short-form-video（短视频）

Short-Form Video Production

Purpose

This skill guides short-form video creation from concept through publication. It emphasizes practitioner methodology: fast prototyping, finding what works before building process, and "sponge then sharpen" creative philosophy.

Core Philosophy: You can't build a franchise system if you're not already selling a billion cheeseburgers. Find what works FIRST, then build process around it.

When to Use This Skill

• Creating original short-form video content (Reels, TikTok, Shorts)
• Cutting podcast clips for short-form distribution
• Developing video formats and testing new concepts
• Optimizing existing video content for better performance

For captions and on-screen text only: Use video-caption-creation skill instead.

The Sponge-Then-Sharpen Method

Phase 1: Sponge (Hunt for What Works)

Be malleable. Try different things. Stay loosey-goosey.

In this phase:

• Prototype fast (15 minutes max per video)
• Publish everything - "I don't have bad clips"
• Test multiple formats, hooks, styles
• Don't build systems yet
• Accept chaos as normal for creative work

Warning signs you're stuck:

• Working on one video for hours
• Building complex processes before proving format
• Perfectionism preventing publishing
• Over-investing in content that hasn't performed

Phase 2: Sharpen (Triple Down on Winners)

Once something works, lock in and scale.

In this phase:

• Build process around proven format
• "I'm not gonna sleep and I'm gonna make 15,000 of these"
• Optimize the working format
• Create templates and systems
• Delegate production

When to shift: You've found a format that consistently gets views AND you could make it repeatedly.

The 15-Minute Rule

If you're working on short-form content for longer than 15 minutes, that's a warning sign.

Short-form prototyping should be:

• Fast to create
• Cheap to fail
• Easy to iterate

Get it 90% good. The last 10% would take 2.5 hours - not worth it.

Mindset: You're a better editor than blank-page creator. Get the draft done, then evaluate.

Hook Construction

The McDonald's Test

If a truck driver wouldn't understand your hook, it's too complicated.

• Accessible language beats impressive vocabulary
• "Stop raising entitled kids" > "Addressing entitlement in childhood development"
• Wider net = better performance

The 3-Second Window

The first 1-3 seconds determine everything. Construct hooks that:

• Stop the scroll immediately
• Create curiosity or emotion
• Promise value without giving it away
• Pass the "would I stop for this?" test

Hook Categories That Work

1. Polarizing Statements

• "George Washington was a screw-up"
• "Your kid's Minecraft addiction is genius"
• Challenge common assumptions

2. Counter-Intuitive Reveals

• "The worst experiences are the best teachers"
• "Don't try to limit screen time"
• Flip expected wisdom

3. Direct Challenges

• "If your kid hates school, please watch this"
• "Never give up on the weird kid"
• Speak directly to specific audience

4. Curiosity Gaps

• "The second person theory"
• "His kids skipped school for 100 days. Here's what happened."
• Incomplete information that demands resolution

5. List Format

• "5 dyslexia myths" (add ding + on-screen number)
• "3 reasons comedians make better marketers"
• Promise structure and takeaways

Hook Optimization

Punchline First: Put the payoff at the beginning, not the end.

• "Don't hire a middle-aged man from Google - that's creepy. Hire me instead."
• Then explain how you got there

No Satisfying Ending? Add: "Check description for how she solved it"

Same Video, Different Hooks: Test the same content with different hooks - space them out over weeks.

Producible Video Formats

Tier 1: No Filming Required (Fastest)

Split-Screen with Oddly Satisfying Footage

• Top half: talking head or podcast clip
• Bottom half: laffy taffy machine, hot metal balls through foam, horse grooming ASMR
• Keeps lizard brain engaged while delivering content

Text on B-Roll

• Stock footage or screen recordings
• Text appears in succession
• Example: "Kids learn to walk at different ages..." [balls falling through foam] "...so why do we make every kid learn algebra at the same age?"

Greenscreen Commentary

• You pointing at/reacting to other content
• Co-signing valuable information
• Quick to produce once setup exists

Podcast Clips (Monologue)

• Single speaker with compelling insight
• Works best with well-spoken hosts in good visual environments
• Add color correction if needed

Tier 2: Light Production

Podcast Clips (Dialogue)

• Back-and-forth creates energy
• Interview format with host reactions
• More engaging but harder to cut cleanly

UGC-Style Talking Head

• You speaking directly to camera
• Casual, authentic feel
• Can use AI restyle for characters (Instagram Edits → Restyle)

Mashup/Supercut

• Multiple clips combined around theme
• "5 different moms on their best homeschool tip"
• Requires library of source material

Tier 3: Produced Content

On-Location Footage

• Your own B-roll with voiceover
• Requires shooting and editing time
• Higher quality but more friction

Scripted UGC

• Crowdsourced footage with guidelines
• Requires legal releases
• Can build library over time

Platform-Specific Considerations

YouTube Shorts

• 8+ minutes = additional ad breaks (algorithm may favor)
• Include #Shorts tag
• Thumbnail + title still matter for browse/search

Instagram Reels

• 60 seconds is acceptable (but shorter usually better)
• Reels get 2-3x reach vs static posts
• Carousels mostly shown to existing audience
• Use Edits app for AI restyle features

TikTok

• Can publish 4-5 times daily
• Clip compilations work (often backed by sponsorship deals)
• "I don't have bad clips, I publish everything" mindset

Facebook Reels

• Comments drive algorithm reach
• Negative sentiment still boosts (algo doesn't distinguish)
• External links hurt reach in main post

Title & Thumbnail (YouTube)

Click-Through Rate Targets

• Below 4%: Low (fix title/thumbnail)
• 4-6%: Ideal range
• Above 6%: Excellent

Thumbnail Principles

Dial It to 11: What's the most extreme version of this title?

• "Teacher lost control" → Show crying teacher with rambunctious kids

Stock Photos > Illustrations: Real photos with expressions work better

Less Text, Higher Contrast: "They ate me for lunch" → "They hate me"

Show Expressive Faces: If using guest, show reaction/emotion

Title Principles

Every Word After Hook Is a Filter:

• "Five dyslexia myths every parent believes" (narrower)
• "Five dyslexia myths" (broader, more clicks)

Universalize When Possible: Make specific topics broadly appealing

Direct Appeal Format: "If your kid hates school, please watch this"

Complementarity

Title and thumbnail should work together, not repeat:

• Title: "6 months with Ray-Ban Smart Glasses"
• Thumbnail text: "It has one problem"

A/B Testing

Low views usually = bad packaging, not bad content. Try:

• Same video, different title
• Same video, different thumbnail
• Space tests out over weeks
• Revising months later can "restart" performance

Content Pillars Strategy

Don't limit yourself to one format. Build multiple independent pillars:

Each pillar has its own methodology. When one works, triple down on that pillar.

Algorithm Optimization

Triple Word Score

Stack four signals - algorithm indexes ALL:

1. Audio transcript (what's spoken)
2. On-screen text (captions, titles)
3. Caption/description (text below)
4. Hashtags (still matter)

First 10 Seconds

Topic words must appear in:

• Audio (spoken explicitly)
• On-screen text (reinforcing, not competing)
• Visual context (environment matches topic)

Caption Best Practices

• Less text on screen at a time
• Check mobile preview (desktop editing looks different)
• Match what successful podcast clips do
• Captions reinforce audio, don't compete with it

Production Checklist

Before Creating

• What format am I testing? (Tier 1/2/3)
• What's my hook? (passes McDonald's Test)
• Do I have all assets? (footage, audio, graphics)
• Time limit: 15 minutes max

During Creation

• Hook appears in first 3 seconds
• Topic words spoken in first 10 seconds
• On-screen text readable on mobile
• Audio quality acceptable
• 90% good is good enough

Before Publishing

• Triple Word Score complete
• Platform-specific requirements met
• Title/thumbnail complementary (YouTube)
• Caption written
• Hashtags appropriate for platform

After Publishing

• Track performance (views, CTR, comments)
• If low views: consider title/thumbnail revision
• If it works: plan to make more of this format
• Don't delete - sometimes old content resurfaces

Common Mistakes

Production Mistakes:

• Working too long on single video (break 15-min rule)
• Over-polishing before proving format works
• Building systems before finding winners
• Not publishing because it's "not good enough"

Hook Mistakes:

• Burying the lede (punchline should come first)
• Fancy vocabulary (fails McDonald's Test)
• Giving away payoff completely
• Hook doesn't match content (clickbait)

Algorithm Mistakes:

• Topic words not in first 10 seconds
• On-screen text competes with audio
• Inconsistent signals (audio says X, caption says Y)
• Same copy across all platforms

Thumbnail/Title Mistakes:

• Repeating same info in title and thumbnail
• Not extreme enough (dial to 11)
• Every word is a filter (too narrow)
• Giving up after one title fails (A/B test)

Related Skills

• video-caption-creation - Detailed caption and hook writing
• podcast-production - Cutting clips from podcast episodes
• social-content-creation - Text-only social posts

References

• Studio/Social Media/ANDREW-MUTO-AUDIT.md - Full practitioner notes
• Studio/Social Media/Format Notes/video-arsenal.md - Producible formats by tier
• Studio/Social Media/Platform Insights/ - Platform-specific heuristics

Version History

• v1.0 (2026-01): Initial skill based on Andrew Muto practitioner methodology

- Sponge-then-sharpen philosophy - 15-minute rule - Producible format tiers - Title/thumbnail optimization - Content pillars strategy