Demand Forecasting
Demand forecasting predicts what will sell and when, giving inventory decisions a number to aim at.
What demand forecasting is
Demand forecasting is predicting what will sell, how much, and when: projecting future demand per product from history, seasonality, trends, and planned events, so buying and stocking decisions have a number to aim at.
Why demand forecasting matters
Every inventory decision is a forecast wearing a purchase order: buy too little and stockouts spend the demand you paid to create, buy too much and capital ages into markdowns. The forecast is where those errors get smaller.
What feeds a forecast
- Sales history, cleaned of stockout-distorted zeros
- Seasonality and the promotional calendar
- Trend and lifecycle signals per product
- Planned launches, ads, and known demand shocks
Frequently asked questions
How accurate should a forecast be?
Accurate enough to beat the naive guess, measured: track forecast versus actual per category, and let safety stock absorb the honest error. A forecast without an error metric is a mood.
Statistical models or machine learning?
Simple models forecast stable sellers shockingly well; ML earns its complexity on promo-heavy, trend-driven, or sparse-history catalogs. Start simple, measure, and upgrade where the error justifies it.