Knockoff
Amazon, without the knockoffs
230 followers
Amazon, without the knockoffs
230 followers
Knockoff filters the trademark-squat pseudo-brands (the SZHLUXes and HORUSDYs) out of your search results, so what's left is brands with a reputation to lose.
Curious how Knockoff decides something is likely a knockoff on Amazon. Is it looking at seller history, review patterns, brand matching, listing text, or some mix of signals? The browser extension angle is interesting, but I’d want to understand how it avoids hiding legitimate third-party sellers or small brands that just don’t have much review history yet.
Knockoff
@crystalmei Three signals. A register of 5,500+ established brands: names on it are left alone. A linguistic score for unknown names, tuned to the signature of trademark-squat pseudo-brands (ALL-CAPS strings, vanishing vowels, improbable consonant runs). And a community list built from user reports, which reaches every install within a day. Verdicts are heuristics plus lists.
Certainly not perfect, but it's improving every day (thanks to thousands of community contributions).
Love the concept! It's becoming increasingly difficult to tell which brands are legitimate on large marketplaces. Curious...how do you decide whether a brand gets a green check or a warning?
Nice work! 👏
Knockoff
@worksforme Full breakdown of how it works in the OSS repo: https://github.com/Shpigford/knockoff#how-it-works
Super clean UI! I've been looking for a reliable Fakespot alternative and this seems perfect. Quick question: How often is the brand register refreshed? Rooting for you guys! 🚀
Knockoff
@derrickzhangdev Brand data is updated automatically every 24 hours!
the community-report list is the part I'd want to stress test before trusting it - a crowdsourced override is great against false positives, but it's also exactly the surface a squatter would try to game by having their own listing mass-"reported" as legit from throwaway accounts. is there any friction on that path, like report weight tied to account age/history, or can a fresh brand get whitelisted purely on volume of reports
Counterfeits are the marketplace problem nobody fully solves. How does Knockoff actually decide something's a knockoff, is it seller signals and review patterns, or something on the listing itself? The tricky case I always hit is the gray-market real product sold through a sketchy third party, technically genuine but you'd never know if it's legit stock. Curious where the line lands for you on that one.
The browser-extension angle makes a lot of sense here because the trust problem happens right inside the shopping workflow. One thing I’d be curious about: when the signal is not obvious, do you show users why a brand was filtered or flagged, or keep the decision mostly automatic?
Knockoff
@yaroslav_stelmakh Yes, we given an explanation via the extension.
the linguistic scoring for pseudo-brand names is a neat idea but I'd worry about false positives - plenty of small legit brands from non-english-speaking founders end up with names that sound machine generated by accident. does the community-report list act as an override that can clear a brand the linguistic score flagged, or does a bad linguistic score cap how high a brand can rank even after being reported as legit