How we score: the Courtesy Score, v1.1

A lifetime object is not a well-made object. It is an object whose failure has somewhere to go. The Courtesy Score is a 0–100 measure of how much second life a thing has in it, computed from six properties a shopper can check by hand — in a store, before paying, without naming a single brand. Everything on this page is the complete method: the weights, the caps, the classes, and the rule that decides when a number is honest enough to show.

The six pillars and their weights

Each pillar is scored 0–10 and the weighted sum is multiplied by ten. The weights are not invented here: they follow the published French durability index, where repairability carries the same weight as reliability and the guarantee term carries one fifth of either — the same 5-to-1 ratio you see between the top two pillars and the last one below.

Three findings cap the score on their own

Averages hide fatal gaps: independent counter-testing of the French repairability index found products scoring above 7 while a spare-parts criterion sat at zero, because five criteria were averaged. The Courtesy Score refuses that failure with hard ceilings. A cap fires only on positive knowledge — an unknown never triggers one — and when more than one applies, the strictest ceiling holds.

• 40

  The energy store is glued or soldered in, with no documented replacement path. A glued-in cell sets the lifespan of everything attached to it.

• 55

  No spare part of any kind is sold for the object. An unrepairable object cannot sit in a top band, whatever it is made of.

• 70

  The core function depends on a remote server the owner does not control. The lifespan then belongs to a company's roadmap, not the object.

Object classes: the questions change with the object

A cast iron skillet has no battery and a wool blanket has no fasteners, so a single universal question set would grade objects on properties they cannot have. Real scoring systems already handle this — the French index publishes a separate calculation grid per product family — and the Courtesy Score does the same with five object classes. A question that does not apply is dropped from its pillar's denominator; a whole pillar that does not apply is dropped and the remaining weights renormalize. Not applicable is never a zero and never a free ten.

Worked examples, from the live tables

These four numbers are read from the same data the category pages render, so this page can never disagree with the index. The pair of bottles is the method in miniature: the single-wall bottle scores 90.5 because it has no vacuum to lose, while the vacuum-insulated bottle — a genuinely excellent object — scores 81.6 because it carries one designed failure that no one can repair. Cast iron plates score 96.0: a monolithic object whose parts pillar is not applicable, with everything that remains near perfect. And enamel and thick glass mugs score 64.0 — the honest middle of the scale, an object that resists damage superbly and cannot be repaired from it at all.

Why Pillar 6 is provisional on every item page

The Guarantee pillar is a fact about a maker: the written term, the claim process, the exclusions, and whether parts are sold to people out of warranty. This index names no makers — that is a founding rule, because properties outlive brands — so Pillar 6 cannot be honestly scored at the item-type level. Every score in the index therefore carries a visible Provisional label, the pillar is left unscored, the weights renormalize around it, and the pillar's note tells you exactly what to ask the specific maker in front of you. A provisional label shown is a method working; a guessed number would be the method failing.

What a photograph cannot see

A photo can identify a thing; it cannot score one. Run the rubric against what an image physically contains: about a third of the score's weight is partially visible — stitching, screw heads, a battery door, a laminate edge — as evidence, never proof. The remaining two thirds is not in a photograph under any circumstance: every parts question, every service question, every warranty question, and whether repair documentation is free. Whether a maker still sells the gasket in year eight lives in a parts catalog, and no camera reaches it. That is why there is no photo-scoring feature here and will not be one.

The display rule

One rule decides when a number appears: the score is shown only when what remains unknown could not move the result into a different band. On these pages every published score is complete across its applicable pillars except the provisional sixth, and the renormalized math makes that explicit rather than hidden. The same rule governs the Courtesy Check, the self-assessment companion page: unanswered questions produce a band range and the next questions to answer, never a false number.

What the score is not

It is not a product rating — it scores item types, and no product or brand is named anywhere in the index. It is not an endorsement of anything for sale, and it carries no prices and no star ratings. It is not a warranty, a guarantee of any individual object's life, or a substitute for the two questions only a specific maker can answer. And it is not a community average: nothing on these pages aggregates anyone else's submitted opinions.

Precedents this method draws on

The weights follow the French durability index, a government score displayed beside the price that weights repairability equal to reliability with the guarantee term at one fifth. The refusal to compute with missing data follows Nutri-Score practice as implemented by Open Food Facts, which names the missing fields instead of guessing. The versioned, published-criteria approach follows iFixit's repairability rubric, which puts its point allocations in public where they can be argued with. Sources: the French durability index, Nutri-Score at Open Food Facts, iFixit's scoring criteria, and the Open Repair Alliance dataset.

Version and change log

• v1.1 — September 15, 2026

  Published with the index. Adds object classes with not-applicable handling, the renormalization rule, the provisional treatment of Pillar 6 at type level, and the display rule. First scored groups shipped.

• v1.0 — September 11, 2026

  Internal draft: six pillars, the 25/25/20/15/10/5 weights, the three caps, and the five bands. Never published; v1.1 supersedes it.