Forecasting
Forecasting
Peter, an aggressive product manager for the widget product line for XYZ Electronics, was in a management meeting. The topic for discussion was his last quarterly sales forecast. It was only half of what sales actually had been, and as a result the manufacturing manager, Mary, needed to run the plan on overtime for six weeks after laying off people during the previous quarter. The controller, Charlie, was upset over the large fluctuations in the inventory level, and the general manager, George, recalled the order they bid below cost because of projected undercapacity. They were all discussing Peter's low forecast.
Mary: You are always low and wrong. Where do those numbers come from anyway, a random number table?
Peter: They come from the salesforce, who are on top of the market. You must realize that they don't like to put speculative orders into their forecast because they look bad if forecasted sales do not materialize. Because of that I actually increased their estimates somewhat.
Mary: We obviously need a better system and fast.
Charlie: I've been looking at the data over the last few years and the forecast is always low at this time of year. I think there is a seasonal effect brought about by the governmental purchases at the end of the fiscal year. Did you consider the seasonal effect?
Peter: No, it wasn't necessary because the salespeople already factor that into their estimates.
Charlie: You hope.
George: Didn't we hear that our competitor, Ajax, was at capacity and therefore probably would raise prices? Did the additional sales come from the region in which Ajax always has been strong?
Peter: That region was high but I've no information to suggest that it was due to a price change by Ajax.
George: Perhaps these forecasts signal a long-term change in the growth of the product line. Perhaps we should be adding plant capacity.
Peter: That's a possibility. I've always said that the widget line would take off with more sales and advertising support.
Charlie: In the meantime we need some short-term forecasts so that this debacle isn't repeated.
Mary: Amen.
Peter: You Monday-morning quarterbacks are really perceptive. I'd like to see you make forecasts instead of criticizing previous forecasts, for a change. Our best quarter ever and you complain about the forecast.
Forecasting, as the story illustrates, provides the basis for almost all planning and control. If the forecasts are unreliable, it is most difficult to make the right tactical or strategic decision. Try to think of decisions made by any organization that do not in some way rely on a forecast of demand or sales. The story also indicates the varied considerations that can go into a forecast, considerations like seasonal effects, salespeople's biases, competitor's price, and long-term growth trends.
Peter was attempting to make short-term forecasts of the sales of an existing brand. There actually is a variety of types of forecasts. Figure 22-1 provides an overview of some of the ways in which forecasting tasks can differ. Forecasts can be short-term (3 to 6 months), medium-term (2 to 3 years), or long-term. The methods used and the forecast accuracy will be sensitive, in part, to the length of time to be forecast. It usually is much
By Time Period
Short term (3 to 6 months)
Medium term (2 to 3 years)
Long term
By Object
Brand sales or usage
Product class sales or usage
Other consumer variables like attitude or intentions
Macro variables like GNP, unemployment, or interest rates
By Segment
Geographic area
Customer type
Product application
By Product Maturity
New products
Growth products
Mature products
Declining products
FIGURE 22-1
Types of forecasting tasks.
easier to forecast for short periods into the future because fewer factors will change in the short run.
There are countless objects that can be the focus of the forecast. Among these possible objects are brand sales (or usage) and product-class sales (or usage). Thus, a symphony manager might be interested in forecasting not only symphony attendance but the attendance at all cultural events, or product-class usage. Product-class sales can help refine the brand-sales forecast and also provide guidance to the marketing program. If attendance to all cultural events were expected to grow, that might have implications for the forecast for symphony attendance and also for ways to promote the symphony.
Often it is necessary or desirable to forecast by market segments or subgroups. A forecast by geographic area might be needed to plan warehouse-stocking decisions or to evaluate regional salesforce efforts. Long-range planning might need forecasts of customer types or of product applications.
In Chapter 21 the focus was on forecasting demand for new products. In this chapter the focus will be on existing products and existing markets.
There is a variety of approaches that can be used for forecasting. Figure 22-2 summarizes those that will be discussed in this chapter. They are grouped into three categories. The first are qualitative in nature, such as the salesforce estimates Peter used. The second are time-series approaches
Qualitative Methods
Juries of executive opinion
Salesforce estimates
Surveys of customer intentions
Delphi
Time-series Extrapolation
Trend projection
Moving average
Exponential smoothing
Seasonal and cyclical index
Causal Models
Leading indicators
Regression models
FIGURE 22-2 Forecasting approaches.
where historical data are projected into the future. The third are causal approaches in which factors causally related to the forecast are identified. For example, the price charged by Ajax might be an important causal influence of widget sales and should be considered explicitly in the forecast.
