How To Build A Unit Forecast For Store Replenishment

Unit Forecast An estimate of how many units will be sold, shipped, or required in a future period. Building a reliable unit forecast for store replenishment requires combining historical point-of-sale data, lead-time variability, store-level factors and promotion calendars into a repeatable process.

Replenishment forecasts should be pragmatic: accurate enough to keep service high while limiting inventory and transportation costs. This article outlines a step-by-step process, common pitfalls and example calculations operators can apply to rebuild or improve their store replenishment forecasts.

Step-By-Step Forecasting Process

Follow these operational steps to convert sales history into a store-SKU unit forecast for replenishment orders.

• 1. Collect And Clean Data: Pull POS sales, on-hand, returns and promotion tags for each store-SKU. Remove erroneous transactions and flag product launches or stockouts.
• 2. Choose Forecast Horizon: Select horizon based on replenishment cadence (e.g., daily for high-velocity, weekly for others).
• 3. Select Forecast Method: Use simple exponential smoothing for stable items; apply seasonal decomposition for seasonal items; consider machine learning for large SKUs with many drivers.
• 4. Add Exogenous Factors: Overlay promotion lifts, local events, holidays and planned price changes as adjustments to baseline unit forecasts.
• 5. Adjust For Availability: Correct historical data for past stockouts so the forecast reflects latent demand if you will be fully available in the future.
• 6. Incorporate Lead Time And Safety Stock: Translate forecasted daily/weekly units into reorder quantities using supplier lead times and desired service levels.
• 7. Automate And Monitor: Schedule daily/weekly runs, track forecast accuracy per SKU-store, and set alerts for large deviations.

Key Calculations For Replenishment

Two calculations often used when converting unit forecasts into orders are reorder point (ROP) and order quantity (EOQ or fixed quantity):

• Reorder Point: ROP = Lead Time Demand + Safety Stock. Lead Time Demand = forecasted daily units × supplier lead time (days).
• Safety Stock: Safety Stock = z-score × standard deviation of demand during lead time. Choose z based on target service level (e.g., 1.28 for ~90%).

Common Pitfalls And How To Avoid Them

Practical experience shows recurring mistakes that reduce forecast usefulness for replenishment. Being aware of these keeps systems reliable.

• Using National Forecasts For Store Decisions: Aggregated forecasts hide store variability; always forecast at the level where replenishment occurs.
• Ignoring Lead-Time Variability: Fluctuating supplier lead times can make deterministic ROPs ineffective; model lead-time distribution or increase safety stock.
• Forgetting Promotions And Price Effects: Promotions shift unit demand massively; treat them as explicit inputs rather than noise.
• Failing To Update Baselines After Assortment Changes: When SKUs are replaced or repackaged, create new baselines instead of forcing old patterns to fit.

Operational Example

A mid-sized convenience chain replenishes a fast-moving snack SKU weekly. Historical daily sales average 30 units with a standard deviation of 6 units. Supplier lead time averages 7 days. Target service level is 95% (z ≈ 1.65). Lead Time Demand = 30 × 7 = 210 units. Safety Stock = 1.65 × (6 × sqrt(7)) ≈ 26 units. ROP ≈ 236 units. If weekly order cadence implies Order Quantity = forecasted weekly demand (30 × 7 = 210) plus safety stock adjustment and round to pack sizes, the replenishment quantity might be set at 300 units to match palletizing and minimum order constraints.

Monitoring And Continuous Improvement

Measure forecast error using MAPE or RMSE at the store-SKU level and track service levels, stockouts and overstocks. Use these KPIs to tune smoothing parameters, safety stock multipliers and to identify SKUs needing special treatment.

• Daily Checks: Monitor item-level exceptions: recent spikes, zero sales, or system miscounts.
• Weekly Reviews: Reconcile forecast volumes with inbound orders and adjust parameters for upcoming promotions.
• Quarterly Audit: Re-evaluate model performance, refresh seasonality components, and incorporate business changes like new stores or lead-time shifts.

In short, the Unit Forecast for store replenishment turns historical sales and business inputs into the piece-counts needed to create orders, set reorder points and size safety stock. Build forecasts at the replenishment level, incorporate promotions and lead-time variability, and monitor performance to reduce stockouts and excess inventory while keeping stores well supplied.