How we score products.
A Worth Score is a weighted average of six factors, each rated 0–10. No factor is secret, and no retailer can move any of them.
The six factors
Weights total 100%Price fairness
20% of the scoreWhat you pay against what comparable products cost today. A high score means you are not paying a premium for the logo.
Build quality
20% of the scoreMaterials, tolerances and how the product holds up to daily handling. Based on teardowns and long-term owner reports.
Performance
20% of the scoreHow well it does the job you bought it for, measured against the best in its category rather than in the abstract.
Longevity
20% of the scoreExpected useful life: battery degradation, software support windows, and how quickly the category moves on.
Repairability
10% of the scoreCan it be opened, are parts sold, and is a repair cheaper than a replacement. Sourced from iFixit-style teardowns.
Support & warranty
10% of the scoreWarranty length, how the manufacturer handles claims, and the cost of out-of-warranty service.
How the number is produced
Each factor is rated 0–10, multiplied by its weight above, and summed. A product scoring 9 on everything scores 9.0 overall. We publish the result to one decimal place and no further — the inputs are expert judgement calls, and a second decimal would imply a precision that does not exist.
Because the score is derived at render time from the factor ratings, the headline number can never drift out of step with the breakdown shown beside it.
Confidence
A score is only as good as the evidence under it, so every product carries a confidence label alongside it:
- High — Backed by four or more independent sources and thousands of owner reviews.
- Medium — Backed by a few sources. Treat small score gaps as noise.
- Limited — Thin evidence so far. Use this as a starting point, not a verdict.
Where the data comes from
Each entry is compiled from independent published reviews and measurements, teardown and repairability reports, manufacturer specifications and warranty terms, and aggregate owner reviews from retail listings.
Every factor links to its evidence. On a product page, expand any of the six factors in the score panel to see the published sources behind that specific rating — publisher, title and an outbound link you can check. The full list also appears at the bottom of the page.
Coverage is uneven, and we show that rather than hiding it. Some products have four independent measurements behind them; some have one. Where we have found no published source for a factor, that factor is labelled “no published source yet — this rating is our own judgement” instead of borrowing credibility from a citation that does not speak to it. The confidence badge is computed from the number of sources actually listed, so it cannot overstate what is there.
Prices come from named retailers and are stamped with the moment we checked them. We never estimate a price, and the headline “best price” is always the cheapest listing a shopper could actually buy — not an average, and not an aspiration. That means price plus delivery: a €254.99 listing with €5.90 postage loses to a €259.00 one that ships free, because the second is what costs you less.
How we make money — we don’t, yet
Brivio earns nothing today. There are no affiliate links, no sponsorships and no ads. The retailer links are searches on each shop’s own site; none of them is tracked and none of them pays us. When we do add affiliate links we will label every one of them, and the rule we are committing to now, before there is any money on the table, is that commission plays no part in the six factors, in which products we cover, or in how retailers are ordered — that ranking is stock first, then delivered price.
Retailers never influence our score. A retailer paying us more cannot rank higher, and a product sold nowhere we could ever earn from is scored exactly the same way.
Keep rate — the number we don’t have yet
Prices, specs and reviews are commodities: anyone can buy the same feeds. The one thing worth knowing that nobody publishes is whether people kept what they bought. So we ask. Mark a product as bought and we’ll ask you at six months and again at a year whether you still use it, returned it, or wish you hadn’t bought it.
We have no answers yet — collection has just started — and product pages say so plainly rather than showing a placeholder figure. We will publish a keep rate only once at least 30 people have answered for a product. A percentage drawn from three people is a made-up number, and the code enforces that floor rather than trusting us to remember it.
When the answers are bad, we will publish those too. A keep rate you can only see when it flatters the product is worth nothing.
Two limits on that number, stated now rather than when it is inconvenient. Every outcome is an unverified self-report — nobody shows us a receipt, and we have no way to confirm a purchase happened. And 30 answers is a floor, not a guarantee of accuracy: it is the point below which a percentage is certainly meaningless, not the point above which it is certainly right. Duplicate answers about the same purchase overwrite rather than accumulate, one address can file only a few reports per product per day, and answers that do not arrive through this site are stored but excluded from the count entirely.
When we are wrong
The model is published so it can be checked, which only means something if being checked can change it. Every score we have revised since publishing it — what it was, what it is now, why, and the evidence — is on the corrections page, including the ones that make a product we recommended look worse. That page is empty today, and the honest reading of that is that nobody has checked us yet.
What we do not do yet
This is a prototype. Prices are not yet live, we do not track price history, and there are no accounts — saved products and comparisons live in your browser only. We would rather show you a small honest index than a large invented one.