Predicting Customer Churn in the Telecommunication Industry by Analyzing Phone Call Transcripts with Convolutional Neural Networks

Abstract

For telecommunication service providers, a principle method for reducing costs and generating revenue is to focus on retaining existing customers rather than acquiring new customers. To support this strategy, it is important to understand customer concerns as early as possible to prevent churn: The customer action of canceling a subscription and moving to a new provider. In this paper, we use actual customer phone call data and develop the convolutional neural network (CNN)-based predictive model to detect churn signals from transcript data of phone calls. Experimental results show that when sufficient training data is provided with our text annotation method, our CNN-based predictive model generates state-of-the-art performance in churn prediction.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 21 citations exist upstream.