Executive summary

A Merchandise Assortment Plan, or MAP, defines what the retailer intends to sell, where it will be ranged, when it will arrive, how much depth it will receive and how the investment will achieve sales, gross margin, service and turnover objectives. It connects the financial merchandise plan to the style-color-size decisions that purchasing and allocation must execute.

1. MAP operating model

Figure 1. MAP requires top-down financial control and bottom-up choice-level feasibility.

2. Merchandise hierarchy and planning grain

The merchandise hierarchy provides financial control; the choice hierarchy provides execution. Planning must support aggregation and drill-down without confusing a style, a style-color choice and an inventory SKU.

Table 2. Recommended hierarchy and the decision governed at each level.

3. Financial merchandise plan

Figure 2. A time-phased plan identifies receipt needs, peaks, winddown and terminal-stock exposure.

Table 4. Core MAP financial measures; valuation bases must remain consistent.

4. Assortment architecture: breadth and depth

Figure 3. Core choices normally receive deeper stock; fashion, test and long-tail choices require controlled depth.

Breadth is the number of customer-visible choices. Depth is the inventory committed behind each choice. In apparel, a choice commonly means style-color; the SKU count expands further through sizes and locations.

Table 5. Illustrative breadth-depth policy by assortment role.

5. Store clustering, range matrix and capacity

Figure 4. A range matrix controls where choices are eligible before depth and allocation are calculated.

Store clustering reduces noise and gives merchants an understandable way to localize the assortment. A store may inherit its cluster range and then receive governed overrides for climate, local demand, strategic role, capacity or fulfilment capability.

Table 6. Range eligibility precedes purchasing and allocation.

6. Lifecycle and seasonal phasing

Figure 5. Lifecycle determines when receipts are useful and when additional stock becomes exit exposure.

Table 7. Lifecycle-specific MAP controls.

7. Choice productivity, ABC and option thresholds

Figure 6. Productivity typically concentrates in a smaller set of choices; ABC helps prioritize service and depth.

Choice productivity should influence depth, participation and replenishment, but it should not be the only design input. Strategic brands, opening price points, visual statements, new concepts and destination items may require controlled overrides.

Table 8. Cybex 85/10/5 ABC convention applied to assortment choices.

8. Color and size planning within MAP

MAP approves the style-color choices and the market participation. The detailed size curve and color-size stock policy then translate each approved choice into SKU demand, purchase quantities and store allocation. This prevents financial choice counts from hiding unproductive colors or broken core sizes.

Table 9. MAP owns the commercial choice; companion modules provide SKU-level operating logic.

9. Receipt planning, purchasing and open-to-buy

Figure 7. Open-to-buy is the uncommitted portion of the approved receipt plan after existing stock and orders.

The MAP defines the receipt need; merchandise purchasing converts that need into supplier commitments. Open-to-buy, or OTB, protects the approved financial envelope while allowing merchants to redirect uncommitted investment as forecasts change.

Table 11. Core receipt and OTB relationships; use consistent valuation and timing.

10. Scenario planning and constraint management

Figure 8. Scenarios make opportunity and commitment risk visible before purchase decisions are fixed.

Table 12. Scenarios should identify the decision, not merely display alternate numbers.

11. Integration with forecasting, purchasing and DC allocation

Figure 9. MAP is the coordinating commercial layer across the companion planning and stock processes.

Table 13. Required information exchange across the merchandise-management series.

12. Worked apparel assortment example

Assume a 26-week women's knitwear plan for three store clusters plus digital. The forecast supports 48 style-color choices. The financial envelope allows an initial commitment plus a chase reserve, while store capacity limits the amount of fashion and extended color placed locally.

Table 14. Illustrative MAP decision chain; all values are examples.

13. Governance, KPIs and review cadence

MAP should be versioned, merchant-owned and transparent. Approved targets, model recommendations, merchant overrides and executed commitments remain separate so the retailer can measure where value was created or lost.

Table 15. Balanced MAP scorecard.

14. Implementation roadmap and concise rules

Executive summary

Apparel inventory should not be managed only at the style or color level. Customer availability is determined at the style-color-size-location level, where a style can appear healthy in total while its highest-demand sizes are already out of stock.

The recommended operating model separates four decisions:

• Assortment width: which styles, colors and sizes the store should carry.

• Depth: how many units are required for each selected color-size combination.

• Lifecycle: whether the style is launch, active, winddown or exit.

• Network fulfilment: which excluded or out-of-stock sizes remain available through another store, a hub or ecommerce.

1. The normal size curve

A normal size curve is the expected percentage of demand by size for an active style, category or fit family. The curve commonly forms a single peak around the core sizes, but it is not a universal population curve. It varies by customer, fit, brand, garment type, gender category, region, season and channel.

Figure 1. Illustrative chain curve: XXS 4%, XS 10%, S 19%, M 28%, L 22%, XL 12%, XXL 5%.

