Forecast demand and inventory

Forecast demand and optimize inventory at retailers

When forecasting SKU demand, setting safety stock, and planning replenishment at multi-location retailers

• · POS data, historical sales, seasonal patterns, promotional calendar, lead times

• · Demand forecast per SKU, safety stock levels, suggested purchase orders

• · Demand planning suite (Blue Yonder, Oracle Demantra, Kinaxis)
• · ERP (SAP, Oracle)
• · WMS for DC inventory
• · POS data feed

• · 40-200 store network (multi-location setup)
• · 300-800 SKUs across categories
• · 12+ weeks historical data for baseline
• · Service level targets defined (fill rate %)

• · Moving average used on seasonal item (lags trend by half-window)
• · Blended MAPE hides deterioration by cohort
• · Promotional lift estimated from sparse promo history (overfitting)
• · Safety stock set without demand variability input
• · Lead time change not factored into replenishment timing
• · EOQ/MOQ constraints not applied before order rounding

• · Forecasting method selection by ABC/XYZ classification and demand pattern
• · Seasonal decomposition (STL, X-13ARIMA) available for shifting patterns
• · Safety stock formula using demand variability + lead time variability + service level target
• · Promotional lift estimation with cannibalization offsets
• · ABC/XYZ analysis framework (A=high volume, C=low; X=stable, Z=intermittent)
• · Forecast accuracy monitoring (MAPE, bias, by cohort)