Reverse-Engineering YouTube Success: How I Built a 90-Day Content Strategy in 1 Hour
You want to start a YouTube channel like Dan Koe or Alex Hormozi. You have expertise and ideas. But reality hits fast. What topics work? How long should videos be? What titles get clicks?
Most creators spend 6 to 12 months guessing. They post randomly, hope the algorithm notices them, and burn out before finding what works.
I took a different approach, focusing on data and AI, of course. I selected the influencer I follow and developed a framework for self-directed learning in a data-driven way. I analyzed Dan Koe’s 200 videos: 1.14 million subscribers and 40.8 million views, using Apify web-service for data extraction and Claude Code for pattern analysis and plan creation. Time spent: 1 hour at most. Cost: ≈$1.
Result: a comprehensive 90-day content strategy I'll share with you.
The Process
Step 1: Data extraction. I used Apify’s YouTube Channel Scrape – a pre-built tool that extracts public video data. You'll get:
Full text transcription of ALL videos!
Date Published
View Count (what were the most popular videos)
Likes (you'll learn what people like the most)
Comments (you'll see what people are asking in each content pillar; you will act as a middleman between the influencer and their fans).
Keywords (see how the influences targets his content)
Description
No coding required. I pasted Dan Koe’s channel URL, asked Claude Code to…
Create a file with all videos from Dan Koe’s YouTube Channel. One link per line, no numbers. Only one link per line: https://www.youtube.com/@DanKoeTalks/videosselected 145 videos, and ran the scraper with this content:
Cost: $1.015. Time: 3 minutes 24 seconds.
The output: a complete dataset with titles, view counts, likes, comments, duration, and publish dates.
Step 2: Pattern analysis. I uploaded the dataset to Claude Code with a simple prompt:
analyze this data and tell me what works. I asked for title patterns, optimal video length, best publishing months, and content calendar recommendations. Find the most popular subjects of videos and take into account comments left by people asking for something new or adding some useful information, and use it in your planning for my Report. Three minutes.
Now export the results in JSON format – the most suitable one for Claude Code and any LLM.
It will look like this. Don't worry it isn't for you, it is for The Robots :)
What the Data Revealed
Title formulas matter. Videos with numbers in titles averaged 302,214 views – 68% above baseline. “How To” format videos hit 285,122 views, up 58%. Top performer: “How To Get Ahead Of 99% Of People (In 6-12 Months)” with 1.66 million views. The formula: action verb, specific percentage, timeframe, benefit in parentheses.
Video length has a sweet spot. Ten to twenty-minute videos performed 91% better than average at 345,310 views per video. Videos under 10 minutes dropped 58% to just 75,618 views. Deeper content wins.
Seasonality drives performance. November videos gained 104% more views than average. October up 59%, July up 43%. September and January dropped 30% or more. Launch timing matters.
Topics cluster predictably. “Life” appeared 37 times in titles, “work” 15 times, “change” 12 times. The audience responds to transformation promises, productivity systems, and success frameworks.
Building the Strategy
My 90-day plan focuses on life transformation and boosting productivity – the proven keys to success. Month one: 8 videos, each 15 to 20 minutes long, using the number-plus-timeframe title formula. Month two: 10 videos incorporating business and success frameworks. Month three: 12 videos featuring full topic rotation and seasonal optimization.
Conservative projection: 50,000 views per video within 90 days based on Dan Koe’s bottom 25% performers. Median scenario: 100,000+ views per video by month three.
The Bottom Line
Total investment: one hour and $1, sure you must have a local Claude Code with a plan. I have Max. Compare that to hiring an agency for $5,000 to $10,000 or spending six months in trial-and-error. Most creators guess, fail, adjust, and repeat until they quit.
I skipped that. I have data showing which titles work, which lengths perform, when to publish, and which topics resonate. This isn’t magic. It’s pattern recognition at scale.
Choose your influencer or multiple influencers – and combine the data into a single report. Follow this process. Create your plan. Carry it out. The question isn’t whether this works; it’s whether you will use it.
Next steps:
Pick a successful creator in your niche
Sign up for Apify and run the YouTube Channel Scraper
Install Claude Code and get used to it
Analyze the dataset with Claude Code
Build your 90-day content calendar
Start publishing
You can find the full breakdown of the Dan Koe analysis, including specific title templates and month-by-month content recommendations, in my Substack. Subscribe for more practical guides on building content systems that work.