Keep in mind that both sites are estimates - there's absolutely no way to determine which reviews are "fake" with 100% accuracy. The difference with ReviewMeta is that you can actually see the numbers behind our calculation and even open the hood and turn the knobs yourself.
I also agree that it should be very easy for Amazon to pick up on. I can't believe it's happening. We call this "Review Hijacking" and have a warning that appears at the top of our report if we detect it: https://reviewmeta.com/blog/amazon-review-hijacking/
We've been noticing a big surge in the number of really inauthentic looking reviews - I'm talking 1,000 written over just a few days, 100% unverified, and some obvious manipulation going on with the product versions. Here are a few examples. I just don't understand how Amazon isn't catching this low-hanging fruit.
We call this "Review Hijacking" and have a bit alert that pops up on our report when detected. It's not a bug, it's a loophole that sellers are exploiting. They are taking advantage of the "product variations" feature that's supposed to be used for listing products of different styles together in one listing (size, color, format, etc).
However these sellers are essentially stealing old, expired listings of products and collecting all the reviews associated with them to pump up their rating and review count.
This has been going on for over a year. I really don't understand how Amazon hasn't figured out a way to close the loophole at this point...
The challenge here is that there's no answer key, so nobody can really know for sure which reviews are authentic and which are not.
We all can look at reviews and make our best guess, but at the end of the day, it's always going to still be a guess. At ReviewMeta, we aren't claiming to know all the answers. We try our best to ensure everyone knows it's an estimate (you're actually forced to click an I agree button before accessing the site) and we also provide all our work so you can use the numbers to help you make a more informed decision.
That's not exactly how it would work. As stated on the reports, deleted reviews don't actually affect our adjusted rating or pass/fail/warn result.
Technically speaking, of course ReviewMeta can be "gamed", but it would require much more effort on the part of the seller to generate reviews in such a way that it wouldn't set off any warnings on our side. Basically force them to be very deliberate in their efforts. It would just add yet another hoop for sellers to jump through, making it that much more difficult on their part.
Just wanted to chime in about this issue. We call it "Review Hijacking" and we've been seeing it more and more. We're actually starting to keep a history of any product title changes and also show you the different product variations that are being reviewed. On your second example, (https://reviewmeta.com/amazon/B07FSW8VMQ), you can click the link under the picture that says "See 3 product variations" and it will show you exactly the breakdown of reviews that come from a log carrier bag and a nut milk bag...
The first example is a little trickier since the first time we analyzed the product, the title had already been changed. (You can click on the "Title History" link for any product).
We're working on being able to detect when these changes/variations are normal or being used to deceive shoppers. It's extremely easy for a human to immediately tell, but much harder to train a computer to do it. For example, a seller might have a t-shirt with different product variations: "Large Blue", "Medium Green". These are very different words, but obviously a legit reason to have a bunch of variations.