Background
Lebanon's largest mobile operator — operating under anonymity per its data-sharing agreement with Berix — manages a subscriber base of approximately 2.1 million active SIMs across prepaid and postpaid segments. Since 2020, the operator has faced an unprecedented challenge: subscriber churn driven not by competitive switching but by economic distress. Subscribers were disconnecting not because they preferred a rival network, but because they could no longer afford to maintain their plans.
Traditional churn models — built on usage patterns, plan tenure, and customer service interactions — were failing to predict this economically-driven churn. The operator needed a new signal layer that could detect financial stress before it translated into disconnection.
The Challenge
- Standard churn models showed only 52% accuracy on economically-driven disconnections.
- Postpaid collections had deteriorated significantly, with 90-day delinquency rates tripling since 2021.
- No internal data signal reliably distinguished "about to churn by choice" from "about to churn by necessity."
- Retention interventions were being wasted on low-risk subscribers while high-risk subscribers went uncontacted.
The Berix Solution
The operator integrated Berix neighborhood economic stress indices and employment sector stability scores into its existing churn prediction pipeline. Berix signals were matched to subscribers at the district level, providing a macro-economic context layer that the operator's internal data could not supply.
The enriched model incorporated 9 Berix variables alongside 23 internal operator signals. The key differentiator was Berix's 60-day leading economic stress indicator — a composite signal that had demonstrated predictive validity for household financial distress in prior Berix research.
Results
Key Learnings
The telecom engagement reinforced a pattern Berix has observed across multiple client engagements: in crisis economies, macro-economic context is a stronger predictor of individual behavior than individual behavioral history. A subscriber's neighborhood economic trajectory tells you more about their likely churn risk than their own usage patterns.
The operator has since expanded the Berix integration to its credit scoring model for postpaid plan approvals, using the same economic stress signals to assess new subscriber risk at the point of acquisition.
“We were flying blind on economically-driven churn. Berix gave us the visibility we needed to intervene before subscribers disconnected.”
— Head of Customer Analytics, Lebanese Mobile Operator (anonymized)