Counterfactual Techniques for Enhancing Customer Retention

Mustafa Zaki, Davis Banks, Mandy Barrett Korpusik, Vaishampayan Pathuri, Kishore Paul

Keywords: counterfactual, e-commerce, BERT

Abstract: In this paper, we introduce a novel method for generating counterfactual explanations to convert customers from an e-commerce company who frequently add items to their cart but do not proceed to checkout. We demonstrate that our method i) allows customization of mutable features, improving the practical applicability of our counterfactual explanations, and ii) outperforms existing techniques including DiCE, GANs, and CFRL in key metrics such as reconstruction error and L1 distance, while maintaining a low latency. Our method can easily be customized to optimize for high coverage or low latency by adjusting the number of nearest unlike neighbors, highlighting the trade-off between these competing goals.

Primary Area: causal reasoning

Submission Number: 2193