DEMAND FORECASTING: REALITY vs. THEORY
WHAT WOULD I REALLY DO DIFFERENTLY , IF I COULD FORECAST DEMAND ?
NATIONAL MANAGEMENT SCIENCE ROUNDTABLE NASHVILLE, TENNESSEE MAY 13, 1991
Steven Robeano Senior Logistics Engineer Ross Laboratories 6480 Busch Boulevard Columbus, Ohio 43229 (614) 624-6124
You know, I must be one of those people the airline has in mind when the pilot gets on the PA system just before take -off and says, "Good morning, you are on Delta Airlines flight 1424 to Nashville." First: I've never gotten on a wrong plane. Second: Don't they look at your ticket when you check-in? It doesn't matter, I still worry about it until I hear the pilot's confirmation. Before we get down to business here, I'll describe the flight we are on for people like me. SUMMARY Formally or informally, the forecast drives the firm. Techniques range from seemingly none at all, or at most "forecasting by desire", to complex and "sophisticated" algorithms and models that require tremendous processing power. Usually, it's a mixture of dozens of methods and options. "What works and what doesn't, rarely depends directly on time, money, or other forms of business horsepower. It is an organizational problem, subject to a visible formal structure and also to an often hidden informal one. This session will look at forecasting as a form of applied common sense. In a commercial business setting, simple doesn't necessarily mean weak and advanced mathematics doesn't necessarily mean accurate or timely. Sales and Marketing functions in a firm often use several forecasting methods in an effort to improve accuracy. The Logistics and Operations functions also often use several methods, with each area complaining that the other's forecasts are not compatible with their area's needs. There is a lot of resistance to reconciliation. Often the cause is simply organizational inertia. Other times each area has good reason for being so parochial. We are going to look at organizational ways to turn these conflicts into constructive cooperation. In the '70s and '80s many firms with large distribution systems, including Ross, installed computerized statistical forecasting systems for logistics. I have talked with many businesses before, during, and after their selection and installation processes. Generally all of them showed an improved level of detail and improved accuracy. The methods and the software may vary somewhat, but most were carefully studied, tested, and evaluated before implementation. Users have lots of rosy expectations when planning a new system. Ross Logistics, for instance, wanted lower finished goods inventory and lower freight costs without compromising customer service. Others firms may want to drive a production or purchasing system. Ross got what it wanted, and so it seems, did most others. But... (did you ever notice that the important stuff always comes after the "but" ? ) But are whiz-bang forecasters like us getting everything we had hoped for? One benefit many of us expected was a total integration of the forecast throughout the business,
Demand Forecasting: Reality vs. Theory
Page 2 of 14
or in other words, getting everyone in the choir to sing off the same script. This is a constant struggle between departments in any large business and by the end of this session you should have a better idea why. Hopefully, you also will be able to identify and deal with similar problems in your own organization, problems which odds strongly lead me to suspect your organization shares, visible or not. As time allows, we would like some information from you on your experiences. I'm not here to teach you forecasting. Any one of you could teach it as well as I could. In planning this session I assumed you had a solid understanding of the forecast methods you are using in your own shop. To be candid with you, I also expected that you were probably at least moderately uncomfortable with both how and why the various parts of your...
Please join StudyMode to read the full document