For a leading Internet Service Provider (ISP), Midtown Consulting Group (MCG) developed a predictive model in forty five days that segmented over 30% of the subscriber base into 13 high churning behavioral clusters.



Create a model to identify the most predictive attributes of churn, then score and segment the customer base to enable prioritization of customer retention efforts. A “common data base” of customer attributes had to be created by extracting from multiple databases: Customer Information System, Help Desk, Accounts Payable, Marketing and an outside demographics database to optimize the predictive modeling. Using the results of the predictive model, formulate retention strategies for each behavioral cluster to enable call center reps to be proactive versus reactive.

1. Developed predictive modeling capability to identify; (a) likely churners within dial-up customer base, (b) which marketing and sales promotions were ineffective, (c) which subscribers should be upgraded to broadband, and (d) which customers should be avoided in their acquisition strategy
2. Designed and implemented an automated process for assigning each customer a churn score and to a predictive cluster as data is refreshed – providing the client with a retention tool for ongoing use
3. Developed a customer database that would be refreshed monthly and enable the predictive model to be continuously updated
4. Utilized Business Objects software to report churn information throughout the firm as well as enabling customer churn scores to be utilized in call center activity

Eliminated ineffective, premium marketing and promotional offers. Transitioned the call center from being reactive to being proactive with respect to customer retention and the client can now focus its limited resources around the identified high churn, high impact behavioral clusters. Instituted an in-house tool that they now leverage and manage on their own without external resources.