short-form-video

name

short-form-video

description

Write production-ready annotated scripts for short-form video (YouTube Shorts, TikTok, Instagram Reels). Use when the user wants to create a short video, write a video script, plan a Reel, Short, or TikTok, or produce short-form content. Covers script frameworks, hooks, pacing, visual directions, text overlays, audio cues, and edit notes for 15–90 second educational and thought-leadership content.

Short-Form Video Scripting

Write fully annotated, production-ready scripts for 15–90 second videos. Based on data from 3.3B+ Shorts views (Paddy Galloway), 35B Shorts views (vidIQ), and frameworks from Jenny Hoyos, George Blackman, and Brendan Kane.

Required Inputs

Before scripting, gather or confirm:

1. Topic/core idea — The single insight, lesson, or argument. One idea per video.
2. Platform — TikTok, YouTube Shorts, Instagram Reels, or cross-post. Affects length, audio, CTA.
3. Production style — Talking head, faceless/voiceover, tutorial/screen share, story/vlog, or hybrid.
4. Context (optional) — Series position, audience level, specific CTA goal, trending audio to use.

If the user hasn't specified platform or style, ask. Default to hybrid style and cross-post if unspecified.

Core Principles (ranked by impact)

1. The hook is a 3-channel simultaneous system

71% of viewers decide in 3 seconds. The hook must work across visual + text overlay + audio simultaneously, all reinforcing one message. No greetings, no setup, no logos.

See hooks.md for the combined hook formula and proven patterns.

2. Optimize every decision for completion rate

Completion rate is the #1 algorithm signal on all platforms. A 30s video at 85% completion beats a 60s video at 50%. Default to the minimum length that delivers the insight (30–45s for single concepts). Structure for watch-to-end, not just watch-to-start.

3. Pattern interrupts every 5–8 seconds

Visual changes every 2–4 seconds. Re-hook the viewer with jump cuts, zoom shifts, B-roll insertions, text overlays, tonal shifts, or SFX. One creator reported 40% retention increase from optimizing clip lengths alone.

See pacing-retention.md for cadence rules and loop structure.

4. Every second must be annotated

The 6-column format (TIME | VISUAL | VOICEOVER | TEXT OVERLAY | AUDIO/SFX | EDIT NOTES) must be fully populated. Sparse scripts produce mediocre videos. The script IS the production plan.

See script-format.md for the full format, vocabulary, and example scripts.

5. Captions and text overlays are non-negotiable

85% of social video is watched muted. Captions boost view time by 12% and make viewers 80% more likely to watch to the end. Reinforce key spoken phrases with text overlays — bold claims, stats, takeaways.

Framework Selection

Default to Hook → Tension → Payoff. Match to content:

Framework	Duration	Best for
HTP (default)	30–60s	Insights, counterintuitive takes
PAS	15–30s	Pain-point content, positioning
Myth → Truth → Proof	30–45s	Contrarian thought leadership
Micro-storytelling	45–60s	Personal lessons, case studies
List/Listicle	30–45s	Practical tips, value delivery
Before/After	30–45s	Results, transformations
One Thing	15–30s	Daily content, single insights
Tutorial	30–60s	Teaching, processes

See frameworks.md for timing breakdowns and structural details.

Workflow

Step 1: Assess inputs + validate topic

Confirm topic, platform, production style. Identify the single takeaway the viewer should remember. Validate the topic against Jenny Hoyos's 4 Criteria (Novelty, Uncertainty, Knowledge Gaps, Complexity) — see frameworks.md. If the topic fails on Novelty or Uncertainty, reframe the angle before proceeding.

Step 2: Select framework + duration

Match framework to content type. Set duration based on framework sweet spot and platform:

• TikTok: 21–34s highest completion, 15–60s range
• YouTube Shorts: 30–60s, algorithm rewards reