How we research, evaluate, and write

How we choose what to cover

We start from what readers actually buy and ask about: the workhorse gear in the middle of the market, the standout upgrades, and the budget options worth knowing about. When a pick draws on first-hand use -- a product we own or have spent real time with -- we say so; otherwise we're clear that the evaluation is research-led.

How we evaluate

Each pick is built from the evidence that actually predicts whether you'll be happy with it: the manufacturer's real specifications, the most credible independent and professional testing, and the consensus of long-term owner feedback -- cross-checked against each other and run through the same framework every time.

For a display that means panel type, resolution and refresh, and the colour and uniformity measured by trusted reviewers; for a keyboard, switch feel and stabilisation; for a chair or desk, adjustability, stability, and build. Where we've used a product ourselves, we lean on that and say so. Where the evidence is thin or the reviewers disagree, we tell you that rather than inventing certainty -- and we never present a number we didn't run or can't source.

How we use AI — and where humans decide

Setup Quarterly is a small operation, so we use AI tools to work faster: to gather background research, pull specs together, draft sections, and keep our reviews structured and consistent. We're saying so plainly, because a review site that hides it hasn't earned the trust it's asking for.

What AI does not do here is decide. Every pick, every score, and every number we publish is reviewed and signed off by a human before it goes live — nothing is auto-published unread. Where we cite a figure, it comes from real testing or a source you can verify, never a number a model invented. If we can't stand behind a claim, it doesn't ship.

How we score

Each pick carries a Setup Quarterly editorial score out of five and a plain-language verdict. To be clear about what that number is and isn't: we do not run first-hand lab benchmarks. Our score is an editorial synthesis, built from three weighted inputs:

• Independent expert reviews. We read the credible testing outlets for each category -- RTINGS, Tom's Hardware, TechRadar, PC Gamer, TechPowerUp, GamersNexus, Trusted Reviews and their peers -- and weight their measured results and verdicts, cited per pick.
• Manufacturer specifications, read critically: where a vendor omits a figure or a spec flatters the marketing more than the buyer, we say so rather than repeat it.
• Long-term owner consensus -- the reliability, quirks and buyer's remorse that only show up after months of real use.

The score reflects value at the product's real price and tier -- a great budget board and a great flagship can both score highly, because we judge each against what it's trying to be and who it's for. It is our editorial verdict, not a lab measurement, and the number shown on the page is the exact rating we stand behind. The "recommended for" and "skip if" lines do what a single number can't: tell you whether it's right for you.

Prices, alternatives, and affiliate links

Prices and availability change constantly, so we describe the tier and link out for the current price rather than hard-coding a figure that goes stale. Most reviews include an "alternatives we considered" section so you can see what we weighed it against. Affiliate links fund our work but never change the verdict; see our disclosure for the full details.

Updates and revisits

A review is a snapshot, and we date every one. When a firmware update, a price change, a recall, or long-term use changes our view, we update the review and note when and why. Material corrections are logged in our corrections policy.