I was flying home from a security conference in Boston on a Friday evening, enjoying an unexpected complimentary upgrade to business class, listening to Dungeon Crawler Carl, and reflecting on some of the conversations I had over the previous few days.
During one of those conversations, I had confidently stated that a YouTube channel I help support had accumulated more than 10 million views. The channel contains playlists of videos created by my team, but the videos themselves are owned and published by our parent organization.
At the time, the number sounded right. It was based on years of rough estimates, manual counting, and occasional spot checks. The more I thought about it, however, the less confident I became. The number wasn’t based on current data, and there was a good chance it was wrong. In fact, I was secretly hoping the real number might be even higher.
That realization led me to a challenge I had been considering for some time. Could I build and publish a useful public project entirely from an airplane using nothing more than free in-flight Wi-Fi, a laptop, and a collection of AI coding tools?
Several years ago, I had built a similar solution using Azure Logic Apps. It worked well enough for a while, but like many personal projects, it eventually fell into disrepair. Connections expired, APIs changed, and over time I lost the ability to easily validate the numbers.
So somewhere over the Midwest, I started vibe coding a replacement.
Part of the motivation was practical. I wanted a solution that was simple, portable, and independent of Azure. Something that anyone could run with nothing more than a YouTube API key. Another part was personal. I wanted to see if I could complete the initial version of a genuinely useful open-source project before the wheels touched the runway. You might call it my attempt at joining the vibe-coding mile-high club.
That challenge led to the creation of the YouTube Playlist Tracker, a Streamlit application and companion Python notebook that uses the YouTube API to collect playlist and video statistics.
When I built the original Logic Apps version years ago, one of the things I learned was that the YouTube API allows you to gather a surprising amount of information about public videos that you don’t own. Since my team’s videos are published under a different channel, that capability was essential. The application can retrieve playlist details, aggregate video statistics, identify top-performing content, and provide a much more accurate picture of overall engagement.
After running the analysis, I quickly learned that my long-standing estimate was significantly off.
Instead of 10 million views, the videos my team created had accumulated approximately 2.6 million views.
Still a number I’m proud of, but also a reminder of why accurate data matters. Assumptions have a way of becoming facts when they’re repeated often enough, and sometimes it’s worth taking the time to verify them.
The project is available on GitHub:
Repository: https://github.com/AndrewBlumhardt/youtube-playlist-tracker
The repository currently includes both a Streamlit dashboard and a Python notebook. The dashboard provides a simple interface for exploring playlist statistics, while the notebook offers a more detailed walkthrough of the code and API interactions for those interested in learning how it works.
This is only the beginning. I’d like to expand the project to support automatic playlist discovery, persistent playlist selections, historical trend tracking, and potentially a hosted or WordPress-friendly version that makes the solution accessible to a broader audience.
The most entertaining part of the project isn’t the code itself. It’s that the majority of the first working version was designed, built, tested, and published somewhere over the Midwest while flying at 35,000 feet using complimentary Wi-Fi.
Sometimes the best projects don’t start with a roadmap. They start with a question worth answering, a few uninterrupted hours, and just enough curiosity to see where the journey leads.