← Back to Projects
Customer Health & Churn Risk Dashboard
Predictive risk scoring and behavioral triggers for proactive customer retention.
0-100
Risk Score Range
94% Precision
Churn Probability
12K
At-Risk Customers
Automated
Recommended Actions
The Challenge
Which customers are at risk of leaving and why? Where should we intervene? Company lost $4.2M annually to churn without predictive visibility.
Technical & Strategic Approach
- Engineered 150+ behavioral features from engagement, transactional, and product usage data
- Trained ensemble model (XGBoost + Neural Network) achieving 94% precision
- Built risk score (0-100) with actionable engagement trend signals
- Created NPS/sentiment correlation to identify root causes of churn
- Automated daily scoring pipeline feeding interventions to retention team
Results & Outcomes
Identified 12K at-risk customers before churn. Proactive outreach achieved 67% retention. Prevented $4.2M annual revenue loss. Retention team shifted from reactive to predictive.
Tech Stack
Python
XGBoost
TensorFlow
SQL
Looker
Airflow
Deep Dive into the Data
Explore interactive dashboards, detailed analysis, and data visualizations
View Full Case Study →