Retail size curves explained: how to set, review and correct them without overbuying
Merchmix ·
Many retail teams can spot a size-curve problem after the fact: core sizes sell out early, slower sizes linger, and stores end up with broken size runs. The harder part is building a repeatable process that catches those issues before buy commitments are locked. In practice, that means using more than one input, checking whether sales data is distorted by out-of-stocks, deciding how much planning detail is worth the effort, and agreeing clear exception rules before orders are finalised. This guide focuses on that workflow, using hypothetical examples rather than case studies.
Start with the signals that actually shape size demand
Before setting a curve, planners should pull together the inputs most likely to explain why size demand differs across the business. Public guidance from SPS Commerce highlights historical sales, point-of-sale data, regional buying patterns and store characteristics such as location and format as useful inputs.
That matters because one standard curve rarely fits every door or channel. Store format, local customer mix and trading context can all shift the balance between sizes. A flagship city store, a college-area location and a regional mall store may all need different assumptions, even within the same brand.
A practical way to begin is to define the planning unit before debating the exact ratio. For example, decide whether a product family should be planned at national level, by region, by store cluster, or with a small number of article-level exceptions. That keeps the team focused on where demand is genuinely different rather than endlessly adjusting percentages.
Do not rely on raw sales alone if key sizes went out of stock
One of the most common planning mistakes is to treat past sales as if they represent unconstrained demand. Vendor-authored retail planning guidance warns that this can be misleading when a size sold out early. In that situation, the missing size becomes underrepresented in the data, not because customers stopped wanting it, but because it was unavailable.
If planners feed that constrained mix straight into the next buy, the error can repeat across future seasons and allocations. In simple terms, the business can end up teaching itself that a size is less important precisely because it failed to keep it in stock.
A better review question is not only what sold, but how each size performed during the periods it was actually in stock. Public material from Style Arcade supports using in-stock-aware size performance measures rather than relying only on total units sold. That does not remove the need for judgement, but it is a useful guardrail before locking a curve.
Choose the planning detail that your demand signal can support
Not every category needs the same level of size planning detail. Public sources support two useful principles here: different stores can have meaningfully different size-demand patterns, and different products may also behave differently. That means planners should avoid both extremes: one national curve for everything, and hyper-granular modelling for every low-volume SKU.
A practical rule is to start broader where demand is stable and volume is lower, then add detail where variation, volume or financial risk are high. For example, a team might use one curve for a stable basics line nationally, regional curves for a major denim program, and article-level exceptions only for a small number of high-risk launches.
This is workflow guidance rather than a formal industry standard, but it helps keep analysis proportional. If the signal is weak, extra granularity can create noise. If the signal is strong and the exposure is meaningful, broader averages can hide problems that are expensive to fix later.
Match pack strategy to forecast confidence
Once the team has a working curve, it still needs to translate that curve into a shipment approach. SPS Commerce notes that established products with predictable demand often suit prepacks, while new products or new channels may be better served by open stock until demand becomes clearer.
The logic is straightforward. Prepacks are efficient, but harder to correct once shipped. Open stock is more flexible, but typically requires more distribution-centre labor and fulfillment complexity. The right choice therefore depends less on preference and more on confidence in the forecast.
Hypothetical example: a retailer launching a proven school-shoe line into familiar store clusters might choose prepacks for initial sets because demand patterns are well understood. By contrast, a new fashion sneaker entering an unfamiliar marketplace could start with a more flexible open-stock approach, giving planners room to respond if one size band over- or under-performs.
Build a practical pre-buy review instead of treating size curves as a one-off task
In the public sources cited here, there is no defined standard review cadence, so the most useful approach is a practical one. A sensible workflow is to run a structured pre-buy review with merchandising, planning and allocation input before final commitments are locked.
A simple review can cover five questions. First, did any key sizes sell out early enough to distort the sales history? Second, are there store or regional clusters where one curve appears to be masking real differences? Third, are there new products, channels or range changes that make last season's curve less reliable? Fourth, does the chosen shipment method still match forecast confidence? Fifth, is the product important enough to justify a more detailed exception review?
The aim is not to achieve a perfect forecast. It is to identify avoidable errors early enough to adjust the buy, split clusters, revise pack ratios or reserve flexibility for later allocation decisions. That is often more realistic than trying to optimise every size decision to the last unit.
Know the exceptions that deserve manual intervention
Some situations should trigger a manual review even in a disciplined process. One is early sellout bias, because constrained sales can hide true demand. Another is low-volume noise, where a small sample can make a curve look more precise than it really is.
New product launches and new channels also deserve caution because past sales may not be a dependable guide. Public evidence also supports being careful about applying the same curve across unlike articles, since different products can show different size distributions.
As a hypothetical internal policy, some teams may separately review materially changed size-range assortments when demand confidence is low. The key point is operational: exceptions should be identified on purpose, not discovered by accident after inventory is already in stores.
Corrective actions to take before buy commitment
If the review suggests the current curve is weak, planners still have several levers available before overbuying becomes locked in. They can adjust the curve itself, split one cluster into two, change pack ratios, reduce the initial commitment, or hold back units for more flexible follow-up allocation.
These actions are especially relevant because poor size curves can lead to stockouts in popular sizes, excess units in slower sizes and broken size runs that create markdown risk. Correcting the mix before commitment is often simpler than trying to trade out of the problem later.
Hypothetical example: an apparel team testing a new outerwear line sees that pilot doors sold out of medium and large quickly while smaller sizes remained. Rather than simply increasing the total buy, the team could revise the size ratio for similar store clusters, keep some depth uncommitted, and use flexible follow-up allocation once a clearer demand pattern emerges.
Connect size-curve decisions to the wider planning workflow
In practice, many teams get better coordination when size-curve review is connected to assortment planning, allocation and replenishment rather than treated as a separate spreadsheet exercise. Size-mix problems may also appear in assortment, allocation or replenishment decisions if the starting assumptions are weak.
For a stronger process, teams should make product history, inventory availability and cluster assumptions visible across functions. Even when the final decision is made by a buyer or planner, the process is stronger when the inputs are visible and the exception logic is consistent.
For retailers reviewing tools and workflows, a useful goal is a repeatable way to compare assumptions, flag exceptions and carry the chosen curve into buy and allocation decisions.
The most useful size-curve process is usually not the most complex one. It is the one that helps teams spot distorted demand, choose the right level of detail, and act on exceptions before inventory is committed. Start with credible inputs, challenge sales data where out-of-stocks have hidden demand, use more granularity only where it is justified, and match pack decisions to forecast confidence. Done consistently, that approach gives planners a clearer way to set, review and correct size curves before inventory is committed, which can help limit avoidable buying errors.