This article explains what the prediction output from a customer churn prediction app contains.
The prediction output from the customer churn prediction app shows the user:
Their current active customers
What their usage/spend/other characteristics look like (e.g.: smallest amount credited during last 30 days, average number of customer service calls made in the last fortnight, number of MMS messages sent in the last 28 days, or other contributing factors from the “customer information dataset” specified during the app creation, such as the customer’s birthday)
What their predicted churn probability is at a given date. A high churn probability indicates a high likelihood for the customer to churn at the the churn date, and vice versa.
Example prediction output from the transactional customer churn prediction app
The definition of “active customers” varies based on the churn option i.e., the user’s business model:
For the transactional option, it is defined as any customer who had at least one transaction since the earliest of the start dates of the rolling date ranges chosen in the contributing factors section.
For the subscription option, it is defined as any customer who hasn’t terminated their subscription, which can be gathered from the latest “subscription start and end dates” dataset.