The challenge
The client, a telecommunications provider, only found out a customer was churning after the cancellation request came in. The retention team had no way to prioritize outreach — every customer was treated the same, and retention offers were sent out broadly instead of targeted at the accounts actually at risk.
Without any signal for why a customer might be about to leave, retention conversations were generic, and marketing spend on save offers wasn't reaching the people who needed them most.
The solution
Vizontra built a churn prediction model trained on usage patterns, billing history, support ticket volume, and contract data, scoring every active customer with a churn-risk probability on a regular refresh cycle.
An explainability layer surfaces the top factors driving each customer's individual risk score — so the retention team knows what to address in the conversation, not just who to call. Risk scores and their drivers are surfaced directly in a Power BI dashboard the team checks daily.
Results
- Every active customer scored with a churn-risk probability, refreshed on a regular cycle
- Retention team prioritizes outreach by risk score instead of treating every customer the same
- Top churn-driving factors surfaced per customer, so offers can be targeted, not generic
- Model validated against historical churn outcomes before rollout
- Churn-risk view built directly into a Power BI dashboard the retention team checks daily
- Retention campaign targeting shifted from broad discounts to identified at-risk segments