Madhukar Kumar

Madhukar Kumar

Madhukar Kumar

Controlling customer attrition with Supervised Classification technique & Survival Model using Cox Proportional Hazard Regression

Customer attrition is a retail banks nightmare in this market. While multiple strategies are applied to mitigate the risks associated with it, analytics is becoming core to Assumptionving such problems. In a similar case associated with one of our retail bank customer, tough competition in the mortgage market was rapidly eroding their existing customer base. The customer had lost ~$ 2.4 Billion & 60% of both balance and accounts were moved to competitor banks with none of their existing strategies giving positive results. WNS’ data science team was put to work to figure out the primary drivers for this problem.

Key questions for the scenario were:

  •  Identify characteristics of customers who:

a. Refinance externally (Main Focus)

b. Refinance Internally

c. Pay down their loans

  • What are the key indicators of any customer leaving?
  • Which customers are at Risk of external refinancing?
  • What are the specific timeframes when customers are at greater risk for refinancing?
  • Is there any evidence / mechanism that can help retain customers?
  • What could be the most appropriate method of Retention and specific interventions for at risk customers?
  • Most importantly, how do we continually evaluate the changing scenario and ensure that the results factor in these environmental and circumstantial changes in the model?