Ways to Improve Sales Forecasts

Sales forecasting as been given a reactive role...inaccurate assumptions do not stem from lack of techniques...knowing drivers of sales demand is crucial to success of sales forecasts...forecasters who rely on single-variable forecasts take risks...sales demand forecasting can be extremely important for strategic decisions. Recent history is inundated with stories of corporations and in some cases entire industries that have made strategic errors because of inaccurate sales forecasts. In fact, for years, sales forecasting has been given a reactive role in these organizations. Its placement downstream in the corporate hierarchy, usually Operations/Planning, has reduced the function to nothing more than a clerical support position. Furthermore, corporate executives have tried to automate the function by investing enormous sums of money in large one dimensional blackbox mainframe systems. Meanwhile, users (management) of sales forecasts continue to question practitioners (preparers) why sales forecast error continues to escalate. Their reluctance to recognize sales forecasting as a viable solution to reducing costs (i.e., lowering inventories), increasing sales volumes (i.e., reduction of backorders/improved Customer Service), and most importantly, identifying those variables who's relationships impact sales demand have only compounded the problem. Inaccurate assumptions do not stem from a lack of sales forecasting techniques because regression analysis, historical trend smoothing, and others have been available to most companies for years. Unfortunately, they share a fundamental mistaken assumption that relationships driving sales demand in the past will continue into the future. They believe the only two factors driving sales demand for any given product are seasonality and trend. Furthermore, both can be derived solely from the items past history. None realize that history for a product can be an unreliable guide as the marketplace becomes more competitive, new products enter the category, line extensions proliferate the shelf, and, most importantly, new channels of distribution emerge (i.e., Warehouse Club Stores). In fact, refocussing sales efforts from one product to another, alone, can change its trend dramatically. Most companies don't foresee changes in consumer behavior or understand the impact of market dynamics, such as saturation points, price sensitivity, coupon redemption and advertising flighting. As a result, many users (managers) have come to distrust traditional techniques. Some actually accept poor sales forecasts as part of doing business and proceed to implement such philosophies as JIT (Just-In-Time Manufacturing) and ECR (Efficient Consumer Response) to better manage inventories and reduce lead times. Their assumption is JIT/ECR are all conclusive in solving their manufacturing inefficiencies and inventory overheads. The results are some reductions in lead times and/or a minimal reduction in Finished Goods inventories. Unfortunately, they missed the basic concept that both JIT and ECR are only as good as the sales forecast driving them. You know the old Junk-In-Junk-Out theory. The end result is they (managers) throw up their hands and continue to develop business plans under the assumption there is no such thing as a good sales forecast. I completely disagree with this metaphor. It is possible to develop reliable projections into the future based on an in-depth understanding of the factors that impact sales demand. Many of which we have absolute control over such as price, advertising, coupon drops, consumer promotions and trade promotions. Understanding the relationships of those variables driving sales demand many times make the difference between an accurate projection and one that flounders. In other words, sales forecasts that are made up of all the factors influencing sales demand (also known as proactive sales forecasts) are much more accurate, besides being informative, than sales forecasts based on shipment history alone (also known as reactive sales forecasts). …