Criteria are defined before testing begins
For each product class we publish the assessment criteria and their relative weight before any unit is evaluated. Criteria are derived from how the product is actually used over time, not from the specification sheet. Setting them first is what prevents a review from being retro-fitted to a conclusion.
What we assess
- Performance under sustained, realistic load rather than short benchmark bursts.
- Build quality, serviceability and the availability and price of replacement parts.
- Total cost of ownership: purchase price, consumables, subscription requirements and expected lifespan.
- Software and support commitments, including the stated duration of security and feature updates.
- Data handling, where the product collects personal data, and what the default settings are.
- Accessibility and the quality of the out-of-box experience for a non-expert.
Scoring
Scores are only published where a product has completed the full protocol for its class. Where a product has been used but not fully tested, we say so and publish impressions without a score. A product we have not handled is never scored, ranked or described as recommended.
Scores are comparative within a class and within a review cycle. When a class is re-tested, previous scores are revisited rather than carried forward.
Samples, loans and purchases
We buy review units where practical. Where a manufacturer provides a loan or sample, that is disclosed in the review, the unit is returned or its retention disclosed, and no condition of any kind is accepted in exchange. We do not accept paid travel, paid placement, or copy approval.
Corrections and re-testing
Reviews carry a last-tested date. Where a firmware or software update materially changes behaviour, we re-test and update the review with a dated note rather than silently editing it.