Collect shopper measurements
One shirt, laid flat, with a bank card in frame for scale. No body photograph, nothing they have to know about themselves.
HOW IT WORKS
How ShirtMath works.
From raw measurements to real recommendations. ShirtMath's fit intelligence engine compares shopper and apparel data to find the right size, in real time.
Last verified: 25 August 2026
THE SEQUENCE, IN ORDER
- Photograph a garment you own
- Extract real dimensions
- Compare against the catalog
- Return a sized recommendation
The process, step by step
An end-to-end fit intelligence workflow
Each step has to succeed before the next is attempted. Where one cannot, the process stops and says so, rather than carrying a guess forward.
Normalize garment data
Chest, body length, shoulder, sleeve and four more points, in inches. Anything unreadable is left blank rather than filled in.
Compare against the catalog
The reference garment is measured against your own product measurements, inside your tenant and nobody else's.
Rank best-fit results
Score and rank candidates by fit, confidence, and the shopper's own reference — never by commission or popularity.
Return recommendations and insights
A size, the evidence behind it, and an explicit statement of how much that evidence is worth — including when there is not enough of it.
Why measurements
A size label is not a measurement
A size label is a name the brand chose, and nothing obliges two brands — or two cuts inside one brand — to agree on what it means. A unisex Gildan G500 in Medium publishes a 20.0″ chest measured flat; a Comfort Colors C4017 in Medium publishes 20.5″. Half an inch apart under the same letter, before manufacturing tolerance is counted.
What shoppers experience
A guided journey that builds confidence
Four moments, none of which asks them to measure themselves or know their own dimensions.
Scan or input a shirt
A shopper photographs a shirt they already own, laid flat.
ShirtMath measures
ShirtMath measures the garment — eight points, in inches. Nothing about the body is captured or inferred.
Recommended size appears
The best size shows up with clear fit notes — or an honest not-enough-information answer.
Matching products show
Recommended sizes carry across similar styles to make shopping easier.
What retailers receive
Actionable data and developer-friendly outputs
Every response is designed to be branched on.
API response
Structured responses: a recommended size with the evidence behind it, alternatives, and an explicit “not enough information” when the catalog cannot support an answer.
Fit confidence
An explainable authority on every response — why this size fits, and where the evidence runs out.
Event telemetry
Every fit interaction captured for continuous improvement: requests, responses, and measurement quality.
Usage insights
An event log of requests, matches and refusals. It carries no numbers yet, because there is no customer traffic yet.
The fit API reference covers the endpoints, the key lifecycle and the session model. Connecting your catalog covers partial imports, failed imports, deletions and disconnection. Why fit returns happen sets out where a fit return actually comes from, and the ShirtMath for Brands overview is the shortest version of the whole thing.
Build fit experiences shoppers trust.
Explore the docs or book a demo to see ShirtMath in action. The consumer site runs the same engine, on the same measurement rules, and the same honest-failure behaviour.
Book a demo Scan a shirt