Inspiration
We wanted to solve a very real commercial problem: companies often realize a client is leaving only after the client has already stopped buying. Our goal was to turn sales data into an early-warning system that helps teams act before churn happens.
What it does
Hacktide is an intelligent commercial dashboard that detects clients at risk, predicts future purchases, and tells the sales team who to contact, why, and with what message. It includes an executive overview, client segmentation, product forecasts, client-level analysis, and a personal call queue for follow-up actions.
How we built it
We built a full-stack app with a React/Vite frontend, a FastAPI backend, and a SQLite analytical database. The data pipeline cleans and loads sales, product, client, potential, and campaign data, then generates alerts, risk scores, client segments, and six-month purchase forecasts. We combined SQL-based business rules with machine learning outputs to make the results explainable and actionable.
Challenges we ran into
The hardest part was making the analytics useful for non-technical users. We had to simplify concepts like ML segmentation, forecast variation, risk signals, and product weight into language that an executive or sales operator could understand quickly. We also ran into performance issues with large analytical views, so we optimized the backend with materialized snapshots and faster API endpoints.
Accomplishments that we're proud of
We are proud that Hacktide is not just a chart dashboard: it turns data into action. A user can go from a high-level risk matrix to a specific client, understand the reason for the alert, inspect product behavior, and send that client to a call queue in one click. We also built an end-to-end data pipeline, forecasting layer, and polished Spanish interface in a short hackathon timeframe.
What we learned
We learned that prediction alone is not enough. For a tool like this to be useful, the model needs to be explainable, fast, and integrated into the team’s workflow. We also learned the importance of designing for business users: the same data becomes much more valuable when it is presented as priorities, actions, and next steps.
What's next for Hacktide
Next, we would connect Hacktide to a real CRM so contact information and call outcomes are synchronized automatically. We also want to deploy the system with a production database, improve the forecasting models with more historical data, and add impact tracking to measure recovered revenue from each alert.
Log in or sign up for Devpost to join the conversation.