5 Forecasting hierarchies

 

I have seen countless companies forecasting demand at irrelevant aggregation levels (material, geographical, or temporal) that do not match the information granularity required to make supply chain decisions. Many supply chains—especially manufacturers—typically rely on populating 18-month forecasts per country by monthly buckets. Should this be considered a best practice, or is it merely a by-default, overlooked choice?

In this chapter,10 we will discuss forecasting hierarchies in detail. You will learn to assess your demand planning process’s relevant material, geographical, and temporal aggregation levels (or dimensions).

5.1 The three forecasting dimensions

Demand forecasts are defined across three dimensions (or hierarchies): materiality, geography, and temporality.

  • Materiality: You can forecast demand using different material levels: per SKU, product, segment, brand, and so on. As well as with different measuring metrics: units, value, weight, type of raw material required, . . . .
  • Geography: You can forecast demand per country, region, market, channel, customer segment, warehouse, store, zip code, . . . .11
  • Temporality: You can also use different time buckets (daily, weekly, monthly, quarterly, or even yearly).

We will discuss in section 5.3 why you could prefer one aggregation level over another. Finally, on top of these three dimensions, we will have to discuss the forecasting horizon:

5.2 Zooming in or out of forecasts

5.3 How do you select the most appropriate aggregation level?

5.3.1 Which aggregation level should you focus on?

5.3.2 What granularity level should you use to create your forecast?

Summary