Sri Lanka’s Dialog Axiata provides telecommunications services to the majority of Sri Lanka, with over 17 million subscribers, which comprises 57% of the Sri Lankan mobile market. The company has many facets, including home broadband, payment platforms, and other fintech services. In a competitive market with high user churn rates, Dialog Axiata needed a way to predict the churn and proactively retain customers. Thus, they created the Home Broadband Churn Prediction Model using Amazon’s SageMaker.
Amazon SageMaker is the ideal platform for building and training machine learning models. (ML). Using SageMaker, Dialog Axiata developed two models for their system: a base model powered by CatBoost, a Gradient Boosting Decision Tree algorithm, and an ensemble model that merges multiple ML strategies. CatBoost is a powerful platform for ML and AI applications, designed to digitally and quickly analyze large amounts of data.
Dialogue Axiata intersects demographic, network usage, and network outage data from across their organization to forecast churn cases 45 days in advance. Their models employ a combination of nearly 100 features, resulting in a highly accurate and efficient process of predicting and mitigating customer churn. The base model trains on these features, with the ensemble model serving as a safety net to ensure no potential churn instances are missed.
Once the models identify customers likely to churn, Dialog Axiata takes action by targeting these customers with personalized retention campaigns. These campaigns may leverage tailored offers, incentives, or other unique forms of communication designed to directly address the distinct needs and concerns of at-risk customers.
Dialog Axiata’s AI Factory, equipped with the SageMaker platform, is central to their operation, ensuring a streamlined approach to AI/ML workloads and the potential to reduce the time and costs associated with these operations. The AI Factory framework also guarantees stringent data security and optimized costs, contributing to the effective management of AI initiatives.
Dialog Axiata’s successful implementation of this churn prediction solution has led to significant business outcomes within a short period, including a substantial reduction in churn rates. This triumph showcases the remarkable power of AI/ML technologies and data-driven strategies in maintaining competitive advantage in the telecommunications landscape.
In conclusion, Dialog Axiata’s pioneering strategy in leveraging powerful AI models using the AI Factory framework and SageMaker has resulted in compelling business outcomes. The company’s forward-thinking, data-driven approach to mitigate and predict customer churn exemplifies the immense potential AI/ML technologies offer in the competitive world of telecommunications.
