Profit From Your Forecasting Software

CHAPTER 10New Product Forecasting

10.1 INTRODUCTION

Making accurate forecasts of the demand for new, or nearly new, products can be challenging. There will be little or no sales history data that we can use for extrapolation or to estimate relationships between sales and predictor variables. In addition, management teams who have invested time and resources in the development of the product may be tempted to make overly optimistic estimates of its chances of success.

Problems like these have led to the development of a range of methods that are intended to improve the accuracy of new product forecasts. For example, intentions surveys involve asking potential customers to indicate the probability that they would buy the new product. Similarly, in choice modeling, potential customers are asked to indicate their preferences between products with different combinations of attributes (e.g., smart phones with different screen sizes, battery discharge times, and prices). The analyst then models customer preferences to estimate the probability that they will choose a particular product in preference to others and, from this, a forecast of the product's market share can be made.

In this chapter, we will focus on the use of statistical time series methods in new product forecasting. When a product has yet to be launched there will, of course, be no demand history to allow the model to be estimated. However, if similar products have been launched in the past, we can fit models to their ...