The operator has been using data mining techniques in predictive models for some time. These models are exploiting an analytic environment of more than 500 variables categorized as call details, contact data, subscriber info, handset info, payment data, usage trend, refill behavior and campaign responses. There are separate models for prepaid churn, postpaid churn and number portability. The telco wanted to increase the predictive accuracy and prevent more customers from churning. In postpaid churn prediction, the addition of the Idiro churn prediction model increased the number of churners accurately predicted by a massive 65% over the operator’s own churn model.

Join for free to read