Weekly Demand Forecasting by SKU to Optimise Purchasing and Stock
The Purchasing and Supply Chain Director needs to predict exactly how many units each reference (SKU) will sell next week across 150 warehouses in order to automate supplier orders.
Where the money comes from
What to do now
- 1Measure current stock-outs, shrinkage and forecast error by category.
- 2Classify SKUs by margin and turnover and set the target service level.
- 3Pilot automatic ordering in 10 warehouses against a control group.
For your committee
One page with the decision, the impact, the risks, the steps and the KPI to track.
The print dialog opens: choose "Save as PDF".
Watch out: more volume is not always more profit
Maximising availability on every SKU raises stock and shrinkage on low-margin references. Set the service level per SKU according to its margin and turnover: not everything deserves 99%.
KPI for the committee
Stock-outs (%), shrinkage (%) and days of stock by category.
Cash
Reduces stock by ~€15M (cash).
Detail for your data team
Proposed technique: XGBoost
Why: Demand by SKU depends on complex interactions between price, promotions, day of the week and weather. XGBoost far outperforms simple models on historical time-series tables with multiple cross variables.
Cost of getting it wrong: Over-buying creates write-offs and storage costs; under-buying causes direct loss of sales and customer penalties.
Data needed: ERP / POS: Historical daily sales by SKU and warehouse over the last 3 years.; Purchasing and Marketing: Active promotions, selling price, discounts and supplier lead times.; Calendar and Weather: Local holidays, day of the week and weather forecast.