Contributors: Ruvisha Pillay, Bill Tonetti, Tracy Roche, Perry Britt

In this article we aim to give you an understanding of the role of measures in a model. Before you continue, be sure to read our introductory article, "Model Introduction."

In Pivot Forecasting, measures are numeric values which are displayed as data rows. Examples of these are Saved Adjustments, Final Forecasts and Synchronized Forecasts. They are time-phased, and they can be factored by financial or unit of measure conversions. Measures can be historic, future, or both historic and future, and each period is a single data point in the model.

We will explore some of the most important Pivot Forecasting measures as we continue.

Actual History is the primary historic measure. It typically is shipment, invoice or sales order history transactional data. It is sourced from your Netstock app. Every other historic measure adjusts, or adds to this measure.

The Copied History measure reflects the combined transaction history, generated using supersession links maintained in your Netstock app. Products or packaging, for example, may be periodically replaced with newer versions. With supersessions, the transaction history of the new product is combined with previous one so that more data is available for accurate forecasting. Both of these measures contribute to the

Resultant history, which is then used for statistical forecasting. In addition to Supersessions, historical promotions and other events can also affect the Actual History. A period of stockout could result in less sales than would have actualized if the item had been in stock. A temporary price reduction or promotion can also increase demand over a period. In both of these examples, forecasts would be more accurate with a process to exclude the events to create a baseline forecast.

Direct adjustments to history can be made in the Adjusted History measure. If you enter a value in the Adjusted History data row, the statistical forecast uses that value instead of the Actual History quantity. Adjustments to history only work on base items.

In most cases, it is better to use Promotions instead of Adjusted History. In Pivot Forecasting, the Promotion row facilitates correcting for non-seasonal historic and future events. If a promotional lift is anticipated in a future period, a positive value captured in the Promotion measure will be added to the Calculated Forecast measure data point. As time passes and the period with the promotion rolls into history, the input value will still be used. It will reduce the Resultant History, thereby "cleaning" the effects of the promotion and making a better baseline forecast. Since Promotions are additive (or subtractive), Pivot Forecasting uses them at any group level and on any hierarchy. Historic values in the Promotions measure are subtracted from actual history to create a baseline of demand, effectively excluding the effects of promotions. A negative Promotion value can be input if an event caused a reduction in historic demand, for example from a stockout.

The statistical forecast is the starting point for all Pivot Forecasting plans. The artificial intelligence of the forecasting engine uses “Expert” model and parameter optimizations to generate the statistical forecast. The outcome of the statistical forecasting process is presented in the Calculated Forecast measure row.

You may find yourself needing to override the Calculated Forecast for a variety of possible reasons. Perhaps the sales team has informed the planning team of a new customer who has the intention of increasing your total sales by 30%. This will be a good reason to override the Calculated Forecast. Similarly, you may need to increase or decrease forecasts based on expected trends or product lifecycle changes.

When you input and save values in either the Adjusted Forecast, Saved Adjustment, or Locked Forecast measure, they will modify the Synchronized Forecast. Adjustments can be added and saved as many times as needed. The Final Forecast row will only be changed once the user finalizes the forecast via the menu options.

It is important to note that the Adjusted Forecast measure row is cleared when the model or items are re-forecasted. These kinds of adjustments are very useful for longer range forecasts. As the history rolls over, there are more data points available for generating an improved calculated forecast.

Sometimes reasons for adjusting the forecast are more permanent. Saved Adjustments are more useful for short term adjustments, such as forecasting a new item. There is no history on which to forecast demand, so Saved Adjustments can be left in place until there is enough history to generate a meaningful forecast.

If there is a need for a changed forecast to remain fixed, the forecast needs to be locked down. The Locked Forecast measure is isolated from statistical forecasts and any other forecast adjustment, including changes made at other levels in the hierarchy. The locked forecast takes precedence over all other sources and reasons for forecast updates.

There is a sequence to the application of forecast adjustments. All adjustment methods override the Calculated Forecast. Saved Adjustments override Adjusted Forecasts, and Locked Forecasts override them all. The Synchronized Forecast measure displays the results of this sequencing.

Pivot Forecasting automatically produces expert forecasts. If the Calculated Forecast is reasonable, then the forecasting review is complete. If there are reasons to moderate the Calculated Forecast, you can make use of the Promotions measure, or any of the 3 Adjustment measure rows to capture the adjustments needed. All changes will be immediately reflected in the Synchronized Forecast measure. Finalizing the forecast will depend on a User’s role. Power users finalize into the Final Forecast measure. Collaborative users will usually finalize into the Sales Forecast reference measure, where power users can review and subsequently copy the results into the Final Forecast row. Final Forecasts will not be updated by the system. Users control this measure and it is the final plan that is used for inventory planning.

For an introduction to how measures flow through hierarchies, refer to our related articles on proration and aggregation:

For more information, you can also check out our Pivot Forecasting collection!