Churn Prediction in Direct Banking: Consumer Credit Customers at Risk

This case study article details how we managed to create a churn prediction solution for a Direct Banking company. Learn how we achieved to reduce the churn rate by 6% after 10 months, and by 15% only two years.

Key Challenges

A large German direct bank with several million customers had an elevated churn rate that was increasing over the years. More specifically, they had to deal with frequent early redemptions of consumer credits. That’s why they initiated the project outlined below to systematically:

• define and identify churners of consumer credits,
• understand their behavior,
• deduct early warning indicators,
• forecast them as precisely as possible in terms of time,
• derive suitable preventive CRM measures to
• reduce the churn rate significantly.

Therefore, the behavior of more than 450,000 consumer credit customers (who redeemed their loans early for various reasons) was analyzed over a period of more than twelve years.

Our Approach

• “Customer Analytics” Status quo analysis as a prerequisite for the project success
• Early conception and development of the customer win-back campaigns, in cooperation with the Digital Marketing department and the responsible campaign managers
• Profitability calculation of the win-back campaigns in the first year
• Analysis of customer (churn) behavior & customer-specific churn forecast
• Accompaniment of the preparation and implementation of the campaign as well as monitoring & success control in the first two subsequent years
• Transfer of the prototype into regular productive operation

Benefits

• Determination of key contract termination indicators for consumer credit customers
• The monthly updated, time-accurate termination forecast enabled customer-specific marketing measures to be taken for the first time to prevent terminations before they occur
• 6% reduction in churn rate (in pilot operation after 10 months)
• 15% reduction in churn rate (after two years of regular operation)
• Upselling effects: Identification of a customer group that even takes out a new consumer credit with a higher credit volume after having been contacted

Team Involved on this Project

Product Owner, Scrum-Master, Data Engineer, ML-Engineer, Data Scientist, Management Scientist, Digital Marketer and Campaign Manager.

Share

Thanks to our
Author

Peter Neckel