Telco Company

Challenges

A leading telecom company faced a growing challenge: customer churn. Every month, a significant percentage of users were leaving for competitors — often without giving any warning or feedback. By the time dissatisfaction became visible, it was too late to react. This not only impacted revenue but also disrupted long-term growth and customer loyalty strategies.

To respond, the company needed a smarter way to understand its customers — before they decided to leave.

The solution

The solution was a data-driven, AI-powered approach that brought together predictive analytics and generative AI to deliver daily, automated insights. At its core, the system used machine learning models built on Amazon SageMaker to predict which customers were at the highest risk of churning. These models analyzed a wide range of internal data—such as billing patterns, service usage, and CRM activity—and continuously learned from new interactions.

But the real breakthrough came from combining prediction with action. For every at-risk customer, the system generated tailored retention suggestions using Amazon Bedrock, which tapped into large language models to craft personalized conversation scenarios. These ready-to-use insights were made available directly to consultants through internal systems, empowering them to engage customers more effectively and at the right moment.

The entire process was automated end-to-end. Data pipelines built on AWS Glue handled daily data updates, ensuring accuracy and speed. The solution required no manual data review, no scripting of messages—just actionable outputs, updated in real time and delivered where they were needed.