Generating Performance Reports
In these example, a company performs sales forecasting each month. Forecasting is done in monthly time buckets. The M3 DMP administrator wants to report the forecast accuracy measured on a two-month horizon, that is,. how accurate the forecast is two months ahead of time. The forecasters are able to modify their forecasts up to one month before, so data needs to be transferred to the report at appropriate times to preserve data history.
Example
To create and maintain the Performance Report starting in July, the M3 DMP administrator must initially generate the report two months before – in May – including the forecast data for July. In this way, the data history for July is preserved so that any future changes to the forecast are not included in the measured forecast accuracy.
A favorite view is created that uses one or more filters with the desired key filtering and includes the desired Measures. The period filter applied must ensure that data is included for all desired Measures in some period. This is necessary for the Measures to be created in the report and to enable Update of those Measures. The favorite view could be aggregated on some keys. If the changes made to the forecast data are not committed, the appropriate change scenario should be linked to the favorite view.
In May sales statistics are available up until March. The periods included in the favorite view must be March and July: March to create the Sales Measure and July to establish the snapshot of Forecast data on a two-month horizon. Note that if Forecast data is included for May-June the horizon for those periods will not be two months.
In June, the Forecast Measure is updated for August. The forecast for July has changed since the report was generated. However, the modification for July made during May should not be included in the performance report with a two-months horizon, so this period is not updated.
The Forecast Measure is updated in this way in July and August for September and October respectively. In September, the Forecast Measure is updated for November and the Sales Measure now available for July is updated.
This procedure is repeated in October where the Forecast for December and Sales for August is updated and in November where the Forecast for January and Sales for September is updated. The report now includes:
- Sales data for March which should be ignored in this context
- Data for July-September which measures forecast accuracy for Q3 on a two-month horizon
- Forecast data for October-January that is the basis to continue to measure forecast accuracy on a two-month horizon for those periods.
A Data Indicator on the Diff% measure has been added, highlighting forecast deviations greater than 4.7%.
The Performance Report on a two-month horizon for Q3 is now ready for distribution.