How should brands allocate scarce stock across wholesale accounts with AI?
When production falls short of orders, someone must decide which retailers get what. How AI optimises allocation against clear rules, and why perceived fairness matters as much as margin.
KEY TAKEAWAYS Summary by the editors
- Scarce stock allocation decides how a brand splits limited inventory across wholesale orders when supply is below demand, by style, colour, size and delivery.
- AI optimisation can balance several goals at once, such as margin, sell-through potential, strategic account priority and fill-rate fairness, but only after the brand has made those goals and their weights explicit.
- Experimental research published in Decision Sciences in 2022 found that under shortage, retailers most often considered equal fill rates fair, and judged demand-based allocations fairer whether a rule or a human made them.
- The same research found that participants acting as retailers claimed more than their exact needs in over a third of cases, which is why allocation should use order and sell-through history, not just stated requests.
- Allocation decisions must be explained to accounts in advance through a published policy, because unexplained shortfalls damage relationships more than the shortfall itself.
When a brand has less stock than its wholesale orders require, AI can help allocate it by optimising across margin, expected sell-through, account priority and fairness, within rules the brand sets. The system proposes how many units of each style, colour and size each account receives, and the sales team reviews exceptions. The quality of the result depends less on the algorithm than on a clear, published allocation policy and on reliable order and sell-through data.
When does scarce stock allocation become necessary?
Shortfalls arise when production comes in below the order book (supplier delays, quality rejections, fabric shortages), when a style outperforms expectations, or when a brand deliberately limits supply to protect full-price selling. Nike's chief financial officer said in December 2024, as reported by Retail Dive, that the company was targeting a significant reduction in the supply of its classic footwear franchises over the following seasons. Planned scarcity of this kind makes allocation a strategic decision, not just an operational one.
Buying patterns add pressure. JOOR's December 2025 wholesale trends paper reports that buyers increasingly place smaller initial orders and hold budget for in-season purchases, wanting available-to-sell stock ready to ship immediately. That means more allocation decisions happen in season, under time pressure and across reorders as well as pre-orders.
What goals should an allocation balance?
| Goal | Typical measure | Tension with |
|---|---|---|
| Margin | Net margin per unit by account | Fairness, strategic accounts |
| Sell-through | Expected full-price sell-through at the store | Accounts with weak data |
| Strategic priority | Account tier, brand image, flagship doors | Margin, smaller accounts |
| Fairness | Equal or comparable fill rates | Margin and sell-through |
| Size integrity | Complete size runs per store | Total units shipped |
| Contractual commitments | Exclusives, guaranteed quantities | All other goals |
Making these goals explicit is the hardest step. Many brands discover that different departments hold different priorities: sales wants to protect key accounts, finance wants to maximise margin and brand management wants to protect flagship doors and image. An optimisation model forces these trade-offs into numbers, which can be uncomfortable but is also useful, because it replaces informal negotiation during a shortage with a policy that management has agreed in advance and can review after each season.
How does AI allocate scarce stock?
Most systems use optimisation rather than prediction alone. Forecasting models estimate the expected full-price sell-through of each style at each account, using order history, sell-out data where available and store profiles. An optimisation model then distributes the available units to maximise a weighted objective, subject to constraints such as minimum fill rates, complete size runs, contractual quantities and pack sizes. The output is a proposed allocation per order line, with exceptions highlighted for review.
Without AI, brands typically apply a single rule, such as cutting every order by the same percentage or serving orders in date sequence. These rules are simple to explain but can send stock to accounts where it will sell slowly, while broken size runs reach stores that cannot use them. Optimisation can avoid these effects, at the cost of being harder to explain.
Why does perceived fairness matter?
Allocation is a relationship decision. Experimental research by Eirini Spiliotopoulou and Anna Conte, published in Decision Sciences in 2022, examined how retailers judge the fairness of inventory allocation. Under shortage, participants most often considered equal fill rates the fair norm. Allocations based on realised demand were judged fairer, whether a rule or a human made the decision. The study also found that participants acting as retailers claimed more than their exact needs in over a third of cases.
Two practical lessons follow. First, allocation logic should be explainable in terms of fill rates and demand that accounts can recognise. Second, stated requests are not a reliable measure of need, so models should rely on order history and sell-through rather than on what accounts say they want during a shortage.
How should a brand implement AI allocation?
- Write the allocation policy: goals, weights, account tiers, contractual commitments and minimum fill rates.
- Clean the data: order lines, confirmed quantities, available-to-sell stock, sell-through by account and size, and pack constraints.
- Start with a transparent rule set and measure its outcomes for one season as a baseline.
- Introduce optimisation for a limited set of styles, compare results with the baseline and review exceptions manually.
- Give reps an explanation per account (fill rate, reason for cuts) to share with buyers.
- Track outcomes: sell-through of allocated stock, fill rates by account tier, complaints and cancellations.
What data problems undermine allocation?
- Stock figures that are not current, causing allocation of units that are already committed elsewhere.
- Missing sell-out data for many accounts, which biases allocation towards accounts that share data.
- Inconsistent account hierarchies, so a group with several doors is treated as separate small accounts.
- Size-level forecasts that are too noisy for small accounts, requiring pooling across similar stores.
Gartner predicted in February 2025 that through 2026 organisations will abandon 60% of AI projects unsupported by AI-ready data. Allocation is especially exposed, because a wrong stock figure leads directly to a broken promise to a customer.
Where should humans stay in charge?
Sales leadership should own the policy and the weights between goals, and review allocations for strategic accounts, new accounts and any account receiving a much lower fill rate than its peers. The model handles the many routine lines; people handle the decisions that carry relationship consequences. That division keeps allocation fast without making it feel arbitrary to the retailers who depend on it.
Frequently asked questions
What is stock allocation in fashion wholesale?
It is the process of deciding how available inventory is distributed across retailer orders, by style, colour, size and delivery. It becomes critical when supply is below demand. Brands use rules, optimisation models or a combination.
How do brands decide which retailers get stock when supply is short?
Common approaches include equal percentage cuts, priority by account tier, first-come-first-served and optimisation that weighs margin, expected sell-through and fairness. The best approach is set out in a policy agreed and shared with accounts before shortages occur.
Is AI allocation fair to smaller retailers?
It can be, if fairness is built into the rules, for example through minimum fill rates. Without such constraints, optimisation may favour large accounts with better data. Research shows retailers tend to see equal fill rates and demand-based allocation as fair under shortage.
What data do I need for AI stock allocation?
Accurate available-to-sell stock, confirmed order lines, sell-through by account and size, account hierarchies and constraints such as pack sizes and contractual quantities. Stock accuracy is the most critical input, since errors lead directly to unfulfilled promises.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
