Home // DATA ANALYTICS 2019, The Eighth International Conference on Data Analytics // View article

Data-driven Direct Marketing via Approximate Dynamic Programming

Authors:
Jesper Slik
Sandjai Bhulai

Keywords: email marketing; Markov decision processes; approximate dynamic programming; recommendation systems

Abstract:
Email marketing is a widely used business tool that is in danger of being overrun by unwanted commercial email. Therefore, direct marketing via email is usually seen as notoriously difficult. One needs to decide which email to send at what time to which customer in order to maximize the email interaction rate. Two main perspectives can be distinguished: scoring the relevancy of each email and sending the most relevant, or seeing the problem as a sequential decision problem and sending emails according to a multi-stage strategy. In this paper, we adopt the second approach and model the problem as a Markov Decision Problem (MDP). The advantage of this approach is that it can balance short- and long-term rewards and allows for complex strategies. We illustrate how the problem can be modeled such that the MDP remains tractable for large datasets. Furthermore, we numerically demonstrate by using real data that the optimal strategy has a high interaction probability, which is much higher than a greedy strategy or a random strategy. Therefore, the model leads to better relevancy to the customer and thereby generates more revenue for the company.

Pages: 63 to 68

Copyright: Copyright (c) IARIA, 2019

Publication date: September 22, 2019

Published in: conference

ISSN: 2308-4464

ISBN: 978-1-61208-741-2

Location: Porto, Portugal

Dates: from September 22, 2019 to September 26, 2019