By Elie Dufeu, CTO & Co-Founder, Metreecs. Published 30 June 2026.
Getting stock allocation right in fashion retail is the difference between a season that pays and one that ends in markdowns. The wrong unit in the wrong store at the wrong time costs margin twice: once in the lost sale, and once in the discount needed to clear it.
For brands managing 500 products or more across multiple stores and channels, the allocation decision is also a forecasting decision. You can only place stock where demand is highest if you know, before the season peaks, which stores will see the strongest velocity.
Key Takeaways
- AI-driven allocation and replenishment has cut lost sales from stockouts by up to 90% in documented fashion retail deployments, while holding in-stock levels near 95% without carrying excess stock
- The biggest allocation mistake is distributing limited inventory evenly across all stores; concentrating stock where sell-through velocity is highest produces better results
- Many buying teams hold back a portion of initial stock (commonly in the 15-25% range) as a reserve, then deploy it to early winners after the first two weeks
- A complete size run in a top-tier store outperforms partial size coverage across several stores. On a flat size curve, a 1,000-unit buy can net roughly 11% less gross profit than the same buy on the correct curve, purely from broken sizes and resulting markdowns
- Retailers adopting AI-driven inventory practices have reported around a 25% reduction in holding costs alongside improved stock turnover
Book a demo to see how Metreecs handles allocation at the store-product level.
Why most allocation decisions leave margin on the table
Most buying and allocation teams operate on a version of the same logic: look at last season's sales by store, apply a percentage uplift or reduction, and distribute accordingly. It feels rigorous. It produces predictable results. And it consistently underperforms.
Historical sales data tells you what sold where it was stocked. It does not tell you what could have sold if the stock had been placed differently. A store that appears to underperform may have been starved of inventory at the moment demand spiked. A store that looks strong may have benefited from a nearby competitor closing, a condition that no longer applies.
A pattern buying teams see repeatedly: a new line gets distributed across stores using the prior year's split, a flagship location sells out of its top sizes within the first weeks of the season, while secondary stores still hold units in those same sizes. The stock eventually gets transferred, but the peak trading window at the flagship has already passed by the time anyone notices.
The root cause is rarely poor judgment. It is a planning model that cannot respond to real-time sell-through signal.
Allocation and in-season rebalancing: two decisions, one strategy
Most teams treat these as separate problems. They are not.
Initial allocation distributes stock at the start of a season. In-season rebalancing (store-to-store transfers, replenishment from the distribution center, reserve deployment) determines whether you hold the second half. Treating them as disconnected decisions leads to poor reserve planning and reactive transfers that arrive too late.
The right approach links them from the start:
- Tier your stores by current sales potential, not historical volume. Use sell-through rate velocity (the rate at which a product moves relative to how long it has been in store), location-level trend, and strategic weighting (flagship vs. outlet, tourist-heavy vs. residential).
- Protect complete size runs at Tier A stores. A size-color variant that sells at full price contributes more margin than a size run spread across several locations where no single store can satisfy a customer.
- Hold back a meaningful reserve, commonly in the 15-25% range. Deploy it to early winners after the first two weeks, when sell-through velocity is visible.
- Pre-set escalation rules for transfers. If a Tier A store sells through 60% of a key product in three weeks and a Tier C store has sold fewer than 10%, the transfer triggers automatically (no escalation call required).
This is a demand planning decision as much as an allocation one. Forecast accuracy at the store-product level is what makes tiering reliable through the season.
The size run problem: why partial coverage destroys margins
In fashion, the size run is the unit of demand. A customer looking for a size 40 blazer in navy is not satisfied by a size 38 in a similar style. Each broken size run is a lost sale. Across a multi-product assortment, broken runs compound.
The standard response to limited stock is to spread it: give every store one or two units per size across several key sizes. This optimizes optics (every store has something) at the cost of performance (no store can fully service demand in any single size).
The alternative, concentrating full size runs in fewer stores, feels riskier until you model it. Industry data backs the concentration approach: a flat size curve, where units are spread evenly rather than matched to actual size demand, can net roughly 11% less gross profit on a 1,000-unit buy than the same buy allocated on the correct curve, once broken-size markdowns are factored in.
This is one reason AI-driven inventory optimization has been documented to cut lost sales from stockouts by up to 90% in fashion retail deployments: the system identifies which store-product combinations are most likely to sell through at full price and concentrates stock accordingly, rather than distributing evenly. Learn more about AI-driven inventory optimization.
Reading early sell-through signal to deploy the reserve
The reserve is not a hedge. It is a tool for compounding winners.
The first 10-14 days after a new collection hits stores contain more predictive signal about full-season performance than any pre-season forecast. Sell-through velocity in week one, by store and by size, shows which locations will burn through their allocation before the season peaks.
The discipline is twofold: identify the winners early, then deploy reserve stock before they run out, not after. A stockout loses not just the immediate sale but the reorder intent. A customer who finds their size at your flagship on Saturday comes back. One who does not goes somewhere else.
