A Two-Stage Model of Consideration Set and Choice

Modeling the demand for a portfolio of products as a function of price is a classical problem in operations with applications in retail and revenue management. Retailers have started to change their products’ prices frequently giving them access to vast amounts of data on aggregate sales at different price configurations. According to a recent Wall Street Journal article, the price of a microwave oven has been changed at least nine times on Amazon.com and at least twice at Best Buy during the course of a typical day. [Angwin and Mattioli, 2012]. In contrast to a “sale” or a “promotion”, these intra-day price changes are frequent, may move in either direction (positive or negative), and have magnitudes that are relatively small ranging from 1% -10% of the base price. It is clear that the price changes have be set optimally, and doing that requires us to accurately predict demand as a function of prices.