New Product Forecasting: Using Structured Judgment
New product launches are the most difficult to forecast. The uncertainty associated with a new product launch is much greater than that of forecasting older, more mature products. This uncertainty and the lack of historical data make traditional time series techniques impractical. Using the historical data associated with similar, previously launched products and incorporating domain knowledge in a structured process can be a useful approach to forecast the demand of new product launches. Since most companies launch many new products (as many as 10 to 15 percent of a company's product portfolio) into the marketplace annually, it makes sense to utilize the past product launches of similar product profiles to forecast the new product. Similarity techniques can be used to determine which previous product launches are likely to be useful in forecasting the new product. The new product is then forecast based on these similar product profiles using analogous forecasting techniques. Once the new product is launched, the sales can be monitored, tracked, and adjusted based on actual sales over time. This chapter discusses a new structured approach to forecasting new product launches using data mining, clustering, statistics, and domain knowledge to forecast what is referred to as evolutionary new products (e.g., line extensions).
DIFFERENCES BETWEEN EVOLUTIONARY AND REVOLUTIONARY NEW PRODUCTS
There are two basic types of new ...