How the desk scores a product.
One rubric, five axes, published weights. The same model and the same anchors are applied to every specimen, so a score means the same thing across the index.
The five axes
Performance
30%The measurement axis. Does it do the job it was sold to do, at the top of its class, and does it hold that under load, heat, distance or repetition?
Build
25%Materials, tolerances, serviceability and the parts that will fail first. We score what we can hold, and we say so when a verdict leans on teardown evidence rather than our own bench.
Value
20%Score against the current lowest verified retail price, not the launch price. When the price moves, this axis and the composite move with it.
Design
15%Form, interface, legibility, and whether the object makes its own operation obvious. Novelty earns nothing; coherence earns a lot.
Ease
10%Daily friction over fourteen days of ordinary use: setup, cleaning, storage, software, the small tax of living with the thing.
Reading a score
Two rules we do not break
Commission is invisible to the score
Retail links on this site are affiliate links and they do pay us. The scoring model never sees them. The price matrix is generated from retailer listings, and a product with no affiliate programme is scored exactly the same way.
A correction is published, not deleted
If a measurement is wrong we re-run it, change the score, and note what changed. Reviews are living documents, not launch-day marketing.