Individual and Company Success, Come Discover the
Understanding Forecast Characteristics
Hal Mather has written: "The forecasting problem is really two
problems, quite different in concept and approach. The first is,
'How do you make better forecasts?" The second is, 'How do you use
forecasts more effectively?' He goes on say that the real question
must be: "How do I use forecast data, with its embedded accuracy
problems, most effectively to run my business?" (1)
1. Forecast is always wrong (or lucky)
The best results will come when the preparation of the forecast is a multi-discipline task. It must involve the users who must have, through education and training, an understanding of the techniques and the results. They must have valid expectations and a recognition why forecasts fail.
Why Forecasts Fail
The eight most common causes why forecasts fail are:
Source data are fundamental to forecasting. Quality forecasts demand quality data which are appropriate to the situation. Source data will be a combination of statistical data and qualitative data, including surveys and managerial opinions. Irrespective of the source, data, to be valid, should meet the following criteria:
Data quality and accuracy
Determine if base data represent orders or shipments, both in terms of timing and quantity. Then be consistent. Avoid recording errors by using single entry and by installing demand filter to flag unreasonable data.
Level of Detail
The marketing forecast must reflect appropriate forecast periods (:i.e. annual, quarterly, monthly) without excess detail which is costly to handle. Planned promotions must be reflected. In the case of dating programs, recognize the difference between demand, shipment and cash flow.
Data External to Demand History
For best results, nontypical demand has been removed from the base demand history as used for quantitative forecasting. This requires the ability to make such separations in the historical data while allowing for use of marketing and sales intelligence. There also should be a capability to identify portions of the forecast as being such "intelligence" and the source.
Dimensions of Data
Allow for conversion of data to the correct manufacturing unit-of-measure. Failure to use a constant unit-of-mea-sure can cause major data errors and bad decisions.
It is a not uncommon practice to state forecasts in non-production dimensions such as dollars, kits, etc. This is acceptable as long as there is adequate definition of any planned pricing changes or any anticipated change of product contents
To be Continued
To stay current on manufacturing competitive knowledge, please subscribe to our weekly bulletin, "Manufacturing. Basics and Best Practices (MBBP)." Simply fill in the below form and click on the " subscribe button."
We'll also send you our Special Report, "6-Change Initiatives for Personal and Company Success."
All at no cost of course.
Your personal information will never
be disclosed to any third party.
Here's what one of our subscribers said about the MBBP Bulletin:
"Great articles. Thanks for the insights. I often share portions of your articles with my staff and they too enjoy them and fine aspects where they can integrate points into their individual areas of responsibilities. Thanks again."
Kerry B. Stephenson. President. KALCO Lighting, LLC
Knowledge you’ll not find at offsite