The most important question in forecasting is that of the determination of the objective or purpose behind the forecast. The purpose determines the accuracy required of the technique and the techniques chosen for forecasting. The techniques vary in the cost, scope and accuracy. What information is necessary for production planning may be different in terms of detail and scope from what is needed for market planning. Supposing there were 100 different products produced by a stamping process. For the production planner to find out what capacities are needed in the future in terms of heavy stamping or light stamping, the detailed forecast and the factors influencing the 100 products is not at all necessary Probably, a simple time-series model provide him with adequate information for planning. This is not the case with market planning.
Similarly, if no changes are being made in the internal company policy, general trend models may be sufficient. Whereas if there are some strategic actions to be taken then very detailed analysis for each of the alternative actions is necessary.
Depending upon the life cycle of the product and the stage at which the product exists at the time of forecasting, the techniques and the detail with which the forecasting information is gathered will vary. It can be easily understood that in the new product development stage, information has to be gathered regarding similar products in the market or the consumer opinion has to be collected regarding what the product market is likely to be. It may also be necessary to find what input output analysis has to offer in terms of projecting the broad future changes in the economy. Even a Delphi method might be useful in such cases; or one may use what it called the Product Differences Measurement Techniques were one compares a proposed product with competitorsâ€™ present and planned products and ranks it on a quantitative scale for different factors. Whereas, in the steady-state of the product life cycle, the main considerations are trend and seasonal effects and the time-series analysis might be good enough in such situations.
In short, there is no single forecasting technique which is always of more value than others; rather one or more techniques may be relevant depending upon the situation. Forecasting although is a science it is also an art in this sense.
Besides defining the purpose of the forecast, it should also be noted that forecasting should be a combined effort organizationally. Since the external and internal environment of the company cannot be perceived in its complete perspective by an individual or a functionary, it is necessary to include a number of functionaries in the forecasting procedure. Such a procedure helps people who use the forecast for different functional planning processes to get an insight into what is behind the forecast figures. A poor understanding of the sales planning by the production planner affects his production planning too.
To do a good job of materials planning it is necessary to know, in addition to what is planned to be produced, also the reasons as to why it is done. It means, a sales planner, production planner and materials planner, must collectively involve in the forecasting process.
Another point to be noted in forecasting is that quite often wrong things are forecasted. For instance, one need not forecast the requirement the requirement for bought out an internally manufactured component parts going into an assembly. The forecast for the final product may be good enough to calculate the requirements. Similarly, the materials requirements for a project oriented industry need not be forecasted. They can be read from the PERT/CPM network. Very often people go for minute forecasting for all the odd products that they manufacture. Often it is enough if one has information regarding the families of products rather than individual products. In fact, the forecasting for a family of products is more accurate.
Another pitfall in forecast is that there is no timely tracking of forecasting. The deviations of the forecast from the actual should be monitored continuously or periodically. This is necessary to find out as to whether the assumed environmental conditions have changed in the present or for the future.