Size-Curve Fragmentation
Assumption: five colours across six sizes turn one dress into 30 stock positions before it reaches a second store. The dangerous number is not 30. It is the first zero in a core size, because the remaining pieces may still count as inventory while no longer forming a credible choice for the customer.
How one style fragments
The arithmetic is multiplication, not addition:
stock positions = styles × colours × sizes × locations
One style, five colours, six sizes, and 100 stores produce 3,000 location-SKU cells. A 900-unit buy sounds deep at style level; spread evenly, it averages only 0.3 units per cell.
The style total hides the break because units are not interchangeable. A customer seeking M cannot consume the three XS units beside it. concept size curve explains the planned split; fragmentation explains what happens when that split is multiplied across colours and stores.
The broken-curve trap
A broken curve is not automatically dead stock. Li, Lu, Lu, and Huang studied about 1.5 million sales and inventory records across 217 stores, 503 footwear products, and 4,024 SKUs over two years. Their 2023 estimates found that 51.7% of unmet demand moved to an adjacent size, 20.2% moved to another style, and 28.1% became lost sales.
Those figures belong to sports footwear, not dresses. They still expose the key error: substitution exists, but it is partial. When several sizes disappear together, the same study found that spillover weakened. A rack can therefore deteriorate nonlinearly. The first stockout hurts one size; widespread stockouts damage the style.
What the spreadsheet misses
| View | Visible number | Hidden failure |
|---|---|---|
| Style | 900 units | No fit information |
| Colour | 180 units | Uneven colour demand |
| Size | 30-unit average | Core-size concentration |
| Location-SKU | 0.3-unit average | Empty cells before launch |
This is why concept store cluster planning matters. Pooling every store into one national curve ignores local bodies, climates, and colour preferences. Giving every store its own curve creates thin cells and noisy forecasts. The operating problem sits between those extremes.
What's contested
There is no universal threshold at which a curve becomes “broken.” The answer depends on fit tolerance, adjacent-size substitution, selling horizon, channel, and whether stock can move between stores. Akchen and Caro’s 2025 footwear research found nearly 25% adjacent-size spillover in its data, while also showing that substitution matters more in low-demand settings.
Size expansion creates a second contest. More sizes increase access but divide a fixed buy across more cells. The practical question is whether pooled inventory, faster replenishment, or fewer colour choices can fund that range without starving each size.
Why this crosses realms
Once M sells out, recorded sales stop measuring M demand. The absence becomes censored data: zero sales may mean zero interest or zero opportunity. That is an concept information theory problem before it is a forecasting problem.
Replenishment adds concept queueing theory. Each size has arrivals, service demand, lead time, and limited substitution. concept quick response reduces the cost of an imperfect first curve, while concept inditex playbook shows why short feedback loops can matter more than a precise preseason guess.
An open question
Can a demand model recover the sale that never happened after M sold out without teaching tomorrow’s size curve to repeat yesterday’s allocation error?
Key Sources
- Li, Lu, Lu, and Huang, “Estimating the Stockout-Based Demand Spillover Effect in a Fashion Retail Setting” (2023): the 217-store estimate of cross-size spillover and lost demand.
- Akchen and Caro, “On Size Substitution and Its Role in Assortment and Inventory Planning” (2025): evidence on when adjacent-size substitution changes stocking decisions.
- Fisher, Rajaram, and Raman, “Optimizing Inventory Replenishment of Retail Fashion Products” (2001): the two-stage treatment of short-life fashion inventory.
- Sung, Jang, Kim, and Lee, “Business Analytics for Streamlined Assort Packing and Distribution of Fashion Goods” (2017): a field case connecting size ratios, pack configurations, and store allocation.
Further Reading
- concept size curve: the demand distribution underneath the fragmentation problem.
- concept quick response: why shorter lead time reduces dependence on the opening allocation.
- concept store cluster planning: how stores can share a curve without pretending their demand is identical.
- concept censored demand: how stockouts hide the demand a forecast needs to recover.
See Also
- concept assortment breadth
- concept inventory pooling
- concept information theory
- concept queueing theory
Abhishek's take
I care less about the 900-unit style total than the first missing M in a selling store. Size-curve fragmentation turns apparently available inventory into a measurement problem: the units remain visible, but the customer’s choice has already disappeared.
Tags: #fashion-buying #inventory #sizing #assortment-planning #demand-forecasting