Why brands, labs and photos disagree about the same shoe

Brands rarely say where they measure. The heel can be measured at the back edge or under the heel bone, the forefoot under the ball of the foot or at its widest point, with or without the insole, which alone is often a few millimetres thick. Labs measure at fixed points along the shoe, usually with the insole in, so their numbers are consistent across shoes but are not the brand's numbers.

In our check of 131 shoes, the stacks shown here (mostly brand figures) and an independent lab's measurements typically differed by about 1.3 mm at the heel and 1.7 mm at the forefoot, and occasionally by 4–5 mm. Drop is the difference of two such numbers, so 1 mm off at each end can move it by 2 mm.

A published weight is for one shoe in one size, usually a men's US 9 or US 10, and brands round. Labs weigh the size they tested, and two pairs of the same model can differ by a few grams. On shoe pages the size is shown next to the weight when the source gives it.

There is no standard definition of where a rocker starts or how high the toe is. Our photo estimates measure the outline of the unloaded sole in side photos, with the toe tip taken at 97 % of the length for a 300 mm shoe, so they sit lower than a lab's toe spring measured at the very tip. What matters is whether they rank shoes the same way: against a lab across about 110 shoes, toe tip height and rise angle did well, toe rocker only for clear differences (typically about 3 points apart), and heel bevel roughly.

Photos differ too: the camera angle, the lens and how the shoe was placed all move the outline a little. Each shoe page shows the median of all its photos and marks a value with * when the photos disagree more than usual.

Bending stiffness depends on how far and where the shoe is bent; firmness depends on the durometer scale (Asker C and Shore A cannot be converted into each other) and on temperature. Our 0–100 scores rescale one lab's readings so shoes can be ranked against each other; they are not forces or hardness values, and the raw reading is shown on each shoe page.

When several values exist, the shoe page shows one by precedence: a lab measurement, then the brand's figure, then a value we derive, then a photo estimate; within each, the higher confidence and then the most recent. Nothing is averaged and nothing is deleted: every recorded value is listed under Technical detail on the shoe page with its source and date, and the change history shows what replaced what.

When a shoe's heel stack, forefoot stack and drop do not add up, because they come from different sources, the page says so. When comparing shoes, differences smaller than the precision of the sources (about 1 mm of stack, 8 g of weight, or the spread between photos) are treated as about the same.