Inspiration

What makes a podcast episode go viral? Lenny Rachitsky recently made the full transcripts of his podcast episodes public online (LinkedIn post below), and we took the opportunity to ~~use up all of our company's Claude API credits~~ analyze what makes an amazing podcast!

Our approach & tools:

Analyzed 285 episode transcripts using Claude's API to extract structured data about guests, topics, products mentioned, intro characteristics, and much more
Called YouTube's API to get episode details (title, description, published timestamp) and viewership metrics (views, likes, comments) for each episode
Mapped transcripts to YouTube videos to understand what drives engagement
Put the Hex notebook agent to work (it deserves a holiday after this 🌴)

Notes: Lenny's Podcast is also available on other podcast apps but we analyzed YouTube activity only. We only analyzed transcripts and metadata (not videos).

Our goal:

Find the secret formula for viral content. What makes certain episodes get 5-50x more views than others?

What we found:

The top 5-10% of episodes receive dramatically more views than the rest, often 10-20x the median
Certain patterns strongly differentiate top performering episodes: guest from super high-profile companies (Airbnb, OpenAI, Netflix), CEO/Founder guests, and specific title keywords
Product mentions evolved over time: ChatGPT dominated from 2023 onward, Cursor is the latest rocket ship, while Notion and Figma remain evergreen
Publishing strategy matters: Tuesday-Saturday posts significantly outperform Monday/Sunday
What surprised us most are the correlations we didn't find:
- Episode length shows zero correlation with views - 80-minute episodes perform identically to 60-minute ones
- Subscriber growth doesn't correlate with episode views - viral episodes don't necessarily grow the channel

We hope you enjoy exploring the data and find lessons to inspire your own analyses or content creation!