What is a size matrix in apparel? Size runs, ratios and prepacks explained.

The matrix problem

In most industries, a product is a SKU. A specific coffee grinder has one barcode, one inventory count, one price. Apparel doesn’t work that way. A style - the “Coastal Twill Overshirt” - is not a sellable thing. What sells is the Coastal Twill Overshirt in Washed Black in size Medium. Every combination of colour and size is its own SKU with its own barcode, its own inventory position and its own demand pattern.

That grid of combinations is the size matrix. One style × 5 colourways × 7 sizes = 35 SKUs, before you add market variations or channel-specific assortments. A 200-style seasonal range easily runs past 5,000 SKUs. Everything operational in apparel - buying, warehousing, wholesale, replenishment, markdowns - happens at matrix level, not style level.

The matrix is also where apparel software earns its keep or fails. Systems built for generic products treat each SKU as an island; apparel systems understand that the 35 SKUs of one style share a BOM, a tech pack, a cost structure and a lifecycle, while carrying independent inventory and demand. Holding both views at once is the defining data-model requirement of the industry.

Size runs

A size run is the set of sizes a style is offered in. Runs vary by category and market:

• Alpha sizing - XS, S, M, L, XL, XXL. Standard for tees, knits, swim and most casual categories.
• Numeric sizing - 28-38 waist for denim and chinos, 0-16 for womenswear, 6-13 for footwear. Often with half-sizes or inseam variants that multiply the matrix again (32×30, 32×32, 32×34).
• Extended runs - plus, petite, tall, big & tall. Structurally the same style but usually with separate grade rules and sometimes separate BOMs (more fabric consumption).
• Kids’ sizing - age-banded (2-3Y, 4-5Y) or height-banded (98cm, 104cm) depending on market convention.

The run decision is commercial, not just technical: every size added increases inventory investment and production minimums while capturing incremental demand. Core styles earn wide runs; fashion styles usually get narrower ones. The discipline is deciding the run per style deliberately at range planning stage rather than defaulting every style to the full run.

Size ratios and curves

A size ratio (or size curve) is the quantity distribution across the run - how many of each size to produce, buy or pack. Demand is never flat across sizes: for a typical US menswear tee, M and L together might account for half of sales, with XS and XXL in single-digit percentages.

A ratio is usually written as proportions across the run: 1-2-3-3-2-1 across XS-XXL means for every 12 units, one XS, two S, three each of M and L, two XL, one XXL. Ratios differ by:

• Category - fitted styles skew differently than oversized ones
• Market - the US curve differs from the Japanese curve for the same style
• Channel - a surf-shop account’s curve differs from a department store’s
• History - the best ratio source is your own sell-through by size for comparable styles, not industry averages

Getting ratios wrong is expensive in both directions: over-buying the tails creates markdown inventory; under-buying the core sizes breaks the run early (covered below). Brands that track sell-through by size and feed it back into next season’s ratios systematically outperform brands that reuse a house ratio for everything.

Prepacks

A prepack (case pack, ratio pack) is a fixed assortment of sizes packed together as one orderable unit. A 6-unit prepack of 1S-2M-2L-1XL means the retailer orders “3 prepacks” instead of specifying 18 units across four sizes. Prepacks simplify ordering, receiving and shelf replenishment for retailers - many mass and department-store accounts order primarily or exclusively in prepacks.

Operationally, prepacks touch every downstream system:

• The size matrix defines which prepack configurations exist for a style
• Wholesale orders are taken in prepacks but often need visibility of the unit-level breakdown
• The warehouse needs pick instructions at prepack level - and needs to know whether packs are pre-assembled by the factory or assembled at the DC
• EDI documents (the 850 order and 856 ASN) carry prepack structures with inner-pack details; getting these wrong is a chargeback trigger
• Inventory has to be visible both as packs and as units, because e-commerce sells units out of the same stock pool that wholesale sells as packs

The pack-vs-unit duality is one of the classic places where generic inventory systems break in apparel: they can model one or the other, but holding both views over the same physical stock requires the matrix-native data model.

Per-market matrices

Multi-market brands quickly discover that one global size matrix doesn’t work. Body dimension distributions differ by population; sizing conventions differ by country (a US women’s 8 is roughly a UK 12 and an EU 40); and demand curves differ even where sizing aligns.

The practical model: the style holds one master definition, with per-market size matrices layered on it. The US matrix runs XS-XXL on the US grade rule; the Japan matrix runs S-XL on the JP grade rule with its own ratio; the EU matrix uses EU size labels mapped to the same or adjusted grades. Same style, same BOM, market-appropriate matrices.

This layering has to survive the trip through every downstream document - line sheets show the right sizes per market, wholesale orders validate against the right matrix, labels carry the right size nomenclature, and EDI catalogues (the 832 document, where used) publish market-correct data. A mislabelled size run reaching a retailer is a compliance issue, not a cosmetic one.

The matrix and inventory

Inventory in apparel is counted, allocated and replenished at SKU level - which means size-and-colour level. This creates analytical requirements that generic systems don’t anticipate:

• Sell-through by size - a style at 60% sell-through may hide a Medium at 95% and an XS at 20%. Style-level reporting alone actively misleads.
• Ratio-aware replenishment - reordering should rebuild the curve, not just top up total units. Reordering 100 units flat across sizes when demand runs 1-2-3-3-2-1 recreates the imbalance.
• Matrix-level allocation - allocating stock between wholesale commitments, e-commerce and retail stores happens per SKU, respecting channel-specific curves.
• Pack-unit reconciliation - the same physical stock may be committed as prepacks to wholesale and as units to D2C, and the system has to prevent double-selling across the two views.

Our guide on apparel inventory management covers the operational side in more depth; the point here is that every one of those capabilities depends on the size matrix being a first-class object in the system rather than an afterthought.

Broken size runs

A size run is broken when core sizes sell out while others remain in stock. The style is technically available but effectively dead at retail: most customers who pick it up can’t find their size, conversion collapses, and the remaining inventory skews toward the hard-to-sell tails.

Broken runs matter commercially because retailers respond to them: reorders stop, markdown support gets requested, and next season’s buy shrinks. For D2C, a broken run quietly kills the product page’s conversion while still consuming ad spend if the style is being promoted.

The operational disciplines that manage broken runs:

• Detection - a report or alert surfacing styles where core sizes (defined per style) are out of stock while total inventory remains above a threshold
• Consolidation - pulling scattered remaining sizes back from stores or channels into one sellable location
• Rebalancing - in-season reorders targeted at the missing sizes when lead times allow
• Exit - marking down or outletting the residual tail sizes deliberately instead of letting them age

Brands that treat broken-run management as a weekly routine recover meaningful margin compared to brands that discover the problem at season-end stocktake.

Why generic systems struggle with the matrix

Generic inventory and ERP tools model products as flat SKUs, sometimes with a “variant” feature bolted on. That works until the apparel-specific requirements arrive: per-market matrices on one style, prepack and unit views over the same stock, ratio-aware replenishment, size-level sell-through analytics, grade-rule-aware sizing, and EDI documents that carry pack structures.

Each of those can be worked around individually - spreadsheet exports, naming conventions, custom fields. Collectively, the workarounds become the operating model, and the team spends its time reconciling views the system should hold natively. This is typically the point where growing brands move from generic inventory tools to apparel-native platforms - the matrix is usually the specific breaking point.

When evaluating any system for apparel, the matrix test is simple: ask to see one style with two market-specific size runs, a prepack definition, and size-level sell-through reporting - live, on real data. Systems built for the matrix show this in minutes; systems with variants bolted on start explaining workarounds.

Frequently asked questions