DYNAMICALLY MANAGING A PROFITABLE EMAIL MARKETING PROGRAM

摘要

We apply the proposed model in a non-contractual context where a retailer operates a large-scale email marketing program. We find that a 3-state HMM model fits the data the best based on the BIC value. The intercept of the State 3 is higher than those of the State 1 and 2. We label both the States 1 and 2 as "occasional" and State 3 as "active"-purchase state, respectively.