Benchmarking Forecasting Practices in Corporate America

Starting in 2002, we have been updating the forecasting practices in corporate America annually. This is the fourth update. Benchmarks are basically averages - average forecast error, average salary in the forecasting profession, average background of a typical forecaster, and so on. Benchmarks can help in two important ways. First, a benchmark encourages you to find out where your company stands and, second, you can see how it compares with other companies. If your company is doing below other companies, then the mission is clear. Something has to be done to bring your forecasting up to the industry level. If it is at par with other companies or above them, then you may like to raise the bar. Benchmarks given in this issue are based on the survey conducted by the Institute of Business Forecasting (IBF) at the five forecasting conferences and tutorials held in 2005 in the United States. The participants who attended the forecasting conferences and tutorials were forecasters or had plans to become one. Of these participants, 67% of them came from companies with sales revenue of $500 million and over. It is difficult to cover all the topics in one issue of the journal; but what we have covered here include forecasting process, forecasting error, forecasting models, forecasting software and systems, salary of forecasters, new product forecasting, and background of forecasters. FORECASTING PROCESS The success of a forecasting function depends on people, process, technology, and resources. Of these, process is the key because it is the one that links people to technology and resources. The better the process, the more effective will be the forecasting function. Therefore, the process should be well thought out and well structured. What kind of process other companies are using can guide you in establishing (or improving) your own process. What is forecasting process? The process covers numerous activities. One way to define it is to describe all the activities involved, which would be the answers to the following questions. What types of forecasts are needed (e.g., order forecasts, shipment forecasts, and financial forecasts) and at what level of details (e.g., forecasts by regions, channels of distribution, and customers)? How often and how far ahead should forecasts be prepared? How often should forecasts be updated and revised? What data/information are needed to prepare them? From where would the data come and how? Where should we place the forecasting function? Who should participate in the process? Which forecasting software/systems and models should we use? Which forecasting philosophy should we follow-onenumber forecast or multiple-number forecasts? Which forecasting approach should we use-bottom-up, top-down, or somewhere in the middle? Do we need a monthly consensus meeting to review the forecast numbers? Who will have the authority to override forecasts? Do we need a Sales & Operations Planning (S&OP) process and/or a Collaborative Planning, Forecasting and Replenishment (CPFR) program? To establish a process (or improve it if it is already in place), we need support from the top, which is the topic of next section. Management Support: As mentioned earlier, the success of a forecasting function depends very much on people, process, technology, and resources, none of which are forthcoming unless there is support from the top. We need resources to start and maintain the forecasting function. We need resources to buy forecasting software and/or systems, syndicated data, and tools to store and analyze data. We need resources to build an infrastructure to communicate data/information/forecasts to stakeholders who are both within and outside the enterprise. Also, for a forecasting function to work efficiently and effectively, we need collaboration among all the stakeholders, internal (Sales, Production, Marketing, and Finance) and external (Distributors and Customers). So, nothing will be accomplished unless you have support from the top. …