Churn Model To Identify Attrition Risk Guide Interventions And Save USD 100MM

Business Objective

Our client is a leading financial institution, offering banking, wealth management, and brokerage services to its customers.

Their most valuable customers have a larger contribution to the client’s business, so it is very important for the client to actively manage attrition, and prevent the top customers from leaving.

  • Identify the customers who are likely to leave within the next 30-45 days with special attention toward the valuable customers – identify the possible reasons for their attrition
  • Although there was a scoring mechanism in place, the outputs were not satisfactory. So our challenge was to significantly improve the mechanism
  • In this specific case, attrition was a rare event with monthly attrition numbers as low as <0.01%
  • Deliver the analysis to the sales team in such a way that they can take specific actions for the identified customers.

Solution Methodology
  • Linked the life events in a consumer’s life with attrition, and collated data pointing to those events from internal and external data sources
  • Accommodated all the past transactions’ history and the point of contact information to create enhanced features predicting attrition
  • Developed a comprehensive sampling framework to accommodate the need to predict attrition for the next 30-day window
  • Used various text analytics algorithms to convert email, phone, and chat communications into features that capture early warning flags
  • Developed an ensemble of models alongside business context and filters to identify attrition risk
  • Developed an actionable plan that the sales reps could use to proactively manage attrition risk.
Business Impact
  • Increased accuracy of existing churn model by more than 50%.

  • Identified the at-risk valuable customers contributing more than USD 500MM to the client

  • Potential savings worth USD 100MM identified by engaging 20% of the at-risk segment with minimal effort.

Copyright © 2023 Tiger Analytics | All Rights Reserved