A curve is most useful when it is stable enough to guide decisions but recent enough to reflect changing demand. Use a hierarchical fallback: store-style history where credible, then store-cluster/category, then chain-category or brand-fit history.

2. Full and shortened size runs

A full run carries every approved size. A shortened run deliberately limits size breadth because the product is winding down, the store has limited presentation space, or local demand does not justify every tail size. The rule should be based on demand coverage and network availability, not a fixed assumption that every small store needs the same sizes.

Figure 2. Example run widths. Actual carried sizes should shift according to the store size index.

Table 1. Illustrative lifecycle and space policies; category-specific governance is required.

3. Weeks of supply by size

Weeks of supply (WOS) must be calculated for every size, not only for the total style. Total style WOS can look balanced while core sizes are near stockout and tail sizes are overstocked.

Figure 3. Overall style WOS is 3.9 weeks, but M is at 3.0 while XXL is at 7.0.

Table 2. Sample WOS calculation across the entire size curve.

4. Color, size and ABC classification

ABC classification converts limited capacity into an explicit priority order. A style, color and size can each have a class. The final color-size priority should reflect all three rather than using style rank alone.

Table 3. ABC may use the retailer's preferred 85/10/5 convention or a category-specific alternative.

5. Store index and size index

Store index and size index solve different problems and should not be combined into one unexplained factor.

6. High-level allocation and distribution algorithm

Figure 6. Recommended decision loop for initial allocation, replenishment and redistribution.

7. Worked allocation example

Assume an A-class active style, 22 available presentation units, four eligible colors and the chain curve shown earlier. The store index determines overall demand; store size indices reshape the curve; color ABC and shelf capacity then establish the priority order.

Table 6. Example decision sequence for the 22-unit color-size matrix.

8. Governance and implementation

The most effective size and color system combines statistical demand with explicit retail policy. Merchants should control thresholds and exceptions, while the system applies them consistently and records results.

Appendix: concise policy rules

Table 10. High-level policy summary.

APPAREL MERCHANDISE MANAGEMENT PAPER

Executive summary

An apparel stock forecast is more than a sales projection. It is a forward-week view of demand, receipts, inventory position, safety stock, availability and exit risk at the level required to make merchandise decisions. The forecast should be built once on governed data and then consumed consistently by assortment planning, purchasing and DC allocation.

1. Forecasting operating model

Figure 1. A governed data warehouse creates a single forecast foundation for the assortment plan, purchasing and DC allocation.

2. Data warehouse foundation

The data warehouse should preserve atomic retail events and publish conformed weekly facts. The forecast engine can then rebuild any history, explain adjustments and reconcile plans without relying on spreadsheets as the system of record.

Table 2. Core warehouse domains for an apparel stock forecast.

3. Normalized and de-seasonalized sales history

Figure 2. Illustrative history: normalization removes known distortions; de-seasonalization reveals the underlying baseline.

4. Seasonality and apparel lifecycle

Seasonality describes the repeatable timing pattern around the baseline. It may be estimated at department, class, attribute, climate cluster or style-family level depending on history depth. A new fashion style normally inherits an analog or pooled seasonal profile until its own evidence is credible.

Table 4. Demand components should remain visible rather than being buried in one number.

5. Lead time, protection period and forward weeks

Figure 3. The forecast horizon must extend beyond the protection period and the decision lead time.

6. Safety stock and target in-stock level

Figure 4. Illustrative relationship: the final points of in-stock service require increasing inventory protection.

7. Parameter-driven stock forecast alternatives

Figure 5. Different methods may be appropriate by lifecycle and data sufficiency; merchants can compare the alternatives before selection.

Table 7. Forecast alternatives should be parameter-controlled and backtested.

8. Forecast parameter system and selection

A parameter-driven design permits different forecasting behaviour without rebuilding the application. Parameters should be effective-dated, versioned, inherited through hierarchy and overrideable only by authorized roles.

Table 8. A practical parameter dictionary for apparel stock forecasting.

9. Forecast use in Merchandise Assortment Planning

Figure 6. The forecast must reconcile across downstream decisions and learn from actual sales, stock and service.

The Merchandise Assortment Plan converts demand opportunity into planned breadth, depth, timing and option productivity. It should receive a baseline demand forecast plus scenario parameters, not only a single seasonal total.

Table 9. The forecast supports both financial and assortment structure decisions.

10. Forecast use in Merchandise Purchasing

Purchasing converts the forecast into time-phased receipts and supplier commitments. It must consider inventory position, forecast through the protection period, safety stock, planned markdowns, existing open orders, MOQ, packs, lead time and the remaining sellable weeks.

Table 10. Illustrative purchase-need calculation. Avoid double-counting presentation stock if it is already embedded in the target.

11. Forecast use in DC Allocation

Figure 7. DC allocation need reflects forecast demand, current inventory, safety stock, store index and size index.

