Churn propensity modeling surfaces at-risk subscribers before they cancel — giving you time to intervene with targeted retention campaigns, personalized offers, or guided experiences. Each subscriber receives a continuously updated risk score based on their behavioral engagement signals, and for Recurly Subscription Management (RSM) merchants, billing and payment history as well.

Available to Recurly Engage customers

1Definition 2Key benefits 3Key details

PrerequisitesRecurly Engage must be enabled on your account. The Churn Propensity Score feature must be activated by your Recurly customer success team before it appears in the Engage console. Approximately two weeks of subscriber data is required for meaningful predictions; full model accuracy is reached after approximately 12 weeks.

LimitationsScore distributions may take time to stabilize for new accounts or accounts that have recently onboarded a large number of subscribers. It's normal for some accounts to have few or no subscribers in the 8–10 risk range — this reflects the model's confidence threshold, not a misconfiguration. Merchants using Recurly Engage without RSM have access to a behavioral engagement-only model; the enhanced model incorporating billing and payment history requires RSM.

Definition

Churn Propensity Modeling is a machine learning feature within Recurly Engage that assigns each subscriber a real-time risk score from 1 to 10 — where 1 indicates low churn risk and 10 indicates very high churn risk. The model trains on your historical subscriber data to learn the behavioral patterns associated with churn, and establishes merchant-level baselines to ensure scores are contextually meaningful for your platform. Scores update automatically on every new engagement event, or every five days in the absence of activity.

Integration type	Data used in model
Recurly Engage + RSM	Behavioral engagement signals + subscription billing and payment history
Recurly Engage only	Behavioral engagement signals only

Key benefits

Proactive retention Identify at-risk subscribers before they make a cancellation decision, enabling timely and relevant outreach.

Unified data model For RSM merchants, behavioral engagement signals are combined with billing data — including payment history, renewal trends, and pause and cancellation events — for a more complete picture of subscriber health.

Granular risk scoring The 1–10 scale gives you more flexibility than a binary flag, supporting tiered intervention strategies based on confidence level.

Flexible segmentation Combine churn scores with other subscriber attributes — such as geographic location, coupon usage history, or prior prompt interactions — to build highly targeted audiences.

Dynamic score updates Scores recalculate on every new engagement event and every five days of inactivity, keeping risk signals current without manual intervention.

Works across merchant types All Recurly Engage merchants can use this feature regardless of RSM status. RSM merchants benefit from a richer dataset that improves model accuracy over time.

Key details

How to use churn propensity modeling

Request feature enablement

Contact your Recurly customer success team to activate the Churn Propensity Score feature on your account. It won't appear in the Engage console until it's been enabled.

Allow the model to accumulate data

Once enabled, the model begins learning from incoming subscriber activity. Meaningful predictions emerge after approximately two weeks of data, with full model accuracy reached around 12 weeks.

NoteIf your account is brand new or has recently onboarded a large number of subscribers, expect score distributions to stabilize over time. It's also normal for some accounts to have few or no subscribers in the 8–10 range — this reflects the model's confidence threshold and is not a misconfiguration.

Create a segment using the churn score

In the Recurly Engage console, navigate to Segments and build an audience based on the Churn Score attribute.

Navigate to Segments in the Recurly Engage console.
Select Add new segment.
Under the Users section, locate the Churn Score attribute.
Use the slider to define the score range you want to target — for example, 7–10 for high-risk subscribers.
Optionally, add additional conditions to refine the audience, such as geographic filters, coupon usage history, or prior prompt interactions.
Save the segment.

Associate the segment with a prompt or campaign

Attach your saved segment to a prompt, guide, or retention campaign within Engage. Common use cases include surfacing a discount offer, initiating a cancel-save flow, or triggering a personalized re-engagement message for subscribers in a defined risk band.