How We Rank
Four axes, published weights, published breakpoints, and a worked example. Every score on this site is arithmetic over numbers you can look up in the same two documents we used.
What we do not do
We have not tested any product on this site. No lab, no owned units, no hands-on. Nobody here has carried these bags, dropped them, soaked them or filled them with weights, and no page will ever imply otherwise.
That is worth stating first because the rest of this category competes on exactly that. Serious outlets buy fifty carry-ons and destroy them, and the results are genuinely useful. We cannot do it and we are not going to pretend.
What we do instead
We compile every specification the manufacturer publishes, check every published dimension against every airline’s published limit, and do the arithmetic nobody else in this category bothers with. Then we publish the formula.
The point is reproducibility. A testing claim is unverifiable — you cannot check whether someone really did drop a bag down a stairwell. A cost-per-covered-year figure is a division you can redo in ten seconds with the same spec sheet. That is a weaker kind of authority and a much more honest one.
- Products compiled
- 39
- With a published price
- 24
- Airline policies compiled
- 14
- Units we claim to have tested
- 0
The four axes and their weights
| Axis | Weight | Computed as | Inputs come from |
|---|---|---|---|
| Value | 35% | price ÷ warranty years, and price ÷ liter | Manufacturer's published price, warranty term and capacity |
| Airline-fit breadth | 25% | count of carriers whose published limit the bag satisfies on all three dimensions | Manufacturer's published dimensions; each carrier's published baggage policy |
| Carry efficiency | 20% | weight ÷ liter of published capacity | Manufacturer's published weight and capacity |
| Warranty & support | 20% | term length, plus points for published repair and airline-damage cover | Manufacturer's published warranty documentation |
Value carries the most weight because our reader has a budget ceiling and the rest of the field does not score for one. The best-known scoring system in outdoor gear publishes its four axes openly — ease of transport, design and functionality, construction quality, capacity — and price is not among them. For someone deciding between an $89.99 bag and a $390 one, that omission is the whole question.
The breakpoints, published
Each raw measure is mapped onto a 0–10 subscore against a linear scale with two published endpoints. Below the “best” figure scores 10; above the “worst” figure scores 0; everything in between interpolates.
| Measure | Scores 10 at | Scores 0 at | Direction |
|---|---|---|---|
| Cost per covered warranty year | $4/yr | $40/yr | Lower is better |
| Cost per liter of published capacity | $1.5/L | $9/L | Lower is better |
| Weight per liter of published capacity | 0.05 lb/L | 0.3 lb/L | Lower is better |
| Airline-fit breadth | Fits every carrier publishing a limit | Fits none of them | Proportional |
How the warranty axis is scored
Out of 10, in three parts:
- Term, up to 6 points. No stated end date scores 6; ten years or more scores 4; five or more scores 2.5; three or more scores 1.5; anything shorter scores 0.5; no published term scores 0.
- Published repair route, 2 points. Repair means the maker keeps a parts supply and a service operation. Replacement-only means they do not.
- Airline-damage cover, 2 points. The most likely thing to destroy a carry-on, and almost no warranty covers it. Exactly one product in our ledger publishes this cover.
The 20-year convention, and why we publish it
A cost-per-year figure needs a finite denominator, and a warranty with no stated end date does not supply one. We use 20 years.
That is a convention, not a claim. We are not saying any bag lasts twenty years, and we are not saying a manufacturer will honor a claim in year nineteen. It is a horizon chosen because it is a long ownership period for luggage, applied identically to every open-ended warranty on the site, and stated here so that anyone who disagrees can rerun the arithmetic with their own number.
When an input is not published
Manufacturers leave things out. Several bags in our ledger publish no capacity; two publish no price at all; several accessories publish no warranty term. Where that happens:
- The cell reads “Not published”, never an estimate.
- That axis is dropped for that product.
- The remaining weights are re-normalized across the axes that do have data, so a product is not penalized for a manufacturer’s silence.
- Every product page lists which axes were scored, because a score built from fewer inputs is a weaker claim and you are entitled to see that.
What we will not do is infer a capacity from the outside dimensions. The loss between box volume and real capacity runs from 7% to 26% across bags we have compiled, so an inferred figure could be wrong by a quarter. The arithmetic behind that range.
How prices work on this site
There are two different price numbers here and conflating them is the easiest mistake to make, so they are kept strictly apart.
| Published list price | Live price | |
|---|---|---|
| Where it comes from | The manufacturer's own store or spec page | A daily automated lookup |
| When it was read | 3 September 2026 | Within the last 48 hours, or it is not shown |
| What it is used for | The Value axis in the ranking | Display next to a buy button |
| Shown as today's price? | Never | Yes, with the date it was verified |
| If unavailable | "Not published", and the axis is dropped | "Check price" |
No buy button on this site will ever show a number we did not verify live. When the live layer has nothing fresh, every CTA reads Check price. A stale price is worse than no price, because it is the one thing on a page a reader acts on directly.
The airline data
Every carrier limit on this site was read from that carrier’s own published baggage policy on 3 September 2026. Each row cites the page it came from and carries the date it was checked. Where a carrier publishes no figure — and several publish no personal-item dimensions — the cell says so.
A bag counts as fitting only when its published dimensions satisfy the published limit on all three sides at once, sorted longest to longest, the way a gate sizer works. Carriers that publish no limit of the relevant kind are excluded from the denominator rather than counted as a pass, because “we do not know” is a different answer from “yes”.
The carry-on table covers 11 US carriers and three international ones. It is re-checked monthly, and the date on each row is the real date of that check rather than a build timestamp.
A worked example
Here is one real product scored end to end, computed live by the same functions that produce every score on the site — so this example cannot drift from the rankings it explains.
What changes a ranking
- A specification change. A revised model with different published dimensions or weight.
- A published price change that moves the value axis. We re-read manufacturer prices quarterly.
- A carrier policy change. Airline limits are re-checked monthly and a change ripples into every fit verdict automatically, because the score is computed rather than typed.
- A discontinuation. Products that stop being sold come out.
- A correction. If we got a number wrong, we fix it and re-date the page. The corrections policy is here.
What does not change a ranking is commission. Every roundup on this site names at least one pick we tell you not to buy, and the order is produced by the formula above rather than chosen. How we are funded.
What this method cannot tell you
Being honest about the boundaries is part of the method, so here they are.
- How it feels. Handle wobble, wheel noise, how a pack’s harness sits after four hours — none of it is published and none of it is scored.
- Component quality. No maker publishes a bearing rating or a zip cycle count. A warranty that names wheels is the closest published proxy, which is why warranty terms carry 20%.
- Real-world durability. We infer from what makers will insure, not from what happens after two hundred flights.
- Whether a spec is true. We take published figures at face value. If a manufacturer publishes a wrong dimension, our arithmetic inherits it — which is why every figure names its source, so you can check.
If the physical-testing question is what you need answered, an outlet that buys and destroys bags will serve you better on that specific point. What we offer is the arithmetic those outlets mostly do not publish, on the axes they mostly do not score.