Viral AI

Predict if your TikToks will go viral or not

Comment

Inspiration

Our team has a combined 12,000 followers on Instagram and is constantly uploading short-form content on TikTok and Instagram. We've always struggled with figuring out what makes a viral video. So we decided to solve this problem. Our tool allows creators to see not just what allows them to go viral, but also revenue potential and how they can improve their videos. By understanding how content performs, creators can make smarter decisions and maximize their earnings.

What it does

"Will I Go Viral" gives insights into key video elements—like the number of faces, audio quality, and revenue potential—to help creators optimize their content. Using a model trained after hundreds of TikToks, we analyze likes, comments, shares, captions, and more to see what really makes a video take off. More importantly, we estimate how much revenue a creator could earn based on virality predictions.

How we built it

  • Backend: Python (Flask) handles API requests and crunches social media data.
  • Frontend: Next.js for a sleek and easy-to-use interface.
  • Machine Learning: We trained our own model using Python + scikit learn to predict views from virality.
  • Infrastructure: AWS Lambda for scalable serverless functions.
  • Containerization: Docker for streamlined deployment.

Challenges we ran into

  • Data Processing Times: Mass-downloading TikToks took a significant amount of time.
  • View Prediction Complexity: Estimating views from engagement patterns required extensive modeling and feature engineering.
  • Deploying on AWS: We ran into multiple Python compatibility issues that delayed the time it took to deploy the model

Accomplishments that we're proud of

  • Successfully trained and deployed a model that predicts how many views a video might get.
  • Built an interactive dashboard for creators to track predicted factors and estimated revenue.
  • Showed creators the revenue potential they might gain based off of their videos

What we learned

  • Engagement is about way more than just views—shares and retention matter too!
  • Predicting revenue from virality requires deep data analysis and machine learning expertise.
  • Different content strategies can make or break virality.

What's next for Will I Go Viral

  • More Platforms: Expanding to Instagram and YouTube next!
  • More Accurate Revenue Predictions: Refining our ML model with more data points to improve earnings estimates.
  • Personalized Creator Tips: Recommending strategies based on successful videos.
  • Real-Time Trend Alerts: Faster insights so creators can hop on trends instantly.

Built With

Share this project:

Updates