Know the Value of your Customer

In today’s world we are using Data Science to solve different problem for different types of business and helping them to make data driven decision. Whenever a problem statement comes I have always realized the business comes with need of the end customer in back of their mind so no matter what problem we are solving using machine learning be it supply chain, pricing, manufacturing etc the objective is directly or indirectly related to the customer. That’s why I always recommend that make data driven decision but put customer in the heart of all your decision making, as I am interacting with multiple people I am realizing that some business does not have a framework to manager their customers life cycle and that’s why today I thought to share basics around customer life time value framework using machine learning for any type of business. In today’s world of startup many companies are trying to find out that how they can predict the value of a customer over the course of their interactions with their business. Let me first start with defining the customer life time value (i.e. CLV).

Let us start with the text book definition:-

“Customer lifetime value is the dollar value of a customer relationship based on the present value of the projected future cash flows from the relationship.”

In simple terms Customer lifetime value is a measure of customer profitability over time. Profit here consist both costs and revenue estimates, as both metrics are very important in estimating customer lifetime value though usually we focus on the revenue side. The reason for this is that revenue is more difficult to forecast than cost so a model is more necessary to predict it and also if you know the revenue form a customer then using that business will also know that how such spend will be there on that customer. So we can say that Customer lifetime value is calculated as a single dollar number where it  summarizes total revenue and costs related to a customer over time, it provides a net profit/loss summary of the customer’s total relationship with the firm,.

Most of the time It is calculated on per customer basis however sometimes it can be done for a specific segmentation in case we have already done the customer segmentation using different parameters.

When we say revenue then we also need to consider that the customer can generate revenue in different ways, of course direct purchases certainly increases the lifetime value of customer but in today’s world indirect marketing from that customers also add value to the customer, for example customer sharing good thing about the company on social media makes him more valuable so we need to remember to include these networking factors too in the framework however here I am going to focus on the direct interactions.

There are many ways to come up with CLV framework but mainly we can divide in two  i.e. Historical Customer Lifetime Value and Predictive Customer Lifetime Value. In Historical Customer Lifetime Value  we look past transaction to come up with the value of customers without predicting what those customers will do next. This works in the cases where customers behave similarly and have been interacting with the company for the same amount of time but has many drawbacks. Whereas the objective of predictive customer lifetime value is to model the purchasing behavior of customers in order to come up with what their future actions will be.

Purchase opportunities are different for different business context and based on the same context and opportunities the model needs to be developed to predict CLV. Let us first look at the business context where the business is contractual such as club membership, subscription based services, insurance policies, streaming services, telecom services, credit cards etc, in these business contexts either the purchase opportunity will be discrete or continuous. The same way there can be business with non-contractual settings such as online/offline grocery purchase, hotel stay, medical appointment etc and here also again the customer will have both the opportunity of purchase in different cases.

Machine learning, Markov Model and probabilistic models are different approaches to customer lifetime modeling however the algorithm needs to be customized as per business context as explained above. In this post I am not going in the details of the algorithm in different business situation as the idea of this post is to give basics around CLV modeling and in case you are interested to know/learn more about the algorithms in your particular business situation then feel free to reach out.

At the end I will just list out few benefits mainly from marketing angle of having CLV model in place for your business:-

·      Measure and demonstrate the bottom-line financial impact of different marketing activities.

·      Clearly align marketing programs with financial objectives and targets.

·      Focus on marketing from an marketing ROI perspective by determining the optimal balance between acquisition, share-of-customer, and enhanced loyalty objectives, increasing customer profitability over time

·      Scenario test a range of possible strategic marketing directions,

·      Determine the impact of internal marketing programs, as well as competitive and environmental factors on long-term customer profitability,

·      The stress testing of various marketing goals and environmental impacts,

·      Balance the competing needs of short-term profitability and longer term goals,

·      Understand the bottom-line profit contribution of different customer segments,

·      Demonstrate a long-term financial return for a range of marketing investments

At the end I would say that by having different predictive model in place to manage your customer’s life cycle will result in enhanced profitability of your business. So as I started use data science to solve your problems and make data driven decision by putting customer in the heart of all your decision making, specially value your customers and you will value when you will know their value..:-)