An algorithm for on-line price discrimination

The combination of on-line dynamic pricing with price discrimination can be very beneficial for firms operating on the Internet. We therefore develop an on-line dynamic pricing algorithm that can adjust the price schedule for a good or service on behalf of a firm. This algorithm (a multi-variable derivative follower with adaptive step-sizes) is able to respond very quickly to changes in customers' demand. An additional advantage of the developed algorithm is that it does not require information about individual customers. Given the growing concern about customers' privacy this can be of great practical importance. Computational experiments (with different customer behavior models) indicate that our algorithm is able to successfully exploit the potential benefits of on-line price discrimination.