I Let AI Grade 148 of My YouTube Videos (Here's What I'm Doing Wrong)

I exported 148 of my long-form videos and let an AI grade every one. Hit or flop. I wanted to know what actually separates the two. The results humbled me. I was wrong about almost everything.

Here's what the data showed:

- Video length was the #1 predictor by far. My hits run a median of 8 minutes, my flops 3.1. Longer wins.
- Publish time, day of week, question marks, exclamation marks, putting the year in the title? None of it mattered. All noise.
- What did matter: longer descriptions, first-person "I" titles, and a specific number in the title.

It trained 6 models in about 14 seconds, hit 88.5% accuracy, and gave me the full Python code behind every prediction. So I could see exactly what it was doing. And it all ran locally on my machine. My channel data never left my computer.

I used MLJAR Studio for the analysis. It's a desktop tool built for non-coders. Drag in a CSV, pick what to predict, hit go. The real Python is right there if you want it. There's a free trial if you want to run it on your own data:
https://mljar.com/studio

MLJAR sponsored this video. Every number you see is real, from my actual channel. I don't recommend tools I wouldn't use myself.

Follow on Twitter: https://twitter.com/jamespelton18
Join the 6 Million Dollar Apps community: https://www.skool.com/6-million-dollar-app-2514/about
Want my team to build your app? https://passive-apps.com

Faith + Money: "In all toil there is profit, but mere talk tends only to poverty." Proverbs 14:23 (ESV)