> ## Documentation Index
> Fetch the complete documentation index at: https://docs.jabrod.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Churn definition

> What counts as a customer leaving.

The model learns from customers who left in the past, so "left" has to be defined first. In a workspace's **Churn** step, answer **When has a customer left?** with one of three definitions:

| Choice | A customer has left when | Needs |
| - | - | - |
| **They cancel** | They have a cancellation date, or a status such as `cancelled` | A **Cancelled on** or **Status** column in the customers file |
| **They stop paying** | No payment has succeeded for a number of days (45 by default) | A payments file |
| **They go quiet** | They haven't used the product for a number of days (30 by default) | A usage events file |

You can change the number of days for the last two. Each choice shows how many of your customers it counts as having left, so you can compare them before training. Pick the definition your team already uses.

For **They cancel**, a customer whose status says they left but who has no cancellation date counts as leaving on their last activity.

## Prediction window

**Predict who leaves in the next** 14, 30, 60 or 90 days. The default is 30. A score is the chance that a customer leaves within that window.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.