A Practical Guide to Building & Using It
What This Automation Does (At a High Level)
Once set up, the system:
- Fetches videos from selected YouTube channels
- Separates long-form and short-form content
- Analyzes performance signals (views, likes, growth)
- Breaks down why content worked (hooks, structure, CTA)
- Stores everything neatly in Google Sheets
- Avoids re-analyzing the same videos again
This turns YouTube research from guesswork into a repeatable system.
Step 1: Schedule the Automation
Purpose: Run research automatically, without manual effort.
- Use a Schedule Trigger
- Example setup:
- Frequency: Weekly
- Day: Sunday
- Time: 7:00 AM
Why this matters:
- Weekly cadence captures trends early
- Prevents over-analyzing daily noise
- Keeps your dataset consistent over time
Step 2: Define Which Channels to Analyze
Tool: Google Sheets (Channels list)
Create a simple sheet with:
- Channel name
- Channel URL
- Optional status field (Pending / Completed)
Why this matters:
- Sheets act as your control panel
- Easy to add/remove competitors
- Non-technical and scalable
Step 3: Loop Through Channels
Purpose: Analyze multiple creators in one run.
- Use a loop node to process channels one by one
- Each loop iteration:
- Pulls channel data
- Scrapes videos
- Runs analysis
Key insight:
- During development, loops only execute one item
- Once connected back, the full list runs automatically
Step 4: Scrape YouTube Content
Tool: YouTube Scraper Actor (API-based)
Configuration highlights:
- Max long-form videos (e.g., 40)
- Max shorts (e.g., 40)
- Sort by views
- Dynamic channel URL
Optional filters:
- Video badges (4K, live, etc.)
- Content types (video vs shorts)
Why this matters:
- Sorting by popularity ensures you analyze signals, not noise
- Limits reduce API costs and improve quality
Step 5: Separate Long-Form & Short-Form Content
Use filters to split:
- Long-form videos
- Shorts
Then:
- Sort each by view count (descending)
- Limit to top performers (e.g., top 20)
Why this matters:
- Shorts and long-form follow different psychology
- Mixing them leads to bad conclusions
Step 6: Analyze Long-Form Videos
What You Analyze
- Title
- Thumbnail
- Performance metrics
What You Skip (Intentionally)
- Full transcript (optional)
Why:
- Long transcripts are expensive
- Titles + thumbnails drive clicks
- Transcript analysis can be added later if needed
How It Works
- AI analyzes:
- Title positioning
- Thumbnail visual cues
- Messaging clarity
All outputs are saved into a structured object.
Step 7: Analyze Short-Form Videos (Deeper)
Short-form is where hooks matter most, so analysis goes deeper.
What Gets Extracted
- Transcript
- Thumbnail
- Views, likes, date
What AI Analyzes
- Hook
- Body
- CTA
- Why the hook worked
- Emotional and strategic positioning
Example insight:
- Contrarian hooks
- Curiosity gaps
- Clear stance early in the video
This is reusable intelligence, not just data.
Step 8: Merge All Analysis Cleanly
Use a merge node:
- Combine title analysis
- Combine image analysis
- Combine video stats
Use:
- Combine mode
- Merge by position
Why this matters:
- Creates one clean data object per video
- Avoids overwriting fields
- Makes Sheets updates reliable
Step 9: Save Everything to Google Sheets
Two databases:
- Long-form content
- Short-form content
Key rule:
- Use Append or Update
- Match by Video ID
Result:
- New videos get added
- Existing videos get updated
- No duplicates
- No manual cleanup
Step 10: Use Pinning During Development (Critical)
When testing:
- Run expensive nodes once
- Pin their output
- Iterate freely downstream
Why this matters:
- Saves API credits
- Speeds up development
- Prevents accidental re-scraping
This is one of the most overlooked productivity tricks.
How to Actually Use the Output (Important)
This automation is not just for “analysis”.
You can use the data to:
- Design better thumbnails
- Write higher-converting hooks
- Identify recurring winning angles
- Train future AI systems on proven patterns
- Spot content gaps competitors ignore
Advanced use:
- Feed this data into another AI that suggests video ideas
- Auto-generate thumbnail prompts
- Build a personal content strategy engine