Why size recommendation doesn't work
The central question isn't: 'What size do similar customers buy?' But rather: How does this garment behave on this body?
of returns are size-related
typical error margin of competitors
standardized sizing systems worldwide
What conventional systems do
Statistical correlation
Systems like True Fit or Fit Analytics collect purchase and return data to derive size recommendations. The problem: the data is noisy because customers return items for various reasons — not just sizing.
Crowd-based prediction
"Customers with a similar profile bought size M." But what does "similar" mean? Two people of the same height and weight can have completely different proportions.
Generic size charts
Size charts are treated as truth, even though they vary massively — between brands, collections, and even within a product line.
What's actually needed
Real body measurements, not self-reported data
0.59cm accuracy through mathematical modeling. Not height/weight, but shoulders, chest, waist, hips, arm length — per area.
Real garment analysis
Not just a size chart, but actual garment dimensions, cut, material, and elasticity. Each product is individually analyzed.
Physics-based calculation
Deterministic simulation instead of statistical guessing. How does this specific garment behave on this specific body?
The core problem of size recommendation
The core problem of size recommendation in e-commerce isn't a data problem — it's a modeling problem. More purchase data won't solve it. What's missing is a physics-based understanding of how clothing fits on individual bodies. maketribe solves exactly this problem: with mathematical modeling instead of statistics, 0.59cm measurement accuracy, and per-body-area analysis.