Objective

• The credit card division of the client bank wanted us to develop a metric to assess the value of its customers (i.e. Customer Lifetime Value or CLV) during the course of their relationship with the firm
• The individual-level transaction data provided by the bank, spanning January 2008 to August 2011, was used for estimating and testing – 6,706,920 transactions by 15,243 customers
  

CRISIL's solution

• Developed probability-based models to forecast the number of transactions using a Pareto/NBD model
• Developed probability based models to forecast average dollar value per transactions using a Gamma/Gamma model 
• Estimated values (CLV) helped to classify customers into different groups
• An open source software R was used for this data-intensive implementation
  

Client impact

• CLV modeling helped the client in answering the following three questions: 1. How frequently does the customer use the credit card; 2. How much do they spend on average; 3. and How much does the firm need to spend to keep them
• The CLV metrics helped the decision makers in focusing more on the attractive and profitable customer segments
• The CLV metrics also helped the customer-facing employee of the client to select the right customer and devise specific campaigns, promotions and discounts
• The model developed has been integrated into the client’s systems and is now a key driver of its sales and marketing strategies

Questions

Looking for high-end research and risk services? Reach out to us at: