The Tyre Reviews Score, rebuilt

From September 2026 the Tyre Reviews Score and the category bars on every tyre page are calculated a new way. The tests, the data and the owner reviews behind them are the same. What changed is how a result in one test is compared with a result in another.

Updated October 2026: a category no longer favours tyres just because they have subjective ratings; a typing error can no longer shift a whole test; results from groups of tests that share no tyre with the rest, such as the North American UHP all-season tests, now count instead of being left out; van tyres and UHP all-season tyres are ranked among their own kind; and the first and last tyres of a small group are no longer scored as if they were the best and worst on the site. On 2 October: summer road tyres stopped being rated off road, and the confidence rules began saying how tyres scored on owner reviews alone are graded.

What changed

• Category bars are percentiles. A Wet score of 90 now means the tyre out-performs 90% of the tyres of its season and class we hold test data for. Before, it meant the tyre sat 90% of the way from the worst tyre to the best tyre in the tests it happened to be in.
• Results are compared across tests. Two tyres that never met in the same test can now be compared, through the tyres their tests share. A tyre that finished mid-field in a strong test is no longer marked down for the company it kept.
• Older tests count for less, not older owner reviews. Owner reviews were being aged in a way that lowered the review score itself. Now age lowers how much weight the reviews carry, and never the score.
• A tyre is scored against its own kind. Summer, all season and winter tyres are ranked among themselves, and within winter, studded and Nordic tyres among themselves. Competition, all-terrain, van and UHP all-season tyres have their own groups. A summer tyre is never rated for snow or ice, and a summer road tyre is never rated off road: the gravel or grass leg of an SUV test says nothing about why anyone buys one.
• A high score needs three tests. A tyre with fewer than three independent tests cannot score above 8.5, however well it did.
• One component was dropped. A "consistency" term that rewarded being equally average everywhere has gone.

Why

The old category scores were built one test at a time. The best tyre in each test scored 100, the worst scored 40, and everything else was placed in between. That is fair inside one test and meaningless between two. A tyre that lost a head-to-head against the best tyre in the world by 3% scored 40. A tyre that won a weak test by a whisker scored 100. And because tests almost never mix seasons, a summer tyre's 85 and an all-season tyre's 85 said nothing about each other, which a data customer of ours was the first to point out.

The overall score inherited the same blind spot. Being behind the winner was penalised three times over, without ever asking how strong the winner was, and a tyre that won its tests in 2016 kept the marks it earned against 2016 rivals.

How the new score works

1. The bridge. For every measured result we fit a simple model: the result depends on the test (its car, surface, speed, weather and field) plus the tyre. Because tyres like the PremiumContact 7 or the CrossClimate 2 appear in dozens of tests, every test they sit in lands on one common scale. For wet braking, 98% of summer tests are linked this way, and the few tests that mix seasons link nearly the whole catalogue. Only named tyres link tests: the unnamed reference tyres some magazines include are a different tyre each time, so they link nothing. A result far out of line with what the rest of the data predict, such as a typing error, counts for less, so it cannot shift its whole test.
2. Percentages, not points. Measured results are compared as ratios, so "brakes 6% shorter" transfers between a test from 80 km/h and one from 100 km/h. Point-scored disciplines such as subjective handling are compared as they are.
3. Ranking within the group. Inside a category, each kind of measurement is first compared with the typical tyre of the group, so a tyre is not lifted or held back by which measurements its tests happened to include. Each tyre's effect is then ranked among the tyres of its season and class, and that rank is the category score you see. The overall score weights the categories the way the site always has, with wet and dry counting most, and snow and ice mattering more for winter tyres and for studded and Nordic tyres most of all.
4. Knowing what we do not know. Every estimate carries an uncertainty. A tyre with one test is pulled toward its predecessor's score, or toward the middle of its group, until more tests arrive. A main discipline a tyre has never been tested in (wet and dry, plus snow and ice for all season and winter tyres) is assumed average, with full uncertainty, rather than ignored. Confidence, shown next to every score, is how much of the final number the tyre's own data decided.
5. Owner reviews. Owner ratings are ranked among tyres of the same group in the same way and blended in at 15%, scaled down as the reviews age.
6. The 0 to 10 scale. The rank is turned into a score on a fixed curve, set once from the distribution of scores the site showed before the change, so a 9.5 is as rare as it always was and a 7.3 still means an ordinary tyre. The curve does not move as tyres come and go.

Reading the numbers

• Category score 90: Better than 90% of the tyres of its season and class that we have test data for, in that discipline, across all their tests.
• Tyre Reviews Score 9.0 and above: Roughly the top 10% of its group. A 9.5 is the top 3 or 4%.
• Confidence: High: at least three tests, with the tyre's own results deciding 90% or more of the figure. Medium: at least two tests and 70% or more. Limited: anything less; expect the number to move as data arrives. A tyre scored on owner reviews alone is graded by how much the reviews decide instead: Medium from 35 reviews, High from 135.
• Across seasons: Scores are within class. A summer tyre's 9.2 and a winter tyre's 9.2 say each is near the top of its own kind, not that they would perform alike on the same day.

What moved

Every score changed a little, because every score was recalculated. This table is live. For the twenty most-viewed tyres of the last three months it shows the score under the previous method and under the new one.

The tyres at the very top are now spread out rather than tied. Several tyres that sat at 9.8 and 9.9 now sit between 9.3 and 9.7, in an order decided by comfort, noise and value, since they are within a few percent of each other in wet and dry.

Tyres that were tested only in strong company gain. Tyres that were tested once and won a weak field lose their 100s. Tyres whose tests are all several years old are re-measured against today's field.

How we checked it

The honest test of a score is whether it predicts a test it has not seen. We removed a test from the data, recalculated everything without it, and asked how well each method predicted the finishing order of that test. On the 2025 Auto Bild summer test the new method, which had never seen the test, matched the actual order better than the old method, which had. The same held for a 52-tyre braking test. That check now runs after every update.

What has not changed

The sources: independent tests from the publications we have always used, our own tests, and vetted owner reviews. The categories. The weighting between wet, dry, comfort, value, snow and ice. The scale. Every result behind every score is still listed on the tyre page.

Questions

The method is documented in full for our data customers in the tyre.info methodology, and the old and new numbers are both published in the data API. If a score looks wrong to you, tell us; the tyre page lists every test behind it.