How We Test

One pick per component, argued rather than guessed. This page explains exactly how a recommendation gets made, what our scores mean, and — just as importantly — what we do not do.

What we are, and what we are not

We are not an independent test lab, and we do not pretend to be. Buying every graphics card in a category, running each for weeks and controlling for every variable is a multi-million-pound operation, and any site that implies otherwise while reviewing dozens of products a year is overstating itself.

What we do is narrower and, we think, more honest: we do the cross-referencing that nobody enjoys doing. The performance data needed to choose a component already exists, produced by labs and reviewers who run controlled tests properly. What almost none of it accounts for is what the part costs where you live, this week. That gap is the work.

How a pick gets made

1. Shortlist. Establish which products in a category are genuinely worth considering this quarter — not every product that exists, and not last year’s winners by default.
2. Research. Pull performance data for each from established, independently verifiable sources. Where a specific claim comes from a specific source, the guide says so.
3. Compare. Set each option against its direct competitors, rather than against its own marketing material.
4. Price. Check what each actually costs at retail right now, in the reader’s market — not the RRP from launch day.
5. Recommend. Weigh performance against price for a stated use case, name one winner, and state the trade-offs plainly instead of burying them.

What our scores mean

A score on this site is an editorial judgement on a 0–5 scale. It is not a measured lab result, and it should not be read as one.

Where a guide carries a score breakdown, the overall figure is a weighted average of four criteria:

Performance and value carry the heaviest weighting because, for the overwhelming majority of buyers, those two are what decide the recommendation. A beautifully built component that loses on price for its performance tier is still the wrong purchase for most people, and we would rather say so.

Two deliberate rules about scores:

• An unscored guide stays unscored. Where a score is absent, it means we have not scored that guide — not that it scored zero. We would rather show nothing than a number we cannot defend.
• A partial breakdown is shown as partial. If only some criteria have been assessed, the average is calculated across those criteria only. We do not pad the remainder with zeros to make the maths look complete.

Hands-on versus sourced

This is the distinction most review sites blur, so we state it explicitly: where a product has genuinely been handled in person, the guide says so. Where it has not, the guide does not imply otherwise. A judgement drawn from someone else’s measurements is still a useful judgement — it is just a different kind of claim, and you are entitled to know which one you are reading.

Pricing

Prices move week to week, which is why value is judged against what something costs now rather than at launch. It also means any price quoted in a guide is a snapshot. Product links route to the closest regional store so the figure you land on is one you can actually pay, in your own currency, from a retailer that will ship to you.

Corrections

Guides are revisited as parts launch, get discontinued or get repriced, and any revised article shows an “Updated” date alongside its original publication date. When we get something wrong, we would rather hear it from you than leave it standing: tell us which guide and what changed. That is the most useful message you can send us, and it is the fastest route to a fix.

Our commercial arrangements never affect any of the above. See the editorial policy for how that independence is maintained, and the affiliate disclosure for how the site is funded.