ncreighton Overview
The Video Transcript to Micro-Content Repurposer transforms long-form video transcripts into ready-to-publish micro-content across all major platforms. Instead of manually extracting highlights, generating captions, and rewriting content for different channels, this skill automates the entire workflow.
Why This Matters
Video creators and marketing teams waste hours manually repurposing content. A 60-minute webinar transcript could yield:
- 8-12 short-form clips (TikTok, Instagram Reels, YouTube Shorts)
- 15-20 quotable pull quotes (LinkedIn, Twitter/X, Pinterest)
- 3-5 blog post sections with SEO optimization
- 10+ email subject lines and preview text hooks
This skill does all that in minutes, identifying the highest-impact moments automatically using AI analysis of engagement patterns, emotional peaks, and key concepts.
Integrations & Platform Support
Works seamlessly with:
- Video Platforms: YouTube, Loom, Vimeo, Zoom transcripts
- Content Platforms: WordPress, Medium, Substack
- Social: LinkedIn, Twitter/X, TikTok, Instagram
- Email: Mailchimp, ConvertKit, ActiveCampaign (via templates)
- Cloud Storage: Google Drive, Dropbox, AWS S3
- Video Editors: CapCut, Adobe Premiere (JSON export for subtitles)
Quick Start
Example 1: Extract LinkedIn Pull Quotes from a Webinar
Transcript: [Paste your 30-min webinar transcript here]
Task: Extract 8 LinkedIn-ready pull quotes with context.
Format: Include the speaker name, timestamp, and 2-3 follow-up engagement questions per quote.
Tone: Professional but conversational, emphasis on insights that drive shares.
Expected Output:
- 8 standalone quotable moments
- Each with attribution, timestamp, and suggested hashtags
- Character count optimized for LinkedIn feed + carousel
- Suggested image dimensions (1200x627px)
Example 2: Generate Short-Form Video Clips with Captions
Transcript: [Your YouTube video transcript]
Duration: Original video is 45 minutes
Clip Length: 60 seconds maximum per clip
Platforms: TikTok, Instagram Reels, YouTube Shorts
Task: Identify hook-worthy sections. Generate captions in SRT format with speaker identification.
Include: B-roll suggestions, pacing notes, and emotional intensity markers (0-10 scale).
Expected Output:
- 10-12 clip suggestions with exact timestamps
- SRT subtitle files ready for video editors
- Hook copy for each clip (max 280 characters)
- Recommended music/sound effect genre per clip
- Technical specs: resolution, aspect ratio, duration
Example 3: Create Blog Post Snippets & SEO Metadata
Transcript: [Your expert interview transcript]
Length: 90-minute deep-dive conversation
Task: Extract 4 standalone blog sections (800-1200 words each).
Requirements: SEO-optimized headers, meta descriptions, internal linking suggestions.
Include: Pull quote callouts, expert credentials, CTA buttons.
Blog Platform: WordPress
Expected Output:
- 4 complete, publish-ready blog sections
- Meta titles and descriptions
- Suggested featured images (dimensions + keywords)
- Internal link recommendations
- Keyword density analysis
- Estimated read time
Capabilities
1. Intelligent Hook Identification
Uses multi-model AI analysis to identify:
- Emotional peaks (excitement, revelation, urgency)
- Key concept introductions (when an idea is first mentioned)
- Story moments (case studies, customer wins, personal anecdotes)
- Contrarian statements (perspective shifts, myth-busting)
- Question prompts (calls-to-action, engagement drivers)
Each moment is scored 0-100 for repurposing potential.
