Bullseye

Problem

A major telecom super app needed an engagement feature to increase daily active usage. Their existing content feed had plateauing DAU metrics. They wanted a gamified prediction marketplace — users bet tokens on outcomes (sports, entertainment, current events) — that would create habitual daily check-ins and increase time spent in the app.

Solution

Bullseye is a token-based prediction marketplace designed for 100K+ DAU. Users receive daily tokens, place predictions on upcoming events, and earn rewards based on outcome accuracy. The system handles real-time odds calculation, outcome resolution, and token distribution with sub-250ms API response times.

Key Technical Decisions

Event-driven architecture — Prediction marketplaces have spiky load patterns: thousands of users placing bets in the final minutes before an event closes, then a burst of resolution calculations when the outcome is announced. An event-driven architecture with message queues handles these spikes gracefully without over-provisioning for average load.

Sub-250ms API response target — In a super app context, the prediction marketplace competes with other features for user attention. If placing a prediction takes more than a second, users switch to another tab. The 250ms target was set based on the client's analytics showing a 15% drop-off for every additional 200ms of latency.

Impact

• Designed for 100K+ DAU on telecom super app platform
• +20% DAU uplift after launch (measured over 30 days)
• Sub-250ms API response time for prediction placement
• Complete delivery in 3 weeks from spec to production
• Event-driven architecture handling spike loads during event closures