What calibration does
When you upload a size chart, Measmerize maps body measurements to your sizes based on the data you provided. That works well out of the box, but no two brands cut their clothes the same way.
Calibration gives you one input to fine-tune that mapping: you pick a real body profile and tell us the size that person should wear. The algorithm then recenters all recommendations around that anchor point.
Here's a concrete example. Say our algorithm would have suggested an L to your reference profile before calibration, but you know that person is a true M. After calibration, that person gets an M, and all recommendations across the entire chart shift relative to that new center point. Every profile, every size, recalibrates proportionally.
One profile. Everything adjusts relative to it.
What it doesn't do
The tool holds one calibration at a time. If you run it twice, the second input replaces the first. The algorithm recenters around the new profile, not both. There's no point calibrating separately for each size; only the last input counts. Storing multiple profiles would also distort the size increment evolution across the chart, which is why we deliberately keep it to a single anchor.
Because calibration shifts recommendations proportionally across the full range, the effect isn't always a flat one-size-down for everyone. A shopper who was an S but sitting close to the M boundary might stay an S after calibration rather than dropping to XS. The shift is distributed across the chart, not applied identically to every profile.
When to use it
Calibrate once, when you're setting up a new size chart. The best input is the profile of your "perfect M" (or whatever your anchor size is): the body measurements you had in mind when you designed that size.
We only recommend recalibrating if you see a sustained pattern of returns for this chart after your initial calibration. A few returns isn't enough signal. A consistent trend across sizes is. And even then, check whether the issue is in the size chart data itself before touching calibration. Calibration helps with systematic offset, but it won't fix missing columns or incorrect measurements.
How to calibrate, step by step
Step 1 - Open the calibration tool
Go to the size chart you want to calibrate and open it in edit mode
In the header, find the Calibration status card
Click Calibration pending to start, or Calibration done to redo an existing one
Step 2 - Enter your reference person's profile
Select the Gender of your reference person
Enter their Height, Weight, and Age (use the cm/ft toggle to switch units)
For women's charts, also fill in the Bra size: select the system (FR, US, UK...), then the band and cup
We recommend using the measurements of your "perfect M"; the body your size was designed to fit.
Step 3 - Select the expected size
At the bottom of the panel, you'll see the available sizes of this chart
Select the size your reference person would normally buy
Step 4 - Confirm
Click Calibrate
The header updates to Calibration done, no need to republish or reassign products
A note on best practice
Use the body profile of the person your size was designed for, not an "average" shopper. The reference profile works best when it reflects genuine design intent. Your pattern maker or buying team usually knows this, so it's worth asking them before you calibrate.
And remember: one calibration per chart. Running it multiple times doesn't stack. The last one wins.
Also, don't aim for a calibration that gives the "right" answer for every profile you can think of. There will always be shoppers who are between two sizes, or who prefer to size up or down for comfort reasons that have nothing to do with fit. Calibration is about aligning the algorithm with your design intent, not eliminating every edge case. Some variance is normal.