The Incremental Value of Unstructured Data in Predicting Customer Churn

Evert de Haan and Elena Menichelli, 2020, 20-105

Firms collect a lot of unstructured written data such as responses to open-ended survey questions, emails, and online comments. Most research, however, focuses on methods to analyze structured data and how resulting insights may be related to customer-, firm-, or market-level outcomes.

This study takes a different approach, asking whether this qualitative data provides insights into the customer database and can be included in econometric models. Evert de Haan and Elena Menichelli investigate if there is value in collecting, storing, and analyzing unstructured data to improve predictions of customer churn.

They analyze data from 50,000 customers from a European telecom provider encompassing: (1) traditional customer database variables like service usage and customer demographics, (2) structured survey information (i.e., the Net Promoter Score), and (3) written customer feedback. The authors extract new structured information, namely the sentiment and the topic of the feedback, from the unstructured written customer feedback, and combine these data with the customer database variables and the Net Promoter Score.

Using a series of Cox proportional hazard models, they show that the extracted information from unstructured data provides incrementally better predictions for churn; use of these additional data results in a potential annual saving of $315,000 for the data provider.

Put into Practice

Firms have a great deal of unstructured data, and this research offers insights into the monetary and managerial value of collecting, storing, and analyzing data for CRM purposes. The authors show how to extract structured information from unstructured data and include in econometric models to improve estimations of churn and retention. 



  • Corporate: FREE
  • Academic: FREE
  • Subscribers: FREE
  • Public: $18.00



Employees of MSI Member Companies enjoy the benefits of complete online access to content, member conferences and networking with the MSI community.



Qualified academics benefit from a relationship with MSI through access to, conferences and research opportunities.



The public is invited to enjoy partial access to content, a free e-newsletter, selected reports and more.




Learn more about becoming a member of the institute

Read More

Stay Informed

The MSI Mailing List

Subscribe to our email list to stay informed about upcoming events, news, etc.

Social Media