Predictive Customer Lifetime Value

Published: 3. April 2016Author: Oliver Staubli, CEO & Data ScientistTags: Predictive Analytics,  E-Commerce,  Marketing,  Sales,  Retail,  Insurance,  Utilities,  Banking,  Telecommunications,  Examples,  Training

"Customer Lifetime Value" (CLV) describes the amount of profit a customer generates over his or her entire lifetime and is therefore a prediction of the net profit attributed to the entire future relationship with a customer. [1] CLV is probably the single most important metric for understanding your customers. CLV helps you make important business decisions about sales, marketing, product development, and customer support.

These crucial questions can be answered easily knowing the CLV: [2]

• Marketing: How much should I spend to acquire a customer?
• Product Development: How can I offer products and services tailored for my best customers?
• Customer Support: How much should I spend to service and retain a customer?
• Sales: What types of customers should sales reps spend the most time on trying to acquire?

The Advantages of Utilizing Customer Lifetime Value

It is easy to always focus on the present, but this is often not the best way to tap into the full potential value of each customer. For example, looking at conversion rates and first purchases while ignoring the long-term value of customers may lead marketers to invest resources in acquiring “cheap” customers with low total revenue value, instead of paying more to acquire customers which will continue to deliver a steady stream of income for years to come. Likewise, marketers and retention experts should focus resources on nurturing customer relationships with those customers who will continue to be a source of substantial revenue over the long term, while conserving resources which would be wasted on low-value customers.

An Simple Example [3]

A suburban golf pro shop determined that their customer base broke down quite naturally by geography. The pro shop determined that most of their customers came from two zip codes, one to the east and one to the west, and accordingly, the pro shop decided to calculate the CLV separately for each zip code. What they discovered was that the CLV for the east customers was 4 times the CLV for the west customers.

Once they dug into their stats, they found out how important this differentiation was. The east zip was 30% of their customer base but 80% of their profit. The west zip represented 70% of their customer base and 20% of their profit.

The pro shop had been spending their marketing dollars evenly east and west. Even though they knew the customers from the east were more valuable, they continued to put 50% towards the west because it seemed more effective. They were simply getting a much better “response” from the west marketing. Calculating the CLV for each segment made it obvious that the marketing dollars were not being allocated effectively.

Calculating Customer Lifetime Value

Calculating CLV requires accurate estimates of future events and is therefore very challenging. It is difficult to predict parameters such as how long a customer will remain engaged with a company and how much the customer will spend in each time period, especially when the customer is new. Further compounding the challenge is the fact that the data required to perform the calculations may be hidden deeply within multiple databases.

Despite the challenges, it is possible thanks to Predictive Analytics to predict CLV with high accuracy and thus to secure a huge advantage over the competition. At a high level, the models work as follows: [4]

1. Observe various individual-level buying patterns from the past – find the various customer stories in the data set.
2. Understand which patterns correspond with valuable customers and which patterns correspond with customers who are leaving for good.
3. As new customers join, match them to these patterns accordingly.

Are you interested in starting an Advanced Analytics project with us?

Get in touch