Churnor
Cherry AI and rule-based customer churn risk assessments
Ask Cherry to assess a customer using recorded integration evidence. Cherry is powered by Groq. Saved AI scores become primary, with automatic rule-based scores as fallback. Scores help prioritize review; they are not probabilities of cancellation.
Ask Cherry
In Customer Detail, ask Cherry to assess a customer or answer a question about recorded churn signals. Groq’s Qwen model receives minimized activity, billing, support, and integration evidence from that project. Cherry returns a risk score, an explanation, confidence, missing-data notes, and suggested actions for your team.
Cherry runs only when requested. Saved AI scores remain primary until another successful analysis and are labeled as older evidence after data changes or 24 hours. Insufficient evidence or a failed request preserves previous scores. AI and rule-based histories are separate.
Signals behind the rule-based fallback score
- Recency: time since the last recorded product activity.
- Frequency change: activity in the last 14 days compared with the preceding 14 days.
- Subscription risk: cancellation, failed payment, and downgrade events.
- Core feature usage: whether a previously used important feature has stopped appearing in recent activity.
- Recent activity: positive activity can offset some risk.
Billing deliveries are excluded from product activity counts. Risk depends on the data available for the customer, so missing telemetry does not establish the cause of disengagement.
Risk tiers
- Healthy: 0–40. Lower risk under the current rules.
- At Risk: 41–70. Signals warrant a closer review.
- High Risk: 71–100. Prioritize the account and inspect its evidence.
These tiers do not estimate how many days remain before a cancellation or guarantee that an account will stay.
When scores refresh
Supported event delivery and imports recalculate rule-based fallback assessments. Authorized dashboard and customer reads refresh missing or older assessments. There is no scheduled background inactivity scan while the application is unused.
Read the evidence before taking action
The customer detail view explains the saved assessment and shows recorded risk history. Recommended actions use that assessment to help your team decide what to investigate.