Peak sales time prediction in new product sales: Can a product manager rely on it?

Managers dealing with new products need to forecast sales growth, especially the time at which the sales would reach the peak, known as the peak sales time (T*). In most cases, they only have a few initial years' data to predict T*. Although product managers manage to predict T*, there is no method to date that can predict T* accurately. In this paper, we develop a new metric based on the diffusion modeling framework that can help in assessing the prediction accuracy of T*. This metric is built on the premise that observed sales growth is affected both by the force that systematically varies with time and by the non-systematic random forces. We show that the two forces must be carefully combined to assess if a predicted T* is accurate enough. In addition, we empirically prove the efficacy of the proposed metric.