Forecasting purchase rates of new products introduced in existing categories

One of the challenges product managers face in formulating strategic decisions is how to effectively forecast the performance of a new product in the marketplace. This paper develops a stochastic model, which is then applied to household purchase data to predict ultimate trial and first repeat purchase rates for a new brand introduced in an existing category of common frequently purchased goods. Previous research directed at the estimation of these two common indicators of new product success, as well as critical components of volume forecasting models, has focused on the elapsed calendar time to purchase or repurchase of the new brand. However, typical scanner panel data also contain information regarding consumer purchasing behavior of other brands in the category. The model developed in this paper utilizes this richer category purchase data from a household scanner data to estimate these two critical forecasting parameters: ultimate trial and first repeat rates. Results from four instances, including two cases involving new products and two cases involving existing brands, are used to demonstrate the value of the category purchase approach in comparison to the elapsed calendar time to purchase approach. Results show that the category purchase approach is more accurate in predicting the percentage of clients who would ever try a new product in the market (i.e., eventual trial) and in predicting the first repeat rate. Results also demonstrate how this projected eventual trial and first repeat rates could be used to develop a better estimate of long-term market share. We demonstrate that this approach provides an improved forecasting tool and recommend that marketers should, when possible, take advantage of this method.