Work out churn prediction value instantly with clear inputs, formula shown and shareable results.
The value of a churn model is not its AUC but the customers a retention action actually saves. That chain has three multipliers: how many churn, what share the model catches (recall), and what share of contacted churners the intervention rescues — typically only 20 to 30 percent. Multiplying through usually reveals a much smaller prize than the model's accuracy metrics suggest.
Retention value
churners = customers x churn rate; identified = churners x recall; saved = identified x save rate (25%); value = saved x value per retained customer
Most flagged customers have already decided, and outreach converts a minority. Measure your own save rate with a randomised holdout rather than assuming the intervention always works.
Only up to the point where precision keeps the contact cost sensible. Contacting the whole base gives 100 percent recall and destroys the economics.