How to Calculate a Customer Life-Time Value (LTV) Without Writing Code

In this article, I will show you how to calculate the LTV in a no-code way using only the Datrics platform.

But first - what is LTV, and why is it important? The customer life-time value (LTV) is a prognosis of the net profit that the customer brings to the company over the whole cycle of their relationship.

The main reason to calculate it is to balance the cost of client acquisition and the company's profit from the customer. Of course, the ideal scenario is when the acquisition cost is lower than LTV. Still, sometimes it is well worth investing in acquiring the new client and then developing relationships to increase the profit over time. It is also important to understand which factors impact LTV to make data-driven decisions.

Here we'll take the dataset from an auto insurance company that contains the following information about the clients: the state, known historical LTV, type of coverage, education, gender, number of policies, the monthly premium, total claims amount, months since the last claim, vehicle class, and other columns.

The analytical pipeline in our case will look like this:

Let's go step-by-step and see how it works.

First, we do some automated data preparation - removing outliers from data and encoding categorical columns to digits (i.e., Employment Status may be only employed, unemployed, and other which is getting converted to 0, 1, and 2 for modeling). Again, this process is done automatically.

Next, we split the data into training and testing (90/10) to validate our created model. Now let's create the predictive model itself. To do that, we will use an AutoML brick. AutoML functionality in Datrics allows non-data scientists to make predictions using a simplified machine learning toolset. First, we choose the Customer Life-time Value column as the one we would like to predict.

And then, we evaluate the model on the testing data not yet seen by the model we preserved one step ago. One of the things we see as the output is the Feature Importance chart - the features in data that impact the LTV most.

As we can see, the number of policies the client purchased is the most important thing, and the next one is the monthly premium. The features such as type of coverage, income, gender, etc., are much less important.

We can also predict the LTV for new customers using the What-if functionality in the Predictive Model - just feed its API the parameters of a potential client:

If I change the number of policies for this person from 1 to 3 (the most important feature), the predicted LTV changes accordingly.

If interested in this kind of LTV analytics or exploring the other predictive and time-series analytical capabilities of the Datrics platform, please give us a shout, and let's talk.

Read more about our other updates and features and use-cases!

The demo is accessible here.