Data Science Nugget - 4

This is the fourth article in this series and will cover the application of Data Science to Customer Lifetime Value (CLTV).

Customer Lifetime Value is one of the most important and difficult metric to be accurately measured in a company. It indicates the reasonable revenue that a company can expect from a customer during lifetime of their business relationship. The metric considers a customer's revenue value and compares that number to the company's predicted customer lifespan. CLTV can be used to identify the customer segments that are valuable to the company and can be accorded preferential treatment in services like Relationship Manager, free locker and checking facilities.

The key benefits of CLTV management are

• Boosts revenue growth
• Customer loyalty and retention
• Helps identify the right customer segment for acquisition
• Reduces customer acquisition costs.
• Helps streamline the customer services/products to the targeted customer segment.

CLTV can be arrived at using a simple RFM (Recency, Frequency and Monetary) model. Another approach could be CLTV = summation (Value of purchases in a year) X average customer lifespan in years. CLTV is reported in the local currency.

While the above approach to CLTV looks quite simple, this is far from the truth. In case of an ecommerce company, the above will apply, however try doing the same for a Bank. The challenge is to define what is Customer Value. The measure of a Customer value in a Bank is quite complex and involves multiple aspects that define the customer revenue to the Bank. Typically Banks generate revenue from customers through the services that they offer as percentage commission or fee based income on wealth management advice, checking account, Bill discounting, Overdraft fees, ATM fees and interest on loans and other credits. Banks can also sell Mutual funds or Insurance products to its customer base and earn commission. It is important that most of these revenue interactions with the customers is captured at a granular level to get an accurate representation of CLTV.

Clustering and regression models are very commonly used to classify customers in terms of their CLTV. This helps the company device products and services suited to that segment at an appropriate cost. The CLTV itself is computed at an individual customer level. When a new customer is onboarded, he/she is mapped to an existing segment and the CLTV forecasted.

CLTV can be improved by improving the customer experience and engaging with the customer to increase the individual transaction value. The definition of preferred customer segments helps to keep the product/service mix relevant to the needs of the customer base.