Inspiration
Customer churn is one of the biggest challenges for subscription and telecom companies. Teams often discover churn too late, after the customer has already disengaged. We wanted to build a tool that helps businesses predict churn early and take action immediately, turning reactive retention into proactive retention.
What it does
RetainIQ is a customer retention intelligence platform that predicts which customers are likely to churn. It analyzes behavioral data such as spending patterns, login frequency, support tickets, and engagement trends to generate real-time churn risk scores. The platform also maps customers across lifecycle stages and automatically drafts personalized outreach emails to help teams intervene before customers leave.
How we built it
We built the frontend using React, Framer Motion, and Recharts to create an interactive dashboard with animated data visualizations and customer journey insights.
The backend was built with Python and Flask, which serves customer data and churn predictions. We used scikit-learn, Pandas, and NumPy to train a logistic regression model that calculates churn probabilities.
The system combines predictive scoring with a template-based email generator to produce personalized retention outreach in real time.
Challenges we ran into
One of our biggest challenges was building a smooth, interactive UI while working under a 24-hour time constraint. Integrating multiple data views, like churn dashboards, lifecycle journeys, and alert systems and required careful state management.
Another challenge was designing a churn prediction system that was simple enough to implement quickly but still meaningful enough to demonstrate real value.
Accomplishments that we're proud of
We’re proud that we built a fully functional end-to-end platform in just 24 hours.
The project includes a machine learning churn model, a live analytics dashboard, an interactive customer lifecycle visualization, and a personalized email generation system.
Most importantly, the product demonstrates a complete retention workflow, from prediction to action.
What we learned
This project taught us how to rapidly prototype a full-stack machine learning product under hackathon constraints.
We learned a lot about integrating data science models with a user-friendly frontend and how important visualization and storytelling are when presenting predictive insights.
What's next for Thunderhacks 2026 (Gold Sponsor Challenge)
Next, we want to improve the platform by integrating real customer data sources such as CRM systems and usage analytics platforms.
We would also expand the machine learning model to include more advanced features and time-series behavior tracking to improve churn prediction accuracy.
Finally, we want to build automated outreach workflows and CRM integrations so teams can act on churn insights directly from the platform.
Note: Due to deployment constraints during the hackathon, the backend service is currently running locally. The full backend code and instructions to run it locally are available in the repository.
Log in or sign up for Devpost to join the conversation.