How we rank

Method last reviewed 9 August 2026

ChoiceTally tracks 7,218 products across 201 categories. We do not test them in a lab and we do not pretend to. What we do is compare every product against the others in its own category on three things buyers can act on: how it is rated, how many people rated it, and where its price sits in the range. This page explains exactly how that works, including the part most sites leave out.

The problem with ranking by star rating

Sorting a category by star average sounds obvious and gives you a bad list. A brand-new listing with eleven five-star ratings will beat a product with 4.7 from forty thousand ratings, every time. The first is barely evidence at all; the second is about as solid as consumer data gets. Rank on the raw average and you systematically promote the least-proven products in every category.

What we do instead: a credibility-weighted rating

We use a weighted average that takes account of how much evidence sits behind a score. In plain terms: a product's own rating counts for more the more people have rated it, and counts for less when hardly anyone has. When the evidence is thin, the score is pulled towards the typical score for that category until enough ratings accumulate to justify moving away from it.

The formula, for anyone who wants it:

The effect is straightforward. A product with far more ratings than the category norm is judged almost entirely on its own score. A product with a handful is judged mostly on what is normal for its category, until it earns the right to be judged on its own. Nothing here is a secret formula: you can check any ranking on the page by hand.

Why we compare within a category, never across

A 4.3 means different things in different aisles. Some categories run hot, with almost everything above 4.5; others are full of 3.8s because the products are genuinely harder to get right. Comparing a kitchen gadget's rating with a pair of running shorts tells you nothing. Every figure on our review pages, including the "above" or "below" comparisons, is calculated against the other products in that category and nothing else.

What the badges mean, and why most products have none

Badges are earned, not decorative. A product carries Highest rated only if it comes first in its category on the weighted score with a meaningful number of ratings behind it, Top 3 only if it is genuinely in the top three, and Most reviewed only if it sits in the top few per cent of its category by review volume. The large majority of the 7,111 rated products we track carry no badge at all, which is how it should be.

Prices

Prices are fetched live from Amazon when a page loads. We never write a price into an article, because a written price is out of date within days and a stale price is worse than none. If a price cannot be fetched, the page says so rather than showing an old figure. Price positioning ("cheaper end", "top of the range") is calculated from the live spread across the category.

What we do not do

• We do not accept payment for a place on a list, a higher position, or a softer verdict. No brand can buy anything here.
• We do not label every product a "top pick". Most products are not top picks, and a badge on everything is a badge on nothing.
• We do not present Amazon's rating as if it were our own rating in structured data. It is Amazon's, it is labelled as Amazon's, and it stays that way.
• We do not adopt manufacturer marketing as fact. Where we quote what a maker says a product does, we attribute it to the maker.
• We do not invent ratings, reviews or test results. If we have not measured something, we do not report a measurement.

The honest limits of this method

Ratings measure satisfaction, not quality, and the two are related but not the same. They can be inflated by review incentives and they say nothing about how a product holds up over three years. Our data comes from Amazon's UK catalogue, so a product that is excellent but sold elsewhere will not appear here at all. And a shortlist is a starting point for a decision, not the decision itself.

We think a transparent, checkable method with stated limits is more useful than a confident-sounding verdict with none. If you think a ranking is wrong, our corrections page explains how to tell us.

How we make money

Through the Amazon Associates programme. If you buy after clicking one of our links, Amazon pays us a commission at no extra cost to you. The commission does not vary in a way that would make one recommendation more profitable than another, and rankings are produced from the data before anyone looks at what earns. Our full affiliate disclosure sets out the relationship.

Common questions