This paper reports the pre-launch calibration and validation of an individual level diffusion model used to focus management actions during a new product launch. Diffusion models have attracted considerable attention in the marketing literature as a way of understanding the market evolution of new products. While these models have proved extremely flexible in explaining dynamics after sales have peaked, there are few management applications prior to launch. Those examples that do exist rely primarily on management judgment and analogy to similar past products rather than consumer feedback for calibration. One specific type of diffusion model developed to address the pre-launch calibration problem, individual level diffusion models, has shown promise in the laboratory and at the aggregate level post launch, but has yet to be implemented in a practical managerial application. We discuss the modeling, estimation and measurement issues in using an individual level diffusion model in practice and demonstrate the performance of one for the pre-launch sales forecasting of a new software product. The model enables us to estimate how many consumers will buy, when they will buy, who will buy, and why they will buy. We validate forecasts for the model and compare them to those using traditional methods of estimating aggregate model parameters.
