Bass Forecasting Tutorial - Overview

Overview

The Bass forecasting model is a tool for forecasting the time-path adoption of new products and new product categories within a target population. It implements the original Bass model (Bass 1969), as well as its extended version, the generalized Bass model (Bass, Krishnan, and Jain 1994). The generalized model expands on the original Bass model by including the effects of advertising and price changes.

The software provides two modes for calibrating the model: (1) by analogy and subsequent refinement (i.e., visual tracking) and (2) by fitting the Bass model to past data using nonlinear least squares (Srinivasan and Mason 1986).

Firms can use the Bass forecasting model to develop marketing programs based on the estimates of future product sales rates for new durables (i.e., products, services, or technologies that are not purchased frequently by a customer).  The model can be “calibrated” either based on parameters obtained from available historical data of the penetration of the product in a target population over a defined time period, or based on the time paths of adoptions of similar or analogous products in the target population.

Technical Note

Bass forecasts rely either on a continuous-time model of the adoption phenomenon, or its discrete-time approximation. While the continuous model is more theoretically justified and is usually recommended, it cannot easily accommodate changes (e.g., growth) in potential market size. For consistency, the discrete model is used throughout in Enginius. The user should note that comparisons with the continuous version of the Bass model
may lead to slightly different parameter estimates or forecasts.