How the scores work — and how to read them honestly

A score you can't see into is just another opinion with a number on it. So here's exactly how the number is made — enough to trust it, and enough to argue with it.

The honest one-paragraph answer. Each option carries sourced facts (objective where possible). You set how much each value matters to you. The score is the transparent combination — facts × your weights — gently pulled toward a neutral 50 wherever we're missing data, so we never fake confidence. The score is yours, not a verdict: change your values and it changes. We never rank by who pays.

Your values, in eight themes

When you set what matters, you're weighting eight universal value-themes. They travel with you across every category:

Set these once and the whole commons re-ranks for you. A category's own sliders just fine-tune within that.

Two honest rules about the weights

• Weights are trade-offs, not wishes. Sliding "Planet" up means you'll accept a little less on the other things to get it. You can't max everything — that's not a bug, it's what choosing means.
• A high overall can still hide one weak axis. Because the score is a weighted average, a great result can mask a terrible single fact. So we flag the weakest axis — and if you weight something heavily and an option scores catastrophically low on it, we cap the score (a dealbreaker can't be bought back by good fees).

Where each fact comes from — the three tiers

Not all facts are equally solid, so we tag every one:

• 📊 Measured — an objective open-data fact (a Nutri-Score, an iFixit repairability score, a real price). Hard.
• 🏷️ Certified — a third-party certification (Fair Trade, Leaping Bunny, B Corp, FSC). Verified by someone independent.
• ✍️ Assessed — our researched judgement from public sources (a bank's fossil financing, a brand's labour record). Because this is judgement, not measurement, we show it as a coarse band — Strong / Good / Fair / Limited / Poor — never a false two-digit number, and every claim carries a source link and a date so you can check it, and dispute it.

This lets you trust the hard facts and weigh the judgements — and it keeps us honest that some of this is curation, not measurement. When a rating is wrong, that's a bug in our honesty: use "dispute or correct a rating" on any entry.

When we don't know

Missing data is shown as missing — never guessed, never a zero. If we lack a fact you care about, the score is pulled toward a neutral 50 and labeled ("based on 4 of 7 facts you value"), and a low-confidence number is muted. We would rather say "we don't know yet" than fake a clean answer.

What the number is — and isn't

• It is a transparent, values-relative fit — "Strong fit for your values," not "good product."
• It isn't a universal verdict, and small gaps aren't meaningful — two options a few points apart are effectively tied. Lead with the tier, not the decimal.
• We never rank by who pays. There are no ads, no sponsorships, no pay-to-rank. The facts come from open data and public sources; you can correct them.

You can argue with it

Every score is meant to be contestable. If a fact is wrong or a judgment is off, that's a bug in our honesty — tell us via Contribute, or (for food/beauty) fix it at the open database the data comes from, where your correction helps everyone. The goal is a commons whose ratings get truer over time because people argue them into shape.

See it in action — explore the map, or start with the primer.