Want more Insights from SAS? Subscribe to our Insights newsletter. Or check back often to get more insights on the topics you care about, including analytics, big data, data management, marketing, risk and fraud.
Practical advice for better business forecasting
3 videos highlight tips from new book
By Mike Gilliland, SAS Product Marketing
There are serious problems with the practice of business forecasting. With the quantity and variety of data growing each year, along with computation power and modeling sophistication, we would expect a progression of improved business forecasting performance.
But this doesn't seem to be happening. There are few endeavors where so much money has been spent, with so little payback.
A recent study of eight companies in the UK found that more than half of their sales forecasts were less accurate than a naïve “no change” model (forecasting no change from the latest sales period). How could this be? How could companies be spending time and resources to generate forecasts that were often worse than doing nothing and just using the naïve forecast?
The new book, Business Forecasting: Practical Problems and Solutions, addresses this and many other vexing issues faced by today’s business forecaster. The book provides practical advice for business forecasting and is intended for those engaged in, overseeing or relying on the output of an organization’s forecasting process.
The ultimate message is that challenges in business forecasting, such as increasing accuracy and reducing bias, are best met through effective management of the forecasting process. But effective management requires an understanding of the realities, limitations and principles fundamental to the process.
When management lacks a grasp of basic concepts like randomness, variation, uncertainty and forecastability, the organization is apt to squander time and resources on expensive and unsuccessful fixes. Business Forecasting: Practical Problems and Solutions provides a critical exploration of these fundamental concepts that are needed to improve organizational forecasting performance.
Check out these videos for three quick tips that can help improve your forecasting, and then download the first chapter for free.
The disturbing 52%
The role of a naïve forecasting models is explored in several of the articles. It is generally agreed that the “no change” model, which forecasts future observations to have no change from the most recent observation, provides a worst case for forecasting performance. In other words, if your forecasts are no more accurate than the naïve forecast, why even bother going through the effort?
The video clip “The 52%” previews a disturbing research study that found more than half of the sample forecasts across eight companies were less accurate than the no change model. It goes on to provide data segmentation strategies to focus improvement efforts where there are the greatest opportunities, and shows how to set appropriate forecast accuracy targets.
Forecasting performance benchmarks – answers or not?
It is only natural for organizations to want to compare their performance to industry peers, but do industry benchmarks provide the answer?
The video clip “Forecasting Performance Benchmarks” provides a preview of two articles that point out the futility of published forecasting benchmarks. While there are legitimate issues with the trustworthiness of benchmark surveys, the fundamental problem is lack of comparability. Different companies don’t have equally forecastable data. A company may have best-in-class forecast accuracy simply because they have the easiest to forecast demand, not because their forecasting methods are the best.
The authors argue instead for “internal” benchmarking against a company’s own data, comparing their forecasting performance to a naïve model.
FVA: Get your reality check here
The typical business forecasting process has considerable costs – for IT systems, software implementations, training and management time for those engaged in the process. But companies rarely step back to evaluate the true value that is added by all this effort in terms of increased forecast accuracy.
The popular method of Forecast Value Added analysis is previewed in the video clip “FVA: A Reality Check on Forecasting Practices.”
Traditional performance metrics, like MAPE, tell you only the size of your forecast error, but nothing about the effectiveness or efficiency of your forecasting process. Using FVA analysis, organizations can identify wasted efforts and streamline forecasting processes. By eliminating the process steps or participants that are just making the forecast worse, fewer resources are spent on forecasting, with more accurate results.
New book offers practical advice
The 49 articles in this new book, with the editors’ commentary, deliver a critical look at basic forecasting concepts, and the realities and limitations of the business forecasting process. Its four main sections cover:
- Fundamental Considerations in Business Forecasting.
- Methods of Statistical Forecasting.
- Forecasting Performance Evaluation and Reporting.
- Process and Politics of Business Forecasting.
Michael Gilliland is the marketing manager for SAS forecasting software, editor of the Forecasting Practice section of "Foresight: The International Journal of Applied Forecasting" and author of The Business Forecasting Deal. He has written numerous articles for forecasting publications worldwide. Mike holds a BA in philosophy from Michigan State University and master’s degrees in philosophy and mathematical sciences from Johns Hopkins University. Follow his blog The Business Forecasting Deal.
- Read the first chapter of the new book, Business Forecasting: Practical Problems and Solutions
- Order now: Business Forecasting: Practical Problems and Solutions
- Find out how forecasting helps Wescom Credit Union save millions of dollars
- Download this white paper for a step-by-step guide to Forecast Value Added analysis
- For more information on large-scale automatic forecasting from SAS, check out SAS Forecast Server. Or, if yours is a small to midsized business, see SAS Forecasting for Desktop