Peter, an aggressive product manager for the widget product line for XYZ Electronics, was in a management meeting. The topic for discussion was his last quarterly sales forecast. It was only half of what sales actually had been, and as a result the manufacturing manager, Mary, needed to run the plan on overtime for six weeks after laying off people during the previous quarter. The controller, Charlie, was upset over the large fluctuations in the inventory level, and the general manager, George, recalled the order they bid below cost because of projected undercapacity. They were all discussing Peter's low forecast.
Mary: You are always low and wrong. Where do those numbers come from anyway, a random number table?
Peter: They come from the salesforce, who are on top of the market. You must realize that they don't like to put speculative orders into their forecast because they look bad if forecasted sales do not materialize. Because of that I actually increased their estimates somewhat.
Mary: We obviously need a better system and fast.
Charlie: I've been looking at the data over the last few years and the forecast is always low at this time of year. I think there is a seasonal effect brought about by the governmental purchases at the end of the fiscal year. Did you consider the seasonal effect?
Peter: No, it wasn't necessary because the salespeople already factor that into their estimates.
Charlie: You hope.
George: Didn't we hear that our competitor, Ajax, was at capacity and therefore probably would raise prices? Did the additional sales come from the region in which Ajax always has been strong?
Peter: That region was high but I've no information to suggest that it was due to a price change by Ajax.
George: Perhaps these forecasts signal a long-term change in the growth of the product line. Perhaps we should be adding plant capacity.
Peter: That's a possibility. I've always said that the widget line would take off with more sales and advertising support.
Charlie: In the meantime we need some short-term forecasts so that this debacle isn't repeated.
Mary: Amen.
Peter: You Monday-morning quarterbacks are really perceptive. I'd like to see you make forecasts instead of criticizing previous forecasts, for a change. Our best quarter ever and you complain about the forecast.
Forecasting, as the story illustrates, provides the basis for almost all planning and control. If the forecasts are unreliable, it is most difficult to make the right tactical or strategic decision. Try to think of decisions made by any organization that do not in some way rely on a forecast of demand or sales. The story also indicates the varied considerations that can go into a forecast, considerations like seasonal effects, salespeople's biases, competitor's price, and long-term growth trends.
Peter was attempting to make short-term forecasts of the sales of an existing brand. There actually is a variety of types of forecasts. Figure 22-1 provides an overview of some of the ways in which forecasting tasks can differ. Forecasts can be short-term (3 to 6 months), medium-term (2 to 3 years), or long-term. The methods used and the forecast accuracy will be sensitive, in part, to the length of time to be forecast. It usually is much
By Time Period
Short term (3 to 6 months)
Medium term (2 to 3 years)
Long term
By Object
Brand sales or usage
Product class sales or usage
Other consumer variables like attitude or intentions
Macro variables like GNP, unemployment, or interest rates
By Segment
Geographic area
Customer type
Product application
By Product Maturity
New products
Growth products
Mature products
Declining products
FIGURE 22-1
Types of forecasting tasks.
easier to forecast for short periods into the future because fewer factors will change in the short run.
There are countless objects that can be the focus of the forecast. Among these possible objects are brand sales (or usage) and product-class sales (or usage). Thus, a symphony manager might be interested in forecasting not only symphony attendance but the attendance at all cultural events, or product-class usage. Product-class sales can help refine the brand-sales forecast and also provide guidance to the marketing program. If attendance to all cultural events were expected to grow, that might have implications for the forecast for symphony attendance and also for ways to promote the symphony.
Often it is necessary or desirable to forecast by market segments or subgroups. A forecast by geographic area might be needed to plan warehouse-stocking decisions or to evaluate regional salesforce efforts. Long-range planning might need forecasts of customer types or of product applications.
In Chapter 21 the focus was on forecasting demand for new products. In this chapter the focus will be on existing products and existing markets.
There is a variety of approaches that can be used for forecasting. Figure 22-2 summarizes those that will be discussed in this chapter. They are grouped into three categories. The first are qualitative in nature, such as the salesforce estimates Peter used. The second are time-series approaches
Qualitative Methods
Juries of executive opinion
Salesforce estimates
Surveys of customer intentions
Delphi
Time-series Extrapolation
Trend projection
Moving average
Exponential smoothing
Seasonal and cyclical index
Causal Models
Leading indicators
Regression models
FIGURE 22-2 Forecasting approaches.
where historical data are projected into the future. The third are causal approaches in which factors causally related to the forecast are identified. For example, the price charged by Ajax might be an important causal influence of widget sales and should be considered explicitly in the forecast.
Comments
Post a Comment