How Dialog Axiata used Amazon SageMaker to scale ML models in production with AI Factory and reduced customer churn within 3 months

This article discusses how Dialog Axiata, a leading telecommunications provider in Sri Lanka, used Amazon SageMaker and an AI Factory framework to develop a machine learning solution to reduce customer churn. The solution involved training two models: a base model using CatBoost and an ensemble model combining multiple algorithms.

Specifically, the article covers:

• Dialog Axiata's challenges with high customer churn rates in Sri Lanka's competitive telecom market
• The solution architecture involving separate training and inference pipelines for the base and ensemble models
• The methodology of using nearly 100 features across various data sources, dual model strategy, and sharing insights with business units for targeted retention campaigns
• Details on Dialog Axiata's AI Factory framework built on SageMaker for ML workload automation, experiment tracking, and cost optimization
• The MLOps process leveraging SageMaker components like Feature Store and Pipelines
• Significant business outcomes achieved, including a substantial reduction in month-over-month gross churn rates within 5 months