How we test

Fews scores products for a trait, not for everyone. A fit score is 0–100 for one trait on one product, and it has to have evidence behind it before it appears anywhere.

The four kinds of evidence

• We tested — an editor with the trait used the product for a set task and time. Scissors: paper, fabric and card for ten minutes. Ladles: ten pours over a clean counter. Shoes: measured with callipers, then worn.
• From specs — a fact from the maker's documentation that decides fit on its own. A right-hand-only single-bevel blade is a hard "no" for left-handers whatever else is true.
• Users report — members with the trait told us what happened, through the "This works for me" button. Reports go into a review queue before they count.
• Brand claims — what the maker says. Shown, labelled, never trusted on its own.

How the score is built

The editor's test carries most of the weight (60%). Member reports adjust the score within fifteen points either way (25%). External signals such as return rates, when we have them, add the rest (15%). Missing parts are left out and the weights renormalised. A hard spec failure caps the score in the "Not for you" band.

Confidence is separate from the score: fewer than two pieces of evidence is "low" and never appears in a ranked list; two to five is "medium"; more than five including an editor test is "high".

Several traits at once

If you have more than one trait, we show the lowest score, not the average. A product that is 90/100 for left-handers and 30/100 for small hands is a 30 for you — one deal-breaker is a deal-breaker. The per-trait breakdown is always one tap away.

What we don't do

We don't show exact prices, because retailers change them daily. We don't rank by commission. We don't use star averages. When we don't know, the badge says "Unverified".