Overview
Customer churn is simply defined to be the percentage of customers in any period that unsubscribe or in some other way stop being a customer. A simple way to measure churn is the following
{% Churn = \frac{\# \; Lost \; Customers}{\# \; Customers} %}
This simple formula hides a lot of the complexity of customer churn. For example, it fails to take into
account any meausure of how long the customer has been a customer to date. For example, were the customers who
left new or long time customers.
One way to account for the age of the accounts that exited is to collect the data into a dtaset, and then to plot the accounts in a histogram or some other measure of density based on account age.
Modeling Churn
A common way to model customer churn is to model length of customer loyalty instead. That is, it is to find a distribution that represents the length of customer loyalty. This is typically done using the tools of Survival Analysis.