Customer Lifetime Value (“CLV”) is becoming a very mainstream data metric used to quickly gauge the value of an individual customer or group of customers. Because of this growing popularity, more businesses are seeking knowledge around the data metric and how it can be used to help them understand the profitability of their customer base.
This article will go through a high-level overview of CLV uses; however, actual CLV calculations will be avoided until later articles as they can become complex and difficult to understand. Instead, the goal of this article is to familiarize the reader with the conceptual types and uses of CLV. Let’s begin with discussing the difference in how businesses choose to utilize CLV analysis.
In CLV analysis, there are two ways to view the value of a customer: Historical and Predicted. Historical looks at a customer’s entire transaction history and shows the total value of that customer at some point in time. Predicted CLV looks at trends within customer behavior to estimate the expected value of a customer who shares similar characteristics with past customers. We will start with Historical CLV and walk through a simple business example to illustrate the importance of this metric.
Historical Customer Lifetime Value
“ABC” Corp. is in the business of selling data system software to large organizations. These data systems are sold as-is and typically have a lifespan of 10 years. The data system in question does not offer maintenance or upgrades and is sold on a non-contracted basis. ABC Corp. currently utilizes standard financial reporting where decision makers are checking the net profitability of the business as a whole. After several quarters of new customers, profitability growth continues to stagnate. What should ABC Corp. do to uncover the stalled growth in the profits of the business? The answer is Historical CLV analysis.
Historical CLV analysis for ABC Corp. begins with a list of customers organized by the business quarter in which they were acquired. The list should detail the total revenues and expenses relating to each of these customers with a corresponding net value. In reviewing the customer list, ABC Corp. uncovers that large expenses for most customers made them unprofitable, and it was only one very large customer that delivered profitability to the overall business. ABC Corp. has just conducted a Historical CLV analysis and concluded that the majority of customers they are acquiring are actually losing the business money. This level of detail is masked within normal financial reporting where all customer revenue and expenses are aggregated. By reviewing the Historical CLV analysis, it is easy to see that most customers are losing the business money and that adding more similar customers will not lead to profit growth. What should the business do? Simple. They should begin to analyze the expenses of these net loss customers to see what trends exist. Once identified, these trends are used to preemptively identify attributes of unprofitable customers and stay clear.
Now that we have a simple business case around Historical CLV, let’s look at Predicted.
Predicted Customer Lifetime Value
JMB Corp. is a major online retailer of t-shirts. These t-shirts are sold as-is and sold on a non-contracted basis. To view the available t-shirts, customers must set up an online membership account for free. JMB Corp. decision makers are looking at ways to increase the total membership signups in an effort to increase sales. The advertising division notifies the decision makers that an opportunity exists with an online website owner to advertise on a Cost per Acquisition (“CPA”) basis. Under the CPA agreement, JMB Corp. will pay the CPA fee for every new membership. How should the decision makers determine what CPA fee makes the membership profitable? They need to estimate, or predict, the value of a membership.
JMB Corp. reaches out to an old consultant who does database analytics and statistics for guidance on the issue. The Consultant informs JMB Corp. that he can perform an analysis to determine the expected value of a membership. He spares the decision makers the tedious details, other than informing them that it utilizes the latest and greatest probability theories generated by the most distinguished academics. After some data crunching, the Consultant informs them that for any membership signup the predicted value of that membership is $100.00. He goes on to share some more statistically relevant terms about margins for error, etc., but ends with this expected value of $100.00. To make it presentable to JMB Corp.’s decision makers, he writes it as E[Membership] = $100.00. Now what?
Armed with this powerful analysis, JMB Corp. knows that if they get a member, they are likely to generate $100.00 AFTER they are acquired. JMB Corp. knows that the $100.00 already takes into consideration the cost of the products sold and the time value associated with the sales, so the only remaining item is how much they can afford to pay to acquire this membership. Since JMB Corp. likes a hefty profit margin of 80%, they are willing to pay $20.00 to acquire this member ($100.00 – $20.00 = $80.00; $80.00 / $100.00 = 80% margin). JMB Corp. goes back to the website owner and offers a $20.00 CPA per member signup generated by the website.
JMB Corp. has just used a Predicted CLV concept to determine how much an acquired membership is worth in order to decide how much to pay for those memberships while maintaining their current profit margins. Now what? JMB Corp. needs to monitor this new program and see if the historical CLV is close to the predicted CLV at the same point in time. If it’s off, then they need to refine how the estimate is made. If it’s on, they need to keep utilizing these techniques and find new CPA opportunities.
ABC Corp. and JMB Corp. had very different problems, yet utilized the concept of CLV analysis to find a solution. The reader should see that although the concept of CLV seems simple, its complexity lies in the intended use of the data metric. To analyze current customer trends, Historical CLV might be the way to go. However, if a business needs to make customer acquisition cost estimates, the only choice is Predicted CLV. In the next article, we will focus entirely on calculating Historical CLV.