Forecasting helps you to be less wrong.
In our personal lives we would be reluctant to identify what we will be doing in three month’s time, over the period of a week. Yet, we are quite willing to identify a single point figure as the sales for an item over that same week in three month’s time.
Why is it that we know to forecast our personal life is full of provisions for error, yet we are happy to state and debate single point forecasts of a commercial product?
A forecast is the anticipation of a number of future orders from customers, hoping the pattern of demand for the future will be similar to the past and assuming no risk of disruption. Although we can modify the time series by applying four analysis of the data set – trend, cycle, seasonality and randomness; we know that the more complex our business and its supply chains, the higher the uncertainty and the likelihood of problems creeping up on us.
While some problems are external, up and down each supply chain, the majority are under the control of your business; within your inbound, internal and outbound logistics operations.
As properties of your performance factors interact dependently, independently and interdependently, they create emergent results (that is, you do not expect them); which become cumulative – what appear to be minor concerns in themselves, become major when linked with other small events.
Changing your approach for better outcomes
Complexity associated with developing a forecast means that having a single point number attached to an item for each period is potentially dangerous, especially if people in your organisation view forecasts as ‘accurate’.
Instead of a single point forecast, use a range, identifying optimistic and pessimistic forecasts, together with the probabilities that reflect your (or your team’s) likelihood of them occurring. Multiplying the forecasts by their probability and weighting the results is likely to provide a more believable indication of future demand by SKU or product line.
In addition to the benefit of this approach to Sales, it is also beneficial in Procurement. The items with long lead times can be ordered at the optimistic forecast level; the items that are easy to obtain and on short lead times can be ordered at the pessimistic forecast level. As cover for ‘unknowns’ provisional orders are raised so the amount required can be increased at short notice.
If your forecasting module does not have the facility to incorporate range forecasts, it is over to a spreadsheet; but however it is done, you will find that range forecasts improve forecast accuracy – especially if you measure the ‘forecast error’ between forecast and actual!