2. Multi-Platform Output Formatting
- Pull quotes (280-500 characters)
- Carousel decks (10-15 slides)
- Document shares (formatted as PDF)
- Video snippets with captions
- Hashtag recommendations (#contentmarketing #videomarketing)
Twitter/X
- Threads (5-15 connected tweets)
- Quote snippets (120-140 characters)
- Thread starters with engagement hooks
- Retweet-optimized versions
TikTok/Reels
- Hook copy (first 3 seconds critical)
- Captions with speaker identification
- Pacing recommendations
- Trending audio suggestions
- Call-to-action variations
- Subject line variations (5 A/B test pairs)
- Preview text (50-85 characters)
- Body copy sections
- Email template recommendations
- CTA button copy
3. Caption & Subtitle Generation
- Automatic SRT/VTT formatting
- Speaker identification and color-coding
- Timestamp accuracy to within 1 second
- Multiple language support (auto-translate)
- Accessibility-optimized (punctuation, pacing)
4. SEO Optimization
- Primary keyword identification from transcript
- Meta tag generation (title, description)
- Header hierarchy (H1, H2, H3)
- Internal linking suggestions
- Featured snippet formatting
- Schema markup recommendations
5. Engagement Metrics & Predictions
- Predicted engagement score (0-100) for each clip
- Optimal posting times by platform
- Hashtag recommendations with volume/competition
- Similar trending topics
- Audience sentiment analysis
Configuration
Required Environment Variables
# OpenAI API key (for transcript analysis and content generation)
OPENAI_API_KEY=sk-xxx
# Anthropic Claude API (optional, for fact-checking and nuanced writing)
ANTHROPIC_API_KEY=sk-ant-xxx
# Optional: Cloud storage access
GOOGLE_DRIVE_API_KEY=xxx
AWS_S3_ACCESS_KEY=xxx
AWS_S3_SECRET_KEY=xxx
Setup Instructions
Get your transcript (plain text, VTT, or SRT format)
- YouTube: Use automatic captions (Settings → Captions → Show Transcript)
- Loom: Download transcript automatically
- Zoom: Export from cloud recording
- Manual: Paste from rev.com, Otter.ai, or similar
Specify your platforms (which channels are you using?)
platforms: ["linkedin", "twitter", "tiktok", "email", "blog"]Set output preferences
tone: "professional" | "casual" | "educational" | "humorous" industry: "SaaS" | "ecommerce" | "health" | "finance" | "tech" clip_duration: 30 | 60 | 90 # seconds target_audience: "founders" | "marketers" | "developers" | "general"
Example Outputs
Output Type 1: LinkedIn Pull Quote Card
╔════════════════════════════════════════════════════════╗
║ "The biggest mistake companies make is assuming ║
║ users want more features. Users want their time ║
║ back." ║
║ ║
║ — Sarah Chen, Product Lead at TechCorp ║
║ [Timestamp: 23:15] ║
║ ║
║ 💡 What's your biggest product assumption? ║
║ 🔗 Reply in comments ║
║ #ProductStrategy #UserExperience #Startup ║
╚════════════════════════════════════════════════════════╝
Character Count: 298
Estimated Reach: 2,400-4,200 impressions
Optimal Post Time: Tuesday, 8 AM PT
Output Type 2: Short-Form Video Clip Spec
CLIP #3: "The Pivot Moment"
Duration: 58 seconds
Timestamp: 34:12 — 35:10
Engagement Score: 87/100
Hook Type: Contrarian Statement + Story
HOOK COPY (First 5 seconds):
"We spent 2 years building the wrong thing. Here's what we learned."
CAPTIONS (SRT format):
00:00:00,000 --> 00:00:03,000
We spent 2 years building the wrong thing.
00:00:03,500 --> 00:00:06,800
Here's what we learned about product market fit.
[... continues through 00:00:58 ...]
PLATFORMS: TikTok, Instagram Reels, YouTube Shorts
ASPECT RATIO: 9:16 (vertical)
RESOLUTION: 1080x1920 minimum
B-ROLL SUGGESTIONS:
- Screen recording of old product (0-5 sec)
- Customer feedback (5-15 sec)
- Current product demo (15-55 sec)
- Call-to-action graphic (55-58 sec)
MUSIC/SOUND: Uplifting, contemplative tone (Spotify: "Indie Focus" playlist)
PACING: Slow reveal, builds to insight at 45 sec
Output Type 3: Email Subject Line A/B Tests
Subject Line Pair #1 (Test Audience: 25%)
A: "We wasted 2 years building this (here's the lesson)"
B: "The $500K mistake that taught us about product-market fit"
Expected Winner: B (curiosity gap + specificity)
Send Time: Tuesday 10 AM
Subject Line Pair #2 (Test Audience: 25%)
A: "Product strategy: The contrarian take"
B: "Why more features actually hurt your users"
Expected Winner: B (concrete benefit)
Send Time: Tuesday 10 AM
Winner Send: Thursday, both audiences (50%)
Control: Your standard subject line to 25%
Preview Text Suggestions (50-85 chars):
"We learned the hard way about feature bloat. Here's our framework..."
Output Type 4: Blog Post Section (SEO-Ready)
BLOG POST SECTION: "Building Products Users Actually Want"
SEO METADATA:
- Primary Keyword: "product-market fit strategy" (monthly volume: 890)
- Secondary Keywords: "feature creep", "user feedback", "MVP development"
- Meta Title: "Product-Market Fit: Why More Features Hurt Growth (Case Study)"
- Meta Description: "Learn the $500K lesson about feature bloat. Real strategy for building products users actually want in 2024."
