Dynamic e-targeting using learning spatial choice models

Abstract

Most CRM applications use static scoring methods to target customers; however, the fast-changing environment of the Internet often demands dynamic models that are able to adapt to changes “on the fly.” Here, we focus on location-based dynamic targeting of customers by proposing spatial models of choice that adapt to dynamically changing environments. These learning spatial choice models incorporate new information as it becomes available and are superior to their static counterparts. We estimate these models using a version of the Expectation-Maximization (EM) algorithm. We illustrate their application in the context of an online publishing firm's data on customer choice and explain how they can be useful in setting targeted e-coupon promotions or dynamic pricing application. Our studies and simulations show that learning models have superior predictive capabilities over a pure static approach and thus are ideally suited for dynamic environments.