Operationalizing this requires two things most teams lack: a reliable sell-through feed at the store-product level updated daily, and a clear pre-defined threshold for reserve deployment. Without the threshold, the decision reverts to politics. The vocal store manager who calls loudest gets the stock, not the store with the highest conversion probability.
AI agents for retail change this by automating the threshold check. When a store's sell-through rate crosses the pre-set trigger, the system queues the transfer without waiting for a planner to notice.
What happens when allocation and demand planning are connected
Consider what a connected system changes in practice.
Under a weekly manual process, sell-through reports get reviewed on a fixed day, with transfer decisions made a few days later. By the time stock moves, the window has often passed.
Connecting the allocation model to a real-time demand planning platform shifts the workflow to daily exception monitoring. Instead of reviewing every store manually, planners act on a short list of product-store combinations that have crossed the deployment threshold. Reserve stock can move in days rather than weeks, freeing up time for the decisions that require judgment: which new delivery to prioritize, which stores need attention on the next season's open-to-buy.
The working capital benefit compounds. Better S&OP coordination and real-time inventory visibility are consistently linked to lower days inventory outstanding, since tighter sell-through and fewer end-of-season markdowns mean stock converts to revenue faster rather than sitting on shelves.
The planning vocabulary that matters
Before implementing tiered allocation and reserve deployment, align on the metrics that drive each decision:
Sell-through rate: units sold divided by units received, expressed as a percentage over a defined period. The primary signal for store tiering and reserve deployment decisions.
Service level: the probability of meeting demand without a stockout. Setting a target service level (typically 90-95% in fashion) determines how much safety stock the reserve needs to cover demand variability during replenishment lead time.
Open-to-buy (OTB): the buying budget remaining for the current period. Demand planning affects OTB directly: if your forecast shows strong sell-through at Tier A stores, you may want to increase the initial buy, but only if OTB allows.
Days inventory outstanding (DIO): total inventory value divided by average daily cost of goods sold. The KPI that tracks whether better allocation is actually reducing working capital tied up in stock. Retail and e-commerce DIO typically ranges from 30 to 90 days depending on category and seasonality.
Getting these four metrics aligned across buying, planning, and logistics is the organizational precondition for the operational changes above. Without shared definitions, a 60% sell-through trigger means something different to each team.
FAQ
How do I tier stores when I do not have reliable sell-through data by store?
Start with what you have: total revenue by store last season, your read on each location's customer profile and foot traffic, and any available context on local competitive dynamics. Imperfect tiering based on judgment and partial data still outperforms flat allocation. As you build a more reliable sell-through feed, refine the tiers.
What sell-through threshold should I use to trigger reserve deployment?
A common starting point is 60% sell-through within the first two weeks for a priority product. The right number depends on your season length and markdown schedule. The key is to set it before the season starts and commit to it, rather than deciding case by case.
How do I handle pushback from store managers who feel they are being shorted?
Anchor the conversation in outcomes. After a season or two of tiered allocation, present the sell-through data: stores that received concentrated, complete-size-run allocation tend to outperform stores that received partial coverage. Managers in lower-tier stores often come around once they see that partial coverage produces earlier markdowns, which affects the category's perceived value in their store.
Is tiered allocation only viable for large retailers?
No. The approach is valid for any brand with more than one sales channel. AI tools reduce the manual effort required at scale, but the underlying logic works at three stores as well as 300.
How does open-to-buy budgeting interact with the allocation decision?
OTB sets the buying budget. Allocation determines how purchased inventory is distributed. If your OTB was set conservatively, you may not have reserve inventory to fund second-wave deployment at winning stores. Demand planning at the pre-season stage, before OTB is set, gives you the flexibility to build a meaningful reserve into the plan.
What is the difference between a stock transfer and in-season replenishment?
A transfer moves existing inventory from one store to another. Replenishment moves stock from a distribution center to a store. The decision logic is similar, and both should be triggered by sell-through velocity crossing a threshold, but replenishment from a DC is generally faster and cheaper than inter-store transfers. A well-structured initial allocation minimizes the number of transfers required.
Conclusion
Full-price sell-through is not won by buying the right product. It is won by putting it in the right place at the right time. Tiered allocation, size run protection, and reserve deployment are the mechanisms. Reliable demand forecasting is what makes them work at scale.
If your allocation model still distributes inventory based on last season's data, you are leaving margin on the table. Book a demo to see how Metreecs handles allocation decisions at the store-product level.
Sources
- PRIME AI: How Fashion Retailers Can Reduce Lost Sales from Stockouts
- Impact Analytics: Enterprise AI Inventory, 4 Real Retail Deployments
- Style Arcade: Size Matters, How the Retail Industry Loses Money on Inaccurate Sizing
- RetailDogma: Broken Sizes in Fashion Retail
- Descartes Finale: Days Inventory Outstanding, Complete Guide
Our articles
Own your operations with AI Intelligence
.svg)