Being able to estimate material needs is a crucial activity for those companies looking to have available stock to cover customer requirements rather than making product to order. By forecasting the top level / finished goods demand, production and planning teams can then explode the bill of material then utilize lead times to quantify and schedule the procurement and supply of component parts.
To avoid stock outs or purchasing/manufacturing within shortened lead times accurate forecasting is critical. Faced with the task of forecasting how much of something you will need can have serious business consequences especially for businesses operating with products that have long lead times. Getting the forecast wrong can result in either surplus materials, where the forecast is higher than the actual demand or at the other end of the scale stocks outs or a failure to meet customer demand.
There are various ways in which organizations can calculate forecasts
Statistical Forecasts
For many companies the primary process for calculating forecasts is to utilize software tools. Whilst these tools are often embedded within the organizations ERP system dedicated forecasting software packages are available.
Data is taken from the ERP system including previous demand and current transactions, a standard set of variables is then configured (typically including desired service level and length of forecast) and the likely demand is then generated.
Statistical forecasts are very much based on actual data created from system transactions. The information used has to be comprehensive enough to ensure that a reliable output is produced. One of the challenges of this approach is that there may be market intelligence that the can influence the forecast that is not transactional – for example the award of a new customer contract, or a new product launch – which maybe difficult (if not impossible) to predict on historical transactions alone. This form of forecasting is also reliant on ERP system data integrity.
Non Statistical Forecasts
Another method of forecasting is without the use of statistical data. These are forecasts that have been devised from subjective estimates of what demand might be. The forecast may take a number of inputs from previous demand through to market intelligence but is often based on current or near term demand. The non-statistical forecasts can take into account sudden changes in production, sudden fluctuations that are not based on statistics, so there are times when both types of forecasts are required.
Summary
Forecasting is obviously difficult, simply because it is not an exact science and it as the supply chain is subject to sudden changes, difficulties and external factors – forecasts must be maintained. Forecasting provides the organization with a guide to future demand. Remember though, despite both experience and knowledge of the current and future environment no forecast will be fully accurate.