Fakespot

» see fakespot.com as it lived, on the wayback machine

~ the last words ~

fakespot started in 2016 when saoud khalifah bought a supplement on amazon that was wrapped in glowing five-star reviews and turned out to be junk. instead of leaving an angry review of his own, he built a tool that looked at the review pattern itself: timing clusters, language similarity, reviewer history, brand-incentivized phrasing. you would paste an amazon link into fakespot.com or install the browser extension and it would grade the listing A through F based on how trustworthy its reviews looked. the model got upgraded to a deep-learning thing somewhere around 2018, claimed 90 percent accuracy on flagged listings, and ate into about ten million users across extensions, mobile apps, and the website.

mozilla bought fakespot in may 2023 and bolted it into firefox as the review checker feature inside the address bar. the idea was that mozilla would have a consumer-facing utility that paid for itself in goodwill, and fakespot would get distribution through firefox. it ran inside the browser for about two years. then in may 2025 mozilla announced a refocus on firefox core, killed pocket in the same memo, and set fakespot's shutdown date at 1 july 2025. the browser extensions, the ios and android apps, and the website all went dark on the same day.

fakespot would have been useful to me in the 2018-2024 window when amazon turkey opened and the cheap-electronics listings started filling up with the same suspicious five-star review patterns as the.com site. i never installed it. i used the old engineer's heuristic of sorting by one-star and reading the bottom of the page, which is a worse version of what fakespot did automatically. the people who installed fakespot tended to be the ones who shopped on amazon often enough for the extension to pay for itself in avoided junk purchases, which fit a us amazon prime household more than a turkish marketplace shopper.

~ the record ~

~ how it landed ~

2016 was a good year to launch a fake-review tool. amazon had spent the previous five years quietly turning into the default product-search engine, beating google for the "should i buy this" query. third-party sellers had flooded in. review brokers were openly advertising on facebook groups. the fake-review economy was visible enough that you could read three trade-press features about it in a month. fakespot landed in that gap with a single-purpose tool and a free tier that did most of what people needed.

the early product was a website where you pasted a url. khalifah added the chrome extension, then the firefox extension, then the mobile apps. amazon spent years trying to get fakespot deplatformed, sending takedown requests to apple's app store in 2020 over the way the ios app overlaid review-grade information on top of amazon's own pages. apple pulled the app for a stretch. fakespot reshipped without the overlay. the friction with amazon kept the tool in the press, which was good for installs.

~ what it actually did ~

the grade was an A-through-F badge sitting at the top of the product page. the underlying model looked at a list of signals nobody enjoys reading individually: how clustered the review timestamps were, how repetitive the phrasing was across reviewers, how often a reviewer had reviewed only one brand, how often the seller had changed listings on the same asin, how many reviews appeared right after a discount campaign. fakespot recombined all of that into a single letter grade and a short paragraph explaining which signal pushed the score down. it was the rare consumer tool that did one thing well and explained itself in two sentences.

the model was always going to be an arms race. fake-review brokers got better at imitating organic review patterns. amazon's own review-moderation got better at scrubbing the worst of it. the gap fakespot exploited got narrower without ever closing. by the firefox-acquisition era the tool was still useful but less obviously useful than it had been in 2018.

~ the inheritance ~

mozilla's may 2025 announcement that pocket and fakespot would both be wound down was the same memo. the company had a new ceo by december 2025 and the strategic message was to stop spreading attention across product side-bets and put the engineering back into firefox itself. fakespot had never become a revenue line. pocket had never become one either. both got cut on the same day for the same reason.

the underlying problem fakespot solved did not go away. the review walls on amazon and aliexpress and temu kept filling up. a handful of clone extensions filled the gap with weaker models and worse interfaces. the firefox review checker that had been built on fakespot's engine was switched off in the same release cycle. ten million users went looking for another grade, found nothing that worked as well, and eventually defaulted back to sorting by one-star.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »