Churn Propensity Model
FirstHive now predicts which customers are likely to churn, and why, so retention teams can act before customers leave instead of after.
The problem
Brands could only see churn after it had happened. RFM churn cohorts and inactivity segments describe customers who have already lapsed. Retention teams in retail, CPG, travel, banking and insurance were acting too late, or relying on blunt rules like “no purchase in 60 days”. That rule treats a weekly grocery shopper and a once-a-year traveller the same way.
How it solves the problem
The model learns from data FirstHive already holds on each unified profile: transactions (what customers buy or pay, and how that’s changing against their own normal rhythm) and engagement (how they respond to messages, use the app or site, and interact with service and loyalty).
- Your definition of churn: admins define churn for their business without engineering help. Use inactivity for retail, CPG and travel, or an event such as a policy lapse, account closure or SIP cancellation for banking and insurance.
- A score for every customer: each profile gets a churn probability, a risk band, its top 3 reasons in plain language, a trend, and whether the score came from the ML model or rules.
- Ready from day one: a transparent rule-based score works immediately. A model trained on your own history through Eddie and ML Canvas takes over once it passes a quality check.
- Fresh daily: every eligible customer is rescored at least once a day.
- Use it everywhere: in Customer 360, Segmentation, Journeys and Campaigns (including a “risk band changed” trigger), and in exports.
- Prove the impact: a Churn Analytics report measures retention against a holdout group.
The score is for marketing and service prioritisation only. It must not be used for credit, underwriting, claims or pricing decisions.
Use cases
| Who | Use case |
|---|---|
| Retention manager | Get a daily list of customers likely to churn soon, with reasons, and start retention journeys before they lapse |
| Journey ops | Trigger a save journey the moment a customer moves into High risk, and branch by band |
| Relationship manager | Check a customer’s churn risk and drivers on the profile before a renewal call |
| Analyst | Audit accuracy and drift, and see whether retention spend actually saved customers |
| Ad and CRM teams | Export churn band for lookalike audiences or CRM prioritisation |