Customer Intelligence & Predictive Analytics Platform
End-to-end data warehouse with predictive ML models for customer churn, propensity scoring, and intelligent sales targeting
The Problem
The Solution
Architected a complete customer intelligence platform using AWS services and machine learning to drive revenue through predictive analytics. The system processes customer data across multiple product categories with automated daily workflows integrating CRM history, callback schedules, and customer lifecycle patterns.
Built medallion architecture data warehouse (bronze/silver/gold zones) with S3, Glue, and Redshift. Developed predictive ML models using SageMaker/Python with PySpark and XGBoost for churn prediction, purchase propensity, and win-back prioritisation. Created RFM-based scoring system combining recency, frequency, monetary value, conversion rates, and product diversity.
Tech Stack
Key Features
- Medallion architecture data warehouse (bronze/silver/gold zones)
- 90-day customer churn prediction model
- 30-day purchase propensity scoring
- Win-back prioritisation using ML
- Automated ETL pipelines with Glue DataBrew and Glue ETL jobs with daily refresh cycles
- RFM-based scoring system with normalised weighted scores
- Automated daily workflows integrating CRM contact history, callback schedules, and customer lifecycle patterns
Results & Impact
Retention & Reactivate campaigns (Sep 1 - Dec 23, 2025) generated £445k+ in directly attributed contracts across both campaigns