How we test
We have not scored anything yet, and this page says so. Here is the standard a score will have to meet before one appears.
As of today, bitcritiq has published no scored reviews. Nothing here carries a number that came from us using a product, because we have not used one — and every article that discusses products says so on the page itself.
This page previously described the process below in the present tense, as though it were happening. It was not. A page describing work nobody has done is exactly the sort of thing this site exists to complain about, so it now reads as what it is: the standard a review will have to meet before one appears. Hold us to it on the day it does.
What we publish instead, today
Shortlists aggregate published scores from named outlets. Every contributing score is shown with its source, and the one number on the page is an average of other people’s testing, labelled as such.
Explainers and guides describe how something works, checked against the manufacturer’s or the standards body’s own documentation, and end with something you can verify yourself.
Neither can carry a bitcritiq score, and that is enforced rather than intended:
allowsScore is false for both formats, so the publishing tool rejects a score
on them.
The standard, for when we do test
A score means a person used the thing. Not a specification comparison, not a summary of other people’s reviews — a named human being with the product in front of them.
The basics, on everything
Every reviewed product will be used as its owner would use it, for at least two weeks, as a primary device wherever that is possible. We do not score anything we have had for a weekend.
Every review states: how many days we had it, what we did with it, where the unit came from, who paid, and what we could not test. If we could not test something important, we say so instead of quietly omitting it.
Where units come from
- Bought — we paid retail. Always our preference.
- Loaner — supplied by the manufacturer and returned afterwards.
- Sample, kept — supplied and not returned. Disclosed on the article.
- Not supplied — we could not obtain one. Then it is not a review, and it carries no score.
No arrangement includes any condition about what we may conclude. We decline those.
Phones and laptops
Battery is measured as screen-on time under a fixed workload at a fixed brightness, run to empty at least twice, and reported as the average. We report the conditions alongside the number, because a battery figure without conditions is meaningless.
Sustained performance is measured over at least twenty minutes, because the interesting number is not the first run — it is what happens after the chassis gets hot. We report the drop from peak, not just the peak.
Audio
We listen across a fixed set of material we know well, on more than one source device. Where independent measurement data exists, we cite it rather than pretending our ears are an instrument.
Smart home
Tested on a real network in a real home, including the part everyone skips: what happens when the internet drops, and what still works locally.
Scoring
| Score | Verdict | What it means |
|---|---|---|
| 8.0–10 | Buy | We would spend our own money on it |
| 6.0–7.9 | Wait | Good, but wrong price or wrong moment |
| Below 6.0 | Skip | Something else is better for the money |
Scores are not curved and are not adjusted after publication except by correction. Price is part of the score: the same product can be a Buy at one price and a Skip at another, and we say which.
Who may write a score
Only a person. Our AI analysts cannot author anything carrying a score, and that is enforced in the publishing tool rather than left to good intentions — the software rejects a scored article with an AI byline. Experience means first-hand use, and an AI persona has none. See the editorial policy for exactly where that line sits.