Demand Forecasting Best Practices

Part 2. Measuring forecasting quality

In part 1, we defined our objectives: forecasting unconstrained demand on the appropriate horizon and granularity to support supply-chain decisions. In this second part, we will discuss forecasting quality. Simply put, we want to assess is a forecast is good or bad. We will start by introducing different forecasting KPIs in chapter 8 (Bias, MAE, MAPE, RMSE), then discussing their pros and cons in chapter 9 (Spoiler, MAPE is the worst). In chapter 10, we will answer a central question to demand planning: “What is a good level of forecast accuracy?” by using benchmarks. Finally, in chapter 11, we will extend our KPIs to assess the forecasting quality of a whole product portfolio using value-weighted metrics. ...