Forecasting new product trial with analogous series

Abstract

This study develops a simple method for forecasting consumer trial for national product launches. The number of consumers who try a brand in its first year on the market is accurately predicted from the number trying the brand in the first thirteen weeks following launch. No information about the specific category or marketing activities is required– just a simple multiplier computed from analogous series in other markets. These analogues provide an empirical generalization that can be easily applied by practicing managers to track and forecast the success of new brand launches. When subject to an out-of-sample test involving 34 fresh data sets, the analogues demonstrated 43 percent reduction in Mean Absolute Percentage Error compared to the most accurate marketing science model.