I Let AI Analyze My Medium Stats…
…and Here’s What I Found
As someone who’s written hundreds of articles on Medium over the years, I’ve always been curious about what makes some posts more successful than others. But, I have to admit, I’ve been a bit hesitant to dive into the data. Maybe it was fear of what I might find, or maybe it was just a mental block. Whatever the reason, I’m glad I finally decided to take the plunge and let AI analyze my Medium stats.
The Setup
I decided to use a new agentic framework called Open Interpreter, which I activated in OS-mode with GPT-4o LLM. This gave the AI access to my computer, which, I know, sounds a bit scary. But, I was eager to see what insights I could gain from my data. I copied and pasted my Medium stats into ChatGPT, which processed them into a CSV file that I could feed into Open Interpreter.
The Analysis
I asked Open Interpreter to analyze my article titles for a range of factors, including:
- Word length
- Presence of numbers (e.g., “Top 5, 7 Steps”)
- “How to” format
- Emotional tone and sentiment
- Trends
- Frequency of keywords
- Views
Open Interpreter used a couple of machine learning models with Python and SciKit-Learn to analyze my data and identify patterns. And, what it found was fascinating.
The Results
Based on the analysis, Open Interpreter provided me with a range of title templates that were more likely to get likes and views. These included:
Comparative Analysis Titles
- Formula: “X vs. Y: Which is Better for [Specific Application]?”
- Template: “YOLOv10 vs. YOLOv9: Which is Better for Real-Time Object Detection?”
Comprehensive Guides
- Formula: “A Comprehensive Guide to [Topic] in [Year]”
- Template: “A Comprehensive Guide to Computer Vision in 2024”
Top Lists and Resources
- Formula: “Top X [Tools/Resources/Projects] for [Field] in [Year]”
- Template: “Top 10 Tools for AI Developers in 2024”
Beginner-Friendly Titles
- Formula: “Beginner’s Guide to [Topic]: How to Get Started with [Specific Technology]”
- Template: “Beginner’s Guide to Machine Learning: How to Get Started with TensorFlow”
Step-by-Step How-To Titles
- Formula: “How to [Achieve Something] Using [Technology/Method]”
- Template: “How to Build a Neural Network Using PyTorch”
The Takeaway
What I found most interesting about these results was that they challenged some of my assumptions about what makes a successful article title. For example, I had always thought that “how to” titles were a bit clichéd, but the data suggests that they’re actually very effective.
The Next Steps
My next step is to create a system where an AI can continuously test and learn from my data, as well as from other top-performing articles. This will allow me to refine my title templates and improve my read rate. And, eventually, I plan to apply this same approach to my LinkedIn articles and beyond.
The Problem
But, here’s the thing: creating an optimal business system with AI is not just about analyzing data and identifying patterns. It’s about creating a feedback loop that allows you to continuously test and refine your approach. And, that’s where many businesses struggle.
The Solution
That’s why I’m putting together an AI Business Systems Handbook, which will provide a step-by-step guide to creating an optimal business system with AI. And, the best part is, it will be available for free when it’s ready. So, if you’re interested in learning more about how to streamline your business with AI, be sure to join me on this journey.
My Journey Ahead
Over the coming weeks and months, I’ll be running various experiments and building my own optimal business systems with AI. And, I’ll be sharing my results and insights with you every step of the way. So, if you’re ready to take your business to the next level with AI, be sure to subscribe to my channel and join the conversation.
