Contributors: Ruvisha Pillay, Bill Tonetti, Tracy Roche
Netstock has created a unique forecasting process to fully leverage dynamic, flexible hierarchies in Pivot Forecasting. In this article, we aim to give you an overview of that process.
In Pivot Forecasting, there is a sequence to adjusting, saving or locking in changes to forecasts:
Promotions modify Calculated Forecasts
Adjusted Forecasts override Calculated forecasts
Saved Adjustments override Calculated Forecasts as well, and their retained so they will continue to override calculations when history is reimported and forecasts are recalculated; and
Locked Forecasts override all other forecasts and adjustments.
These changes can happen at any level of aggregation, on any hierarchy; even in alternative units like price or cost. Our article on Pivot Forecasting - The Role of Measures in a Model describes these measures further.
In the diagram, above, note the sequencing of the rows. They are shown in this order to help users to understand the prioritization of overrides. The Calculated Forecast is the starting point. Next are Promotions, which modify the Calculated Forecast. The remaining rows are overrides, with the Locked Forecast being the most “powerful.” All adjustments impact the Synchronized Forecast. And, as a final step, Synchronized Forecasts are “Finalized”, which copies them into the Final Forecast row. The system will never modify the Final Forecast. Only users can do this.
Pivot Forecasting is a 3-step process:
The system generates a Calculated Forecast
Users review and adjust forecasts
Forecasts are finalized and moved into the Final Forecast measures
Step 1
The Calculated forecast is automated. However there are steps that can be taken for improvement, to reduce the number of manual forecast adjustments that may need to be made in the future. For example, you can turn off forecasting for discontinued items, correct for non-repeating events using Promotions, or capture and reuse seasonal indexes.
Step 2
Next, users apply their judgment by making adjustments. Users typically begin their review at aggregate levels, or by working with filters that narrow down the number of items to review. Most adjustments should be entered into the Saved Adjustment row. These adjustments are retained until a user removes them. If a forecast is far into the future, you may not want to save the adjustments because that will cause the system to ignore calculated forecasts which will improve as new history is imported. In that case you can use the Adjusted Forecast row. Changes made in the Adjusted Forecast row are the least permanent. They aren’t saved so your future forecasts will be able to adapt as new demand history is imported. On the other extreme are Locked Forecasts, which override them all.
Step 3
The last step is to finalize the plan, which will then store the scenario to the Final Forecast measure. The Final Forecast is passed to your inventory planning app and it will be the first step in determining your optimal supply plan. The system will not change the Final Forecast measure and it will log changes with the date, user, previous and changed values. Users finalize by clicking on the “Finalize” button near the top of the screen.
You may also be interested in the below articles: