Client: Streaming Service with 10M+ Subscribers
Tools: AWS Glue, SageMaker, XGBoost, SHAP
Challenge:
Churn rate was increasing but the company had no actionable insight into which users were at risk and why.

Solution:

• Built ETL pipelines via AWS Glue to process behavioral and transactional data
• Trained XGBoost churn models, validated with AUC/ROC
• Used SHAP for model explainability, identifying key churn drivers
• Integrated with CRM to trigger proactive retention campaigns

Outcome:
Retention improved by 17%. Customer Success team leveraged insights for personalized outreach.