Putting a new forecasting system in place usually meets with a good deal of resistance for a variety of reasons – some good and some are just smoke screens to avoid accountability. Let’s look at the good, the bad and the ugly to try to uncover the real truth.
- “The forecast is always wrong so why do it.” True, but there are degrees of error and a consistently well thought out forecast is the best attempt at predicting product demand. This is the key upstream driver in many downstream enterprise operational processes such as inventory planning, capacity planning and vendor management.
- “Our business can’t be forecasted.” If it’s a pure make-to-order business with customers that will tolerate long lead times and sometimes uncompetitive prices then don’t worry about forecasting. Most businesses are not afforded that luxury and at a minimum must do some rough cut forecasting to obtain better material pricing and improved customer delivery rates. There are many elements of demand planning that can be forecasted ― even in a semi make-to-order environment.
- “High touch is usually high error.” When it comes to overriding a system-generated forecast, research has shown that small changes to the forecast produce more error. On the other hand large changes can produce a significant improvement. This seems counter intuitive, but the real reason is that small changes represent an attempt to tweak a scientifically generated forecast that in the end is best handled through automation. A large change represents special knowledge of future events like the gain or loss of a customer, which a time series-based, historical forecasting system would have no knowledge of.
- “Automation can solve more than one problem.” Very true. Automating the forecasting process by using time series statistical forecasting tools can not only improve accuracy, but add consistency and predictably into the outcome by taking the human bias out of the equation. Also, automation can reduce the time to produce the forecast and free up demand planners to add value to the overall sales and operations planning process.
- “Collaborative forecasting is more pain than it’s worth.” Again research clearly shows that the best forecast is produced using a combination of time series-generated baseline forecasts combined with a multiple stake holder review and override process. The problem in many cases is getting the stake holders such as Sales to employ the discipline and step up to the accountability of producing a companywide, useable forecast at a level of detail that will help planners. Results produced have shown that a carefully structured, formalized sales and operations planning process is a necessity. Additionally, the best results are produced when technology is carefully used to meet the needs of collaborators rather than having them meet the technology requirements.
- “More detail is better sometimes.” True, if we are referring to the amount of historical data to use in the time series forecasting process; but false if we are referring to the depth and breadth of the data used in the process. Adding too much granularity such as customer, ship-to, SKU, size, style and color can over complicate the process and ultimately produce bad results. When consistently bad results appear (from inaccurate forecasts), then there is lack of credibility in the forecast and the process. The end is near for that planning process, and unfortunately a reversion back to “old ways” usually occurs.
- “More work upfront usually produces less work in the end.” Absolutely true. If we avoid paralysis by analysis and clean and link the necessary data required to do both a good time series baseline forecast and provide the important data needed by the collaborators, then the process will move forward successfully.
There are many more of these forecasting myths that in some cases can help us to understand what not to do and how to resist some of the headwinds facing progress. Change in these types of processes has always been difficult because the old ways are deeply entrenched by a company’s old corporate knowledge workers.
This post was written by John Hughes