Product Growth Models for Medium-Term Forecasting of Short Life Cycle Products

Abstract

Short product life cycles are becoming increasingly common in many industries. Traditional approaches to medium-term forecasting are not designed for the type of information available (or the lack thereof) in the short life cycle environment. A typical demand curve for these products consists of rapid growth, maturity, and decline phases coupled with seasonal variation. With reference to product demand curves of a personal computer (PC) manufacturer, we suggest the use of information on total life cycle sales and the peak sales timing to obtain initial monthly forecast in the absence of a sales history. Three growth models are presented in which such information can be utilized to estimate the parameters. We also outline procedures that use demand history of prior products to estimate the seasonal variation in demand. Using data on PC products, we empirically validate the models and compare their fit and forecast performance with ARIMA models. We show that the accuracy of the forecast made multiple periods ahead using two of the three models investigated is comparable to that made one period ahead using ARIMA models. Empirical observations and issues relating to the implementation of the models at a PC manufacturer are also discussed.