YouTube Trend Hunter

Problem

Finding breakout YouTube niches before they saturate requires analysing thousands of videos' view velocity, comment sentiment, and topic clustering — a task too slow for manual research and too noisy for simple keyword alerts.

Results

Fully automated since June 2025. Delivers daily trend reports to email via GitHub Actions. Successfully flagged 3 niche breakouts 48-72 hours before mainstream creator coverage.

Challenges

• Staying within YouTube Data API v3 quota limits (10,000 units/day) while fetching enough data for statistically meaningful trend signals.
• Distinguishing genuine trend breakouts from evergreen content that simply resurfaces — required normalising by channel age and historical velocity.
• Formatting a dense analytics report into a scannable daily email without overwhelming the reader.

Architecture

A Python pipeline queries the YouTube Data API v3 for trending videos by category and region. Videos are clustered by topic using keyword extraction, then scored by view-velocity (views-per-hour since publish) and comment sentiment (VADER). A daily summary report is generated and emailed automatically via GitHub Actions — zero infrastructure cost.