Modeling And Analysis Of Multi-stage Marketing Messages Delivery Problem

In the industry of E-commerce,Internet etc, Marketers usually use the delivery of marketing messages to deliver information to the customers and attract the customers to buy their products, Considering the delivery of marketing messages, marketers in the interactive environment must decide to send which messages to their customers and the numbers of each message. We assume that customer are already clustered into homogeneous segment according to predetermined rules. We consider a multi-stage marketing message match problem to a customer segment. We build the delivery model using the method of multi-stage dynamic programming, and we assume that success probability of each message has a beta prior distribution, every stage we use the result of the stage before to update the parameters of the success probability of message. And then we give a solution method of this multi-stage marketing message delivery problem, a approximate dynamic programming approach based on the Lagrangian multipliers. In the numerical analysis, we show that our method outperforms approaches which ignore the effects of information gain.