Size Curve
A 1,000-unit style can still fail if the size curve is wrong. The lost sale often hides in XS and XXL, not in total stock. A size curve is the planned split of units across sizes: 8% XS, 18% S, 28% M, 26% L, 14% XL, 6% XXL, or whatever the customer body and store history actually support.
How it works
Retail buyers do not buy "a dress." They buy a dress multiplied by color, size, store, channel, and week. The size curve is the quiet multiplier that decides whether the stock lands where bodies exist.
The mistake is easy: treat the middle as safe. A bell-shaped curve feels sensible because M and L usually carry volume. But fashion demand is not only body distribution. It is also fit block, fabric stretch, silhouette, age band, city, channel, and price. A bodycon knit and an oversized cotton shirt can need different curves even when the printed size labels match.
A useful curve starts with history, then gets punished by context. If a style sold out in XXL by week 2, the recorded sales understate XXL demand. If XS remained because the fit ran one size small, that is not proof that XS demand died. Size curves are forecasts with censorship built in: stockouts hide the demand they created.
Where it shows up
A 10,000-unit buy with the wrong curve can look healthy at aggregate level while failing in the aisle. The total line says "stock available." The customer sees "my size is gone."
| Signal | What it may mean | Buyer risk |
|---|---|---|
| M and L sell fast everywhere | True core demand | Underbuying the center |
| XS sells slowly in one fit block | Fit may run large | Cutting XS too hard |
| XXL sells out early | Censored demand | Mistaking stockout for weak volume |
| Online has more edge-size demand | Wider search behavior | Store curve cannot be copied blindly |
A useful benchmark is concept queueing theory: a size is a queue with almost no substitution. A customer who needs XL does not buy S because S is available. In that sense, broken sizing behaves less like normal inventory and more like a capacity constraint.
What's contested
The hard question is not whether size curves matter. That is settled in practice. The contested part is how much history to trust when the history was produced by bad availability.
One school trusts sales data after stockout correction. Another trusts anthropometric and fit data first, then lets sales adjust it. Both can be wrong. Human bodies change slowly; fashion silhouettes change in one season.
Cross-realm bridge
The size curve is the fashion version of concept information theory. The missing bit is not "how many units sold?" It is "which demand signal was never observed because the size was absent?" Shannon's 1948 frame is useful here: surprise carries information, and an early edge-size stockout is a high-information event.
It also links to tech jacquard loom. Jacquard turned fabric into addressable instructions in 1804. Modern sizing turns bodies into addressable demand. The label on the garment is crude, but the operating problem is already computational.
An open question
If online returns expose fit pain at size level, should a retailer let return reasons rewrite the next size curve faster than sales do? That question belongs near concept inditex playbook and concept quick response.
Key Sources
- Claude E. Shannon, "A Mathematical Theory of Communication" (1948) — the clean frame for missing information and surprise.
- Marshall Fisher, Janice Hammond, Walter Obermeyer, and Ananth Raman, "Making Supply Meet Demand in an Uncertain World" (Harvard Business Review, 1994) — the Sport Obermeyer case behind early-demand correction.
- ISO 8559-1:2017, Size designation of clothes, Part 1: Anthropometric definitions for body measurement — the standards layer beneath apparel sizing.
- ASTM D5585, Standard Tables of Body Measurements for Adult Female Misses Figure Type — one reference point for body-measurement tables.
- to verify: public apparel-retail studies separating size-level lost sales from aggregate sell-through.
Further Reading
- concept queueing theory — size availability behaves like capacity, not decoration.
- concept quick response — the shorter the production loop, the less a bad initial curve hurts.
- concept inditex playbook — the famous speed story is also a demand-sensing story.
- tech jacquard loom — fabric was programmable before software had a name.
Abhishek's take
The size curve is where fashion stops being taste and becomes arithmetic. I care about it because the buyer's visible mistake is overstock, but the deeper mistake is usually unobserved demand. The clean move is to treat stockouts as missing data, not as proof that the customer disappeared.
Where I've used this
I use this on the buying floor when a style's total sales look healthy but the edge sizes tell a different story. The useful question is not "did it sell?" but "which size-level demand did the system never get a chance to see?"
Tags: #fashion-buying #inventory #sizing #demand-forecasting #retail-operations