Demand forecasting is one of the central features of Supermind. Fully automated and statistics-driven demand forecasts allow you to improve forecasting accuracy and reduce the number of human errors.
The benefits of automated demand forecasting
- Provides daily forecasts for all items at all locations
- Automatically takes into account campaign-related sales spikes and previous availability issues
- Allows you to react quickly to changes in demand
The forecasts provided by the system are time series forecasts, which are calculated on a daily basis for one year ahead. Daily demand forecasts are used in the demand calculation, and they are converted into concrete purchase orders based on the balance forecast.
The demand forecast is calculated fully automatically at the location/item level for all items. The assortment links and categories of items and locations can be imported from the back-end system or managed in Supermind.
Assortment management
Life-cycle management
The system has several different demand forecast models that work according to the set parameters and conditions. The parameters and conditions are defined in cooperation with the customer during system implementation. The parameters can also be easily changed by the customer.
Forecasting models
Supermind has several different forecasting models that enable, for example, trend forecasting, seasonal forecasting, and moving average forecasting.
All models provide time series forecasts based on past demand. The forecasts take into account e.g. previous availability and campaign-related demand spikes. In other words, the sales history used in the calculation of forecasts is converted into demand history.
The system calculates a forecast for all items in all locations using all forecast models. The forecasts, which are ultimately used in demand calculation and automated replenishment, can be generated, for example, by product group.
The trend forecast model usually works best for products with steady demand, while the seasonal forecast model is a useful option for products with seasonal demand.
The system saves the forecasts in time series, which can be viewed in the system’s user interface. With time series selections, you can retrieve data for a single time series, e.g. for the past three years and the next one year.
The system divides the sales time series into trend and seasonal components. The forecast is therefore more accurate than, for example, a comparison with the previous year’s sales.
Examples of exceptional situations
Availability issues in the past
When there are problems with product availability, it also affects sales. SUPERMIND corrects the sales history from the time when the product has not been available. The corrected sales figures are then used to calculate the forecast.
Sales spikes during previous campaigns
Campaign sales can be seen as a clear spike in the sales history. SUPERMIND can automatically filter significant spikes to the level of normal sales. The spike-filtered sales figures are used to calculate the forecast.
Information about campaigns can also be imported into the system, in which case automated filtering is not needed. In this way, you can also filter campaigns in which sales growth has not been statistically significant.
Would you like to know more?
We would love to show you how SUPERMIND works and what the results look like. We believe we can convince you in just one hour. Contact our team and book a demo meeting with us!
Petri Lindholm
+358 400 420 583
[email protected]