Sales forecasting and demand planning: a brief note

When studying the statistical techniques on sales forecasting, you are told that there is a need to assess accuracy performance of your chosen forecasting model from time to time so as to tune it or replace it with a better one in the recurring forecasting exercises in corporations. The fact is, forecasting (and prediction) is a complicated topic at the theoretical level; the challenge of using forecasting is not solely at the technique level, see for example Ho (2011). It is useful to see how the statistics-based forecasting exercise fits into the overall operations management subject domain, see the following diagram:

Basically, you use real demand statstics to develop your forecasting models. Your model can involve the calculation of trend line, seasonality indexes or it can be based on exponential smoothing. You also need to take into the consideration of impacts of various events in the past and in the future, notably marketing promotion efforts by your company as well as your competitors.

You then make use of the sales forecast as major input to support your company's demand planning, which should be based on your company's Sales and Operations Planning Framework. Very  often, the demand planning process is a collaborative one involving your company's staff as well as your business partners; this planning process can take the form of CPFR. Thus, your company will not directly use your sales forecast as the demand plan nor as the figure for your sales budget.

From time to time, you do need to assess the accuracy performance of your sales forecast (e.g. using mean squared error), demand plan as well as sales budget. Moreover, you need to assess how the quality of these three sets of data affect your operational performance, such as customer service quality, inventory level, and distribution costs, etc..

At the end of the day, you do need to tune your sales forecasting model to improve its performance; at the same time, you can also try to reduce the various lead times in your logistics cycle so as to reduce your company's reliance on forecasting for demand management. You can try to compress your logistics cycle lead time with the JIT approach or with the technique of postponement, for examples.

References

1. Demand planning: http://searchmanufacturingerp.techtarget.com/definition/demand-planning
2. Forecasting: http://en.wikipedia.org/wiki/Forecasting
3. Gattorna, J.L., Ogulin, R. and Reynolds, M.W. (editors) (2003) Gower Handbook of Supply Chain Management, Gower.; note Chapter 2.7: "Forecasting and demand planning" by Scott F. Githens.
4. Morris, C. (2003) Quantitative Approaches in Business Studies, Prentice Hall.; note Chapter 17: "Forecasting: time-series, semi-log graphs and exponential smoothing".
5. Sales and Operations Planning: http://en.wikipedia.org/wiki/Sales_and_operations_planning