DC allocation distributes constrained supply to the SKU-location combinations where it protects the most expected demand and margin. The process should calculate need first, then optimize against available DC stock, packs, store capacity and shipment rules.

Table 11. Forecast elements used in initial and replenishment allocation.

Illustrative need: SKU-store forecast = style forecast x store index x color share x size index. Net allocation need = target stock through the protection period + safety stock - inventory position, then constrained and pack-rounded.

12. Worked apparel forecast example

Assume a carryover knit style with enough history for a trend-seasonal forecast. The selected 10-week unconstrained demand is 820 units. The first effective receipt arrives after a six-week lead time and one-week review period. Forecast error over the seven-week protection period supports 72 units of safety stock at the selected service target.

Table 12. One forecast version, transformed for three downstream decisions.

13. Backtesting, KPIs and governance

Figure 8. Backtesting should compare accuracy and bias by horizon, lifecycle, class and forecastability segment.

Table 13. Forecast accuracy is necessary, but operational and financial outcomes complete the scorecard.

14. Implementation roadmap and concise rules

Executive summary

Stock optimization is not the pursuit of the highest possible inventory level. It is the disciplined selection of the lowest practical inventory position that can deliver the required customer service, presentation, replenishment and financial outcomes for each SKU-location.

The operating model separates six decisions:

• Service objective: the in-stock or fill-rate target appropriate to the product and store.

• Safety protection: the buffer required for demand and lead-time uncertainty.

• Operating band: minimum, target and maximum WOS for each policy segment.

• Investment priority: which product-store combinations receive scarce units first.

• Network action: replenish, transfer, retain, consolidate, markdown or exit.

• Closed-loop learning: compare service, turns, forecast error and inventory cost after each action.

1. The concept of optimum stock

Optimum stock is the policy-driven inventory position at which the next unit of stock is no longer expected to create enough service or margin benefit to justify its holding, transfer, space and markdown risk. It is a range, not a permanently fixed number, because forecasts, lead times and lifecycle conditions change.

Figure 1. Service gains diminish as investment rises; the economic optimum occurs before the theoretical maximum in-stock rate.

2. Service levels and safety stock

An in-stock target expresses the probability or percentage of customer demand the retailer intends to satisfy from available inventory. Cycle service level and unit fill rate are related but different measures; the policy should name which one it governs. A higher service target requires disproportionately more safety stock as the target approaches 100%.

Table 1. Standard normal service factors; use only when the statistical assumptions are appropriate.

3. Minimum, target and maximum WOS

WOS turns the stock policy into an understandable operating band. Minimum WOS is the replenishment trigger or starvation threshold. Target WOS is the desired post-action position. Maximum WOS limits excess investment and identifies transfer or markdown candidates. Each threshold should vary by product class, store class, lifecycle and lead time.

Figure 3. Illustrative active-product WOS bands; long-lead categories may require materially higher coverage.

Table 2. Illustrative active-lifecycle WOS policy. Launch and winddown require separate bands.

4. ABC classification for stock and stores

Product ABC and store ABC solve different allocation questions. Product ABC ranks the economic or service importance of the item. Store ABC ranks the demand productivity, strategic role and capacity of the location. The two classifications form a service and investment matrix.

5. Optimizing in-stock percentage and stock investment

The optimization objective should maximize expected full-price demand served and margin protected, subject to an inventory budget, store capacity, availability, pack sizes and lifecycle. It should not maximize service independently of cost, because the last fractions of service can require disproportionate inventory.

Figure 5. Optimum stock occurs near the minimum of expected shortage cost plus holding and markdown cost.

6. High-level stock optimization algorithm

Figure 6. Closed-loop optimization for initial allocation, replenishment, transfer and exit decisions.

7. Cybex Store Stock Planner operating model

The Store Stock Planner provides a practical merchant interface for the policy. The WOS slider defines starvation, healthy and surplus bands; capacity identifies required exits; the style grid ranks service gaps; and color-size detail prevents a healthy style total from hiding broken selling combinations.

Figure 7. Cybex Store Stock Planner example supplied by the user; values shown are application data, not policy recommendations in this paper.

Table 4. Connecting the planner interface to the optimum-stock policy.

8. Worked optimum-stock example

Assume an A product in an A store. Forecast weekly demand is 20 units, lead time is two weeks, review period is one week, protection-period demand is therefore 60 units, and governed safety stock is 18 units. The presentation minimum is 12 units and current inventory position is 49 units.

Table 5. Illustrative optimum-stock calculation for one SKU-location.

9. Governance, measurement and implementation

The policy should be transparent and merchant-controlled. Parameters must have owners, version history and effective dates. Recommendations should explain the forecast, service target, safety stock, WOS band, ABC classes, capacity effect and alternative source.

Table 7. Governed parameters for a controlled optimization process.

Appendix: concise optimum-stock rules

Table 9. High-level policy summary.