About

What Drives Successful Usage on Zerve?
Zerve AI Hackathon 2026 | by neehashiva44
Analyzed 409,287 real Zerve platform events from 4,774 users (Sep-Dec 2025) to answer: what behaviors predict long-term success?
Key Findings: Upgraded users generate 270x more events than casual users. 95.3% of users who hit credit limits never upgrade. Coder Agent usage in first 72 hours predicts upgrade. Germany leads all markets with 7% upgrade rate. Success Score framework identifies high-potential users.
What was built: User Journey Funnel, Credit Cliff Analysis, Random Forest ML Model (AUC 0.873), Early Warning System, Cohort Retention Heatmap, User Segmentation, Success Score Framework, Geographic Analysis, Growth Revenue Calculator and Live Streamlit Dashboard.
Tools: Python, Pandas, Scikit-learn, Matplotlib, Streamlit, Zerve AI Agent
Live App: https://zerve-user-success.hub.zerve.cloud