Same body. Same-looking jeans. One fits at 28, the other at 29. The size chart swears they're identical. They're not, and the difference is physics. One pair is raw selvedge denim with almost no give. The other is stretch denim that forgives the thigh. A size chart is four flat numbers with no idea which fabric it's describing, so it ships the wrong size in one of them. We built Contour to feel the fabric the chart can't. For a body and a garment it computes the strain at each zone (waist, hip, thigh) against that fabric's stretch limit. Run the shopper Riley through it: a 29 in the raw selvedge, because at 28 the thigh blows past the denim's limit, and a 28 in the stretch pair, because the fabric gives. Same body, two correct answers. The size is decided in plain, unit-tested code, not the language model. Agents advise, code decides. On a labeled 105-pair synthetic golden set, that engine beats the waist-only chart most brands still use by 40 percentage points, and the gap is widest on rigid denim and tailoring, where fit bites hardest. (Riley and the jeans are synthetic test data, not a real shopper.) If you sell denim: how much of your return rate is really one shopper ordering two sizes of the same jean, keeping one, and shipping the other back? #RetailTech #FashionTech #ReturnsManagement #ReverseLogistics #ApparelRetail