Prof. Aurélie Lemmens on managing churn to optimize profit
Proactive retention programs are dependent on the proper strategic decisions on who to target, which size to address with an incentive or when to target those customers. Prof. Lemmens’ presented research focuses on the first question of who to target and the optimal target size in order to optimize the profitability of retention campaigns. Her work includes the definition of a profit-based loss function for the financial impact of a retention intervention for each customer. The conducted experiments in her research showed that this approach leads to more profitable retention campaigns. Concluding, Prof. Lemmens argues that the loss function should become a general part of model specification, especially when targeting specific individuals.
Furthermore, during the research seminar she explored the problem of model validation with the PHD students. For one, in order to overcome overfitting problems, she suggests to adopt a 3-step process, that is being applied within machine learning for some time already. By separating the data sample into three parts, which are then independently used for calibration, model selection and validation, the outcome is expected to avoid the issue of in-sample performance. This approach amongst others, can also be found in her research on churn management.
The Marketing department at WU thanks Prof. Lemmens for her time and the sharing of her knowledge.