s too bad it’s the way it is, because a very few fixes could really help a lot.