Academy
ESGet my diagnostic
All guides

Guide 4 of 20

Demand planning

Forecasts per product, model comparison and ABCD classification.

Open the module
Who uses it: Planning and sales4 min

What it's for

Knowing how much will sell is the starting point for purchasing, inventory, production and distribution.

Step by step

  1. 1

    Overview

    Forecast accuracy, average error (MAPE), forecast versus actual month by month, and alerts such as products with high error.

  2. 2

    Forecast studio

    Pick a product and an algorithm (or let the app optimize it) and generate the forecast with its confidence band. "Compare all models" tries every method and tells you which is most accurate, with MAE, MAPE, WMAPE and bias.

  3. 3

    ABCD classification

    Classifies products by their share of revenue and how variable their demand is. Use it to decide where to pay more attention and which inventory policy to use.

Example: A catalogue filter

For one Acme filter, the studio picks Holt-Winters with a 13.3 % error. That forecast, with its 95 % band, then feeds MRP, DRP and S&OP.

Tips

  • A positive bias means the forecast tends to run high: check whether past promotions inflate the history.
  • Scenarios let you try "what if we sell 10 % more?" without touching the official plan.