Extending Bass for Improved New Product Forecasting

Forecasting demand for new products is increasingly difficult as the technology treadmill drives product lifecycles shorter and shorter. The task is even more challenging for electronic goods, where product lifecycles are measured in quarters, manufacturing processes in months, while the market volatility takes place on a day-to-day basis. We present a model that perpetually reduces forecast variance as new market information is acquired over time. Our model extends Bass’s original idea of product diffusion to a more comprehensive theoretical setting. We first describe how forecast variances can be reduced when combining predictive information from multiple diffusion models. We then introduce the notion of demand-leading indicators in a Bayesian framework that reduces forecast variance by incorporating a wide variety of information that emerges during the product lifecycle. We describe a successful implementation of this model at Intel, where one-third of the microprocessor products were tested. When compared to the current forecasting method, our model reduces forecasting time/effort from 3 days to 2 hours while decreasing forecast errors by 33%, which leads to $11.8M in cost savings over four months of demand realization.