How it works
The mechanism, in 4 layers.
Signal ingestion
Reads from Stripe (payment cadence), GHL (engagement metrics), support tickets (sentiment), product analytics (usage). Updates customer health score hourly.
Pattern matching
Trained on historical churn signals — the patterns that preceded past cancellations. Surfaces customers tracking those patterns before they actually cancel.
Expansion detection
Identifies customers using the product more than their tier supports, asking advanced questions, or signaling expansion intent in conversations. Routes to expansion team.
Intervention triggers
On health score drop ≥ 20 points in 7 days: route to Churn Saver Agent or human CSM with full context.