How we test.

Every product promises something. This page is how we check those promises against the actual thing — and how we decide how sure we’re allowed to sound about it.

What “tested” means here

It means the thing came out of the box, got set up, and got used. Not unpacked for photos and put away.

What it doesn’t mean is that I keep everything forever, or that every product is one I’d buy for myself. Plenty of them aren’t made for me at all. So each file says who it was tested as — the kind of buyer I had in mind while using it, whether that’s someone setting up a first desk, a parent buying for a teenager, or somebody who needs it to survive a work van.

That’s a smaller promise than “this is my daily driver,” and a more useful one. A product can be wrong for me and exactly right for you, and you can’t tell which from a review that never says who it was judged for.

How we know what we know
MeasuredWe put a number on it — a reading, a temperature, a stopwatch. The strongest thing on the page, and the rarest.
ObservedWe saw it happen while using the thing. No number, but first-hand and repeatable.
EstimatedA reasonable guess based on what we did measure. It always says what it's guessing from, and it's the first thing an update overturns.
OpinionHow something felt to use. Worth saying, clearly labelled, and it never counts toward how sure we are.

What the maker says about its own product is kept separate from all of this. Claims appear on the page as claims — marked verified, partly true, or not checked — and nothing becomes a fact here without one of the four labels above attached to it.

How sure we are, and why
Limited evidenceEarly days. Not much recorded yet, and nothing measured.
ModerateEnough to be useful. Not enough to be certain.
HighSeveral measurements, something to compare it against, and months of use behind it.
Very highOwned a long time, with notes added along the way — not just time on the shelf.

That rating isn’t a judgement call. It’s worked out from what we actually wrote down: how many findings there are, how many of them were measured, whether we had something to compare it against, how many photos we took, and when we last added anything.

Owning something longer doesn’t make us more sure of it. If a thing sat in a drawer after the first two weeks, that’s two weeks of evidence no matter how long ago we bought it. The only way a file gets more confident is if someone goes back and records something new.

Annoying by accident, or annoying on purpose
Nobody chose itThe plain fault. It's there because no one checked — a cable that's too short, a port that won't seat properly.
Chosen, and it's for youNot a fault at all. A latch that takes two hands so a battery never drops mid-job is good design, and gets credited.
Chosen, and it's for themBuilt to suit the seller rather than the owner: parts that only work if they're the maker's own, a feature that stops when you cancel a subscription, glue where screws would do.

The difference that matters is whether somebody decided it. We can’t read a company’s mind, so we don’t pretend to — we go on what’s visible, and we say what made us think so. Does the problem disappear the moment you buy the maker’s own accessory? Does it only happen with other brands’ parts? Is it protecting you from something real? Those are things you can check, and the reason is printed right there on the page.

When a file gets updated

Only when something actually changed. It broke, we got something wrong, the maker altered the design, or we found out it works with something we didn’t expect. “Still fine” doesn’t get written up, because that isn’t news.

When we get something wrong, the correction goes on the same page as the mistake and says what we said before. A file with no updates means nothing worth telling you about has happened.

What this site won’t do
No best-of lists

We only write about things that came through the door — bought, or sent free through Vine. That’s not a fair basis for ranking anything, so we don’t.

No score out of ten

Nothing here gets squashed into a number. How sure we are and how good a product is are two different questions, and only one of them can be measured.

No hoops

No pop-ups, no email wall, no five paragraphs before the answer. If the point is saving you hassle, the site can’t add any.