Verizon deploys generative AI to analyze 170 million annual calls and prevent more than 100,000 churn cases

Background

Telecommunications providers face persistent churn pressure. Verizon's 170 million annual calls represent both a cost challenge and a retention opportunity: dissatisfied callers who reach poorly matched agents are more likely to defect. Traditional IVR and skill-based routing systems had limited ability to predict the nature of a call before agent assignment. In 2024, Verizon moved to generative AI to close that gap.

What Was Implemented

• Generative AI system deployed to analyze and predict call intent for all inbound calls with 80% accuracy
• Intelligent call routing directing customers to the best-matched agent among 60,000 call center staff
• In-store AI for personalizing offers during 70 million annual store visits, targeting a 7-minute reduction in visit time
• All models and data processed within Verizon's own network (no external cloud exposure) using 1,500 customer data points per telephone number

Results

The AI-driven routing and intent-prediction system is reported to have prevented more than 100,000 potential churn cases in 2024. Call intent is predicted with 80% accuracy , enabling better agent matching at scale. In-store AI deployment targets a 7-minute reduction in average visit time across 70 million annual visits — though verified outcomes for this initiative were not available in sources fetched.

Lessons

• Predicting call intent before routing — rather than after — allows a large contact center to match customers to specialists more effectively at scale
• Data-residency controls (on-premise LLM processing) can be a practical enabler for enterprises that hold sensitive customer data at the required regulatory risk level
• Linking quality routing to churn-reduction outcomes (not just handle-time metrics) reframes contact center investment as a revenue-retention lever