Customer retention using Survival model

August 16, 2020

Leading Life Insurance Company in Poland

Problem Statement

• Look out for an approach to retain policies and riders and to set up an early warning system to identify a potential attrition cases to prevent further revenue leakage
• Loss out on premium on riders which was lost due to rider and policies lapsation.
• Loss around 369 MN PLN in ANP due to lapsation

Analytics Led Approach

• After observing Attrition trend in the data over the years an approach was framed to provide a solution to contain further revenue leakage
• A Cox proportional hazard regression model was built to predict the propensity of policies to survive over a period of time
• Given are high level process steps which were followed
  • Data Exploration – Combining Policy; Customer; Prospect and Agent Data
  • Data processing and Variable Selection
  • Analytical Model Development
  • Model Output
  • Reports & Insights

Business Impact

• Identification of product affinities of segments of customers
• A product recommendation engine to provide new product recommendations (5 each) with their probability of conversion for all 3.5 million lapsed customers was designed
• Primary users, retention team were provided with insights for customer retention
• Integrated the attrition prediction with every customer touch point system through a web service and formulated a growth strategy to retain the existing customers as well as grow their business
• Model output was segmented to optimize the approach to follow the case and provided insights to treat each policy based on the segmented matrix Re-evaluate;Retain;Grow;Protect– for example, Retain is the quadrant high value and high risky policy, retaining such policies will optimize the effort as well as cost.

Critical Success Factors

• identified attrition and revival opportunity of worth 37 MN PLN for one particular month. 1% retention will result in 3.7 MN PLN recovery in a month.
• Provided most probable list of policies and riders to be retained on a monthly basis