Continuous Improvement Training Program
A Continuous Improvement Program
Featuring 8-Training Modules by Bill Gaw

Increase the effectiveness of your
Lean Manufacturing Training Program
Lean Manufacturing Simulation Game 

What to Expect in an Improvement Effort

At its core, demand forecasting is just another application of business judgment, but in a setting that is uncomfortable for most of us, namely, the future. As with many business functions, persistence, attention to detail, and creativity help to improve the results. On-again-off-again efforts, sloppiness in the details, and "quick fix" approaches tend to yield disappointing results. You don't need expensive software and PhD mathematicians to produce useful fore­casts. You can start with a PC, some common software packages, and a small team of analysts. In fact, the costs of forecasting, in most examples we have seen, are small in comparison with the amounts typically expended for JIT, Quick Response, systems projects, and other operations changes.

Think of a forecasting improvement effort as a reengineering project, rather than a mathematical or technical exercise. It needs a champion, contributions from many people throughout the organization, time for reflection and testing of ideas, and importantly, an implementation phase to embed the new processes within the organization. Its success will depend as much on these management issues as on its technical strengths.

A few key characteristics to expect in forecasting improve­ment efforts include:

1. The effort transcends many departments and functions within the company. Better forecasting means re­quires a better understanding of customers, and that knowledge may reside several places, in bits and pieces. Sales and Marketing are obvious candidates, but other useful sources may include Inventory, R & D, Quality, and Discussions with upper management, manufacturing, purchasing, and possibly some suppli­ers, help to build a picture of when forecasts are needed and how they will be used.

2. Plan for improvements in stages, not one "giant leap." One key to forecasting is to break down the uncertain­ties affecting demand for your products into their component parts. Often these parts make good targets as stages in the improvement effort. As an example, you might start by improving forecasts for just one group of customers, or just work on better forecasting of the timing of demand, before moving on to other sources of uncertainty.

3. Getting early information usually pays handsome divi­dends. Test markets, finding "bellwether" customers, observing or visiting trendsetters, and similar advance ways to sample customers' interest almost invariably improve your forecasts or reduce the level of risk. Even so-called fashion products have probably been tried somewhere before, for example, in Paris studios. For one "new" technology in electronics, we found similar concepts that had been patented and used as far back as 1947! Brainstorming and creativity sessions may generate fresh ideas for getting early insights into customers' interest in a new item. Typically, advance scouting efforts pay for themselves many times over.

4. Automating the process (modeling) contributes more than just numbers. Over time, even a simple model will usually outperform human forecasters. If the model­ing process is managed well and doesn't become lost in details, it will uncover gaps in knowledge and sharpen people's thinking. Operationally, models have obvious benefits for forecasters' productivity, by freeing them from mechanical calculations. But, there is a Dark Side to the Force of modeling:
The more the forecast process resembles a technical "black box," the more resistance it will encounter. Neural nets, Box-Jenkins models, and heteroscedasticity are fine for shop talk among technical professionals, hut they fail the "user friendly" test with operating units. Add about a month to the implementation phase for every time you say one of these words in public.

5. Improved forecasting processes save time for upper management. Most complaints about poor forecasting relate to current operations, and forecasting improve­ment efforts often originate with operating managers. But, in developing a new forecasting system, most companies expand their understanding of such things as the drivers of demand for their products. Such knowledge, by reducing the range of speculation and debates about such things as market trends, or the likely impact of a proposed marketing program, often helps decisions about future actions move faster through upper management reviews and committees.
Rather than being a technical assignment for mathematics specialists, a forecasting improvement project is a mission to better understand your customers. As with quality improvement programs, it should yield new insights for all levels of management about your business and the funda­mental forces driving it. And, as with quality programs, its benefits build upon one another, year by year.

To be Continued