Skip to main content
All CollectionsHow to use SKU Science
🚀 Why and how to create a forecast at an aggregate level
🚀 Why and how to create a forecast at an aggregate level

This article describes how to improve forecasts by calculating them at a higher level.

Updated over 11 months ago

Why create forecasts at an aggregate level

For the reasons listed below, it can be beneficial to create your forecasts at an aggregate level rather than at the article level, which is the default setting.

  • The accuracy of forecasts is always better when calculated from larger volumes.

  • If in your business, you regularly replace certain items with newer versions (e.g. if you market mobile phones) due to the product life cycle, it is challenging to detect seasonality and trends unless you work at the product family level.

To do this, you must first go to the Forecast Edition in order to create an aggregation.

If you are not yet familiar with this part, you can go to this article to learn how to create aggregations and save them. Once you have at least one aggregation, you will learn below how to use it to recalculate all your forecasts.

How to calculate forecasts at an aggregate level

In addition to viewing your aggregations, you have the option to recalculate all forecasts using one of your saved aggregations.

To relaunch the calculation of forecasts from this level, click on the calculator icon located at the top on the black toolbar.

The following window appears. Simply check "Compute forecast from aggregation" and choose the saved aggregation from the list available in the dropdown menu.

Click on Confirm & Launch to recalculate all forecasts starting from the Family level. The platform then automatically initiates a series of calculations.

It begins by calculating the best forecast for each family.

Then, based on these forecasts made for each product family, it will automatically allocate each value found for each period at the level of each lower-level forecast, as defined in your account.

If your forecasts at the lowest level are defined at the SKU level, the platform will update all SKU-level forecasts with the new data from this higher-level calculation.

If your forecasts at the lowest level are defined at the SKU x Customer level, all these forecasts will be updated using the same allocation method.

Example from the demo account

By using the data from the demo account, it is possible to quickly understand how it works.

Below, the platform displays the forecast for Family 4 before launching the calculations at the Family level. The displayed forecast is the sum of all forecasts for the individual items that make up Family 4. In this case, the forecasts were calculated at the item level.

If we launch the forecast calculations at the Family level, we then obtain the new result below for each Family. It can be observed that the forecast results differ at the Family 4 level. For instance, in June 2019, in the first case, the displayed value is 1,675, whereas in the second case, the value becomes 1,728.

The more underlying items you have, the more noticeable the impact will be on your forecasts. In the demo account, the number of references is limited, so the impact is slight.

In the two examples below, one can similarly compare the values for 8/2019. Indeed, there is a change from a quantity of 1,613 items to 1,648 when calculating the forecasts starting from the aggregate Family level.

How to calculate forecasts at the lowest level

To cancel out the effect of this calculation and obtain forecasts again for each item independently of the Family, simply click on the calculator icon and uncheck the checkbox. By confirming, the platform restarts the forecast calculations using the default level set on the platform.​

Did this answer your question?