What it is
One of the key components of the order proposal calculation is demand. Demand can be made up of
a variety of sales and promotion related data series on finished items or
demand from components used in the preparation of finished items.
The statistical forecast is used in calculating demand unless you override it. Open the item card to view total demand for an item.
Special order items
AGR calculates demand for items marked as special order items from their open sales orders.
AGR automatically calculates a statistical forecast for each item in a location once a month by default, although the forecast procedure can be set to run weekly. Some events can trigger an item/location reforecast outside of the regular cycle. These include
opening an item card for an item with no forecast or an invalid forecast
changing the First Sale Date Into Forecast setting in Item Details to limit historical sales data
editing the sales history of an item in the item card or plans.
AGR automatically selects the best forecasting model for each item from its
sales pattern and
length of sales history available for the Item
When the sales pattern for the item in that location changes, AGR automatically selects a new best fit model to generate the next statistical forecast.
Statistical forecast and time periods
AGR uses a maximum of 36 complete months of sale history by default when calculating monthly statistical forecasts, although you can change this in settings. If you have selected weekly forecasting, AGR uses the item's sales history up to the last complete week of sales.
The amount of sales history AGR uses to calculate the statistical forecast can be overridden by changing the First Sale Date Into Forecast setting in Item Details.
For example, you could change the date when the item's history prior to a certain date was not typical for this location and is unlikely to be repeated in future.
Use the Forecast Start Date and Forecast End Date settings in Item Details to specifically define when a forecast should start or end. This is a useful setting to use for items which are coming to the end of their active life. If you do not require an item to have a forecast, toggle the Exclude From Forecast setting in Item Details to Yes.
By default, a statistical forecast will be calculated for future months. The monthly statistical forecast values are then distributed equally over the weeks within a month.
The monthly Statistical Forecast values can be configured in settings to be distributed by weekly trends in historical sales within each month.
A weekly statistical forecast is distributed equally between week days, although it can be configured to include Saturdays and/or Sundays.
Alternatively, the day distribution of a weekly statistical forecast can be configured in settings to use the day distribution seen in historical sales.
Forecast key points
AGR uses historical sales data, among other things, to calculate forecasts. Items with no or little sales data or with seasonal sales patterns have some forecast limitations.
For a new item, AGR needs a minimum of 1 sales point in the prior month or week before calculating a statistical forecast.
Where less than 5 data points exist, AGR will calculate a simple moving average statistical forecast
To reflect seasonal patterns in the forecast, an item in a location needs to have at least 2 years of historical sales data. AGR applies a Seasonal label to items it detects as being seasonal and sets the value in the Seasonality column to Yes.
Monthly or weekly forecasting
Both forecasting durations have their pluses and minuses. Monthly forecasting captures seasonality better and is more suited to items where the order period > 1 month. Weekly forecasts are more responsive to recent changes in demand and should be considered for items with a short order period, for example < 7 days.
Ending a forecast
To prevent order proposals being calculated when an item is coming to the end of its life, the forecast can be ended by using one of two forecast settings in the Item Details tab of the Item Card.
I don't like the forecast calculated. What can I do?
AGR's statistical forecast uses actual sales data from the ERP. AGR does not know whether the sale was made in the normal course of business or was exceptional. You may need to make manual adjustments to the sales history based on your knowledge of a specific sale. Adjusting an item's sales history will impact the calculation of the forecast.
Alternatively, if you think the forecast predicted by AGR needs changing to, for example, pick up an upward trend more quickly, we recommend overriding the forecast in Plans instead of manipulating the historical sales. It is likely that the forecast will eventually pick up on the upward trend on its own. This can easily be done in the Plans tab of the Item Card or through the Plans module itself.