The previous blog ended with a description of how a sales campaign is decomposed into a possible period, weekly, and daily product-level forecast in the master schedule. Before the discussion can proceed much further, it is important to review the demand section of the MPS grid. A schematic of the demand portion of the MPS grid is illustrated below.
Some of the nomenclature on the illustration is familiar to master scheduling. The demand time fence (DTF) is defined in the APICS Dictionary as “That point in time inside of which the forecast is no longer included in total demand and projected available inventory calculations; inside this point, only customer orders are considered.” Beyond the demand time fence, period gross requirements are determined by adding the unconsumed forecast and actual orders. The planning time fence (PTF) is defined as “A point in time denoted in the planning horizon of the master scheduling process that marks a boundary inside of which changes to the schedule may adversely affect component schedules, capacity plans, customer deliveries, and cost.”
To these two functions, three more need to be added to the master schedule grid: forecast campaign start, forecast campaign end, and forecast memory. The forecast start is easily defined as the date specified by marketing and sales as the beginning of the campaign forecast. Usually, any unsold forecast from the previous campaign is not carried over. The forecast end is the reverse: this is the date when the forecast campaign ends.
The forecast memory is the span of time back from the current date that is used to view historic forecast and order consumption and to determine past forecast carry-forward to future periods. This elastic block of time is determined from the forecast start to the current date, or a set number of “look back” days from the current date, but not less than the forecast start date. When the past history in forecast memory is either positive or negative, it must be added or subtracted from the future forecast if the integrity of the campaign forecast is to be maintained in the face of actual sales variability. There are two options possible. In Option 1, the forecast is recalculated at the current date. With Option 2, the forecast is recalculated at the demand time fence.
As will be detailed in the next blog, the forecast memory holds the key to effectively managing the future forecast. It will enable the master scheduler to see the impact of actual orders on the forecast and to ensure the proper forecast is being used in the MPS.