Why star ratings can't tell you a brand is real

The rating is about a page, not a company

The star rating is the most-read piece of information on any listing, and the one least able to answer the question shoppers are actually asking. A high rating gets read as a statement about the maker: this is a company that does good work. But the rating is attached to a product page, and a page is something a seller creates, merges, retires and replaces. Nothing about it is tied to a business with a history.

The brands most worth being wary of are precisely the ones with least to lose by treating the page as disposable. A company that has traded for thirty years has one reputation and cannot start again. A brand registered last spring can be replaced in the time it takes to file a trademark, and the rating stays with the old page.

A disposable brand looks best just before it disappears

Consider the lifecycle. The listing goes up with a new name and an empty history. Early units are often the best units, because the seller wants the average to start high, and early buyers are frequently the least critical. For the first year the average is made almost entirely of first impressions: it arrived, it looked like the picture, it worked when switched on.

Durability failures arrive later — a handle loosens, a battery swells, a coating peels — and they arrive in a trickle. By the time they are numerous enough to move the average, the seller has seen the trend and has a replacement brand ready. So at any given moment, the disposable brands you see are the ones in their honeymoon. The ones whose reviews turned are gone. The system shows you survivors, and survivorship makes the category look better than it is.

The established brand beside them has been on the same listing for years and has collected every unit that failed in year four. Its average is lower because it is more honest, not because the product is worse.

Reviews that belong to a different product

A listing can offer several variations — sizes, colours, models — under one parent page, and Amazon pools the reviews across them. That is reasonable when the variations are the same product in different colours. It stops being reasonable when a seller attaches an unrelated product as a new variation so it inherits a history it did not earn. This has been a documented abuse for years, and Amazon has acted against it, but the mechanism survives because the legitimate use survives.

The tell is reviews that describe something other than what you are looking at. Read a handful of the older ones and notice whether they mention a different item, a different size, or features this product does not have. A related practice is keeping a listing alive across products: an old page with a long history is repurposed by changing the title, images and contents while keeping the identifier. Dated reviews that praise a thing with a different name are the sign.

Reviews that were never independent

Then there are reviews that were bought, seeded or encouraged in ways that break the assumption behind the average. Amazon prohibits incentivised reviews and has removed large numbers of them, but the practices persist in forms hard to detect from outside: refund-for-review schemes run through messaging groups, inserts offering a gift card for five stars, free units sent to reliably generous reviewers, and clusters posted in a short window at launch.

None of this needs a conspiracy theory. If the name is temporary, the only asset is the listing, and the listing is worth exactly its rating. Spending money to raise that rating is, from the seller's side, just marketing.

You cannot reliably spot a seeded review on its own. What you can notice is the shape: a high proportion of very short five-star reviews, a burst dated within days of each other, and text describing the experience of receiving the product rather than using it.

Why a big number is not a big signal

The instinct is that scale fixes this — a 4.6 from twelve thousand ratings surely cannot be engineered. But most of those ratings are not reviews. Since Amazon introduced one-tap ratings, a star can be left without writing anything, and a large count is mostly taps. It tells you a great many units shipped, which for a heavily advertised, aggressively priced listing is what you would expect. It does not tell you those buyers still own a working product, and it says nothing about who made it.

Put another way: the rating measures early satisfaction at scale, and a disposable brand is optimised for early satisfaction at scale. A number a strategy is built to maximise cannot be evidence against that strategy.

What reviews are still good for

None of this makes reviews useless. It makes the average useless for the question of whether a company exists, and puts the useful information in the text rather than the stars.

Go to the one- and two-star reviews and read for specifics. Vague complaints about shipping are noise. A description of exactly how the thing broke — the pin that sheared, the seam that split, the charger that got hot — is signal, and three reviewers describing the same failure is close to a defect report. Look at the dates: failures clustering eight to twelve months after purchase say something about durability the average cannot.

Photographs are worth more than words, because they show what arrived rather than what was promised. Reviews that mention contacting the company, and what happened, are the closest thing the page offers to evidence there is a company to contact. If you want to know whether the brand is real, though, the page will not tell you; the public record will, and our guide to checking a brand in ten minutes covers the steps.