Methodology

A tape measure, and a hypothesis.

Most online sizing tools ask for measurements and pretend that's enough. Real fit depends on the geometry of a garment relative to the geometry of a body — plus an entire layer of intent the brand baked into the cut. Here's how we handle all three.

The numbers in a size chart aren't your body

When a brand publishes a size chart, it's usually telling you the garment's flat measurements, not the body those clothes are designed for. The gap between them — what tailors call ease — is the room you can move in, and it varies wildly. A slim shirt for a 100cm chest might measure 104cm flat. A regular shirt for the same chest might measure 110. A relaxed t-shirt, 116.

So the same body wears a different garment dimension depending on the cut. Knowing only your chest measurement doesn't tell us whether you'll like a shirt — we also need to know what kind of fit you're after, and what kind of body you're trying to fit.

The reverse-calc trick

Most people don't own a tape measure but do own clothes that fit them. That's a usable signal. If your Zara M slim shirt fits well, and we know what a Zara M slim shirt measures (104cm chest, 96cm waist), and we know how much ease Zara intends for that fit (4cm at the chest, 6cm at the waist, after correcting for Zara's tendency to run small) — then we can solve backwards for your body.

body_zone = (garment_zone + brand_sizing_bias) − ideal_ease − shape_correction

It's lossy — we end up with confidence around 80%, not the 100% a real tape measure gives. But it's enough to start. And if you give us a second reference (say, a Uniqlo regular trouser size 32), the two estimates cross-validate and confidence climbs into the high 80s.

Why we ask about body shapes

Two people with identical 96cm waists can have very different needs from the same garment. Someone with a slim build at 96cm carries that circumference evenly around the torso. Someone with a beer belly at 96cm carries it concentrated at the front, which means they need an extra ~4cm of ease at the waistband to avoid pulling — and 1.4× the usual weight on getting waist fit right, because nothing else matters if the waistband doesn't button.

Our shape modifiers do exactly this: they shift the ideal ease and the relative importance of each zone based on what kind of body you're describing. An athletic upper body weights shoulders 1.3× because a tight-shoulder shirt is unwearable. An hourglass figure weights the waist 1.3× because the silhouette only works if the garment follows the curve.

Scoring a size: the comfort window

For each size of each garment, we compute a margin per zone — how much the garment exceeds your body in centimeters — and compare it to the ease window prescribed for that fit, gender, and apparel category.

Inside the ideal plateau (centered 40% of the window): 1.0 score, marked Perfect Fit.
Within the window but off-center: 0.4–1.0 score, marked Slightly Tight or Slightly Loose.
Just past the bounds: 0–0.4 score, marked Tight or Loose.
Far outside: zero, with a 0.65× penalty applied to overall score per critical zone affected.

The penalty matters. Without it, a size that fits perfectly at the shoulders but is laughably loose at the waist could outscore a size that fits acceptably everywhere. The penalty makes balanced fit win.

Length zones are different

Sleeves, top length, and trouser inseam aren't about ease — they're about whether the garment ends in the right place on your body. Your preferences here matter: do you want sleeves that hit the wrist bone, or sit half an inch past it? Do you want a cropped trouser, no break, slight break, or full break?

Length zones are scored against your ideal length (body + the offset your preferences imply). The verdict ladder for these is Short / Perfect / Long, not Tight / Loose, because the problem is geometric, not about ease.

Confidence, surfaced honestly

Every measurement we use carries a confidence score, and the final recommendation includes the average. Direct measurements are 1.0. Reverse-calculations are 0.7–0.85. Anthropometric fallbacks (estimating thigh from hips, neck from chest) are 0.4. The system tells you, before you commit to a size, how reliable its own answer is.

Below ~65% confidence, we surface a banner suggesting you add measurements or a brand reference to improve it. We never hide uncertainty.

What this is not

This isn't a body scanner or a virtual try-on. It can't account for fabric drape, stretch, or the difference between woven cotton and a knit jersey of nominally the same dimensions. It can't see how a garment will photograph. What it can do — better than guessing, often better than the brand's own size guide — is tell you which size will fit closest to the cut the designer intended for someone with your specific body.

For tailorable adjustments (shortening sleeves, taking in a waist, hemming trousers), we surface specific suggestions on the result page so you can factor them into your decision.

Find my size →