Audience Lens

✨ Inspiration

Scrolling through thousands of YouTube comments is a nightmare for creators—especially the ones with huge audiences. But those comments often contain valuable gold: feedback, future content ideas, or straight-up appreciation. I wanted to help creators actually hear their audience without wasting hours. That’s where Audience Lens came in—a tool that does the heavy lifting and shows only what really matters.

💡 What it does

Audience Lens takes a YouTube video URL and:

Fetches all comments using pagination (even if there are thousands)
Sends them to the Google Gemini API for deep analysis
Extracts:
- Sentiment breakdown (Positive, Neutral, Negative)
- Viewer suggestions & questions
- Criticism & common dislikes
- Key recurring themes and content signals
Displays everything in a simple dashboard so creators know exactly what their audience is saying—without reading every comment.

🛠️ How I built it

Frontend: React.js for the UI and routing
Backend: Node.js/Express server to interact with YouTube API
Comment Fetching: Used YouTube Data API v3 with custom pagination to go beyond the 100-comment limit
AI Analysis: Integrated Google Gemini API to extract sentiment, feedback themes, and viewer intent
Result Display: Dashboard with summary sections and visual categorization of comments

🧩 Challenges I ran into

YouTube API limits: Fetching all comments while managing quotas and pagination logic was tricky.
AI tuning: Getting consistent, relevant results from Gemini took prompt tweaking and output filtering.
Data noise: Had to clean up emoji-filled, slang-heavy, or low-effort comments before sending them for analysis.

🏆 Accomplishments that I'm proud of

Made the YouTube comment section actually useful for creators
Integrated AI in a meaningful way—not just for sentiment but full insight extraction
Handled large volumes of user-generated data smoothly and efficiently
Designed a clean UI that even non-technical creators can understand instantly