What is Churn Prediction?

Churn prediction is the use of machine learning to estimate the probability that a customer will stop buying, cancel a subscription, or otherwise disengage within a defined time window. The output is a score per customer that downstream systems use to trigger retention actions.

Definition

A churn model is trained on historical data that includes both customers who stayed and those who left, labeled with relevant features such as recency, frequency, monetary value, support contacts, satisfaction signals, and product usage. The model learns patterns that precede churn and applies them to active customers. Scores are typically refreshed on a daily or weekly cadence.

Why it matters

Retaining an existing customer is consistently cheaper than acquiring a new one, so the economic case for predicting churn early is strong. Churn scores feed retention campaigns, loyalty programs, win-back emails, and customer service prioritization. Used well, they help brands intervene before the customer has already mentally left.

Operational use

A score by itself does nothing; it needs a system that can act on it. In a composable architecture, churn scores live in the CDP and are exposed to marketing automation, the storefront, and the service desk. Concrete actions include personalized offers, proactive outreach, or routing high-value at-risk customers to a human agent. Measuring whether the intervention actually reduced churn closes the feedback loop.

Related

Explore Agentic Frontend Management Platform · Personalization.