- Word Count: 1,247 words
- Estimated Read Time: 5 minutes
CONTENT:
---
## Why More Features Actually Hurt Your Growth
Sarah Chen, Product Lead at TechCorp, made a discovery that changed how her team thinks about product development.
"We spent 2 years building features," she explains in an recent interview. "And we completely missed what users actually wanted."
### The $500K Mistake
[Full section with blockquotes, subheaders, examples...]
> "The biggest mistake companies make is assuming users want more features. Users want their time back." — Sarah Chen
### The Framework We Use Today
[Detailed methodology section...]
---
INTERNAL LINKING SUGGESTIONS:
- Link "MVP development" → your guide on building MVPs
- Link "user feedback loops" → your customer research article
- Suggest this post link back from: "Feature Prioritization Framework"
FEATURED IMAGE:
- Dimensions: 1200x627px
- Alt Text: "Product roadmap showing prioritized features vs. MVP scope"
- Suggestion: Chart showing feature adoption curve
- Designer Brief: Show contrast between complex vs. simple product interfaces
CALL-TO-ACTION:
- Primary CTA: "Download our Product Strategy Checklist" (pdf lead magnet)
- Secondary CTA: "Read our case study on market fit" (internal link)
- Placement: After section 2, end of post
SCHEMA MARKUP RECOMMENDED:
- Article schema
- FAQPage schema (if FAQs included)
- Author schema (Sarah Chen credentials)
---
Tips & Best Practices
1. Transcript Quality Matters
- Use auto-captions when possible (YouTube, Zoom are highly accurate)
- Clean up common errors before processing:
- Remove timestamps if using raw text export
- Fix obvious speech-to-text errors (names, technical terms)
- Remove filler words if you want more polished quotes
Pro Tip: If using Otter.ai or Rev, request "verbatim" transcripts (they preserve natural speech patterns better).
2. Optimize for Your Audience First
Don't just repurpose for all platforms equally. Ask yourself:
- Which platform does my audience actually use? (B2B execs on LinkedIn, creators on TikTok)
- What's the maturity of my audience? (Students need simple language; CTOs want technical depth)
- What action do you want? (LinkedIn drives B2B leads; TikTok builds awareness)
Pro Tip: Use the target_audience parameter to ensure all outputs match your ICP.
3. The 70/20/10 Rule for Content Mix
- 70%: Educational/value content (frameworks, tips, lessons learned)
- 20%: Behind-the-scenes/personal stories (builds relatability)
- 10%: Direct promotion (product, course, book, CTA)
Use this skill's engagement prediction scores to verify your mix aligns.
4. Batching Creates Compounding Returns
- Repurpose 1 long-form video → 40-50 pieces of micro-content
- Schedule this content over 3-4 months
- Compound effect: Each piece drives backlinks, cross-platform discovery, brand presence
Pro Tip: Use the skill on your competitor's videos too (public, educational content only). See what resonates.
5. Captions Dramatically Increase Video Completion
- Videos with captions have 25% higher watch-through rate
- 85% of video is watched with sound OFF (mobile use case)
- Use this skill's SRT export directly in CapCut, Premiere, or YouTube Studio
Pro Tip: Add captions even to platform-native uploads (not just external embeds).
6. Email Subject Lines Are Your Biggest Leverage
- A/B testing one subject line can increase CTR by 20-50%
- Use the predicted winner recommendation, but always test in your market
- The skill suggests 5 pairs; test the top 2 pairs first, then scale the winner
7. SEO Compounds Over Months
- Each blog snippet is a separate keyword target
- Link them to each other (internal linking boost SEO)
- Republish the same content in 6 months (Google loves fresh updates)
Safety & Guardrails
What This Skill Will NOT Do
❌ Fabricate information. All outputs are extracted/derived from your source transcript. We don't add facts, stats, or claims not in the original video.
❌ Bypass copyright. You must own, have license to, or have permission to repurpose the source video. Check:
- Your original video rights
- Guest speaker permissions (especially for interviews)
- Licensed music/imagery rights
- Platform ToS (YouTube, LinkedIn, etc.)
❌ Publish without review. AI-generated content should always be reviewed by a human before publishing:
- Verify tone matches your brand voice
- Check facts and attributions
- Ensure context isn't lost in extraction
- Validate punctuation and emoji use
❌ Create deepfakes or manipulated video. This skill generates text and captions, not altered video content. All output is authentic to source material.
❌ Guarantee engagement metrics. Predictions (87/100 engagement score) are estimates based on content analysis, not guarantees. Actual performance depends on:
- Your audience size and loyalty
- Timing of posts
- External trending topics
- Platform algorithm changes
Boundaries & Limitations
🔸 Transcript accuracy: Skill quality = source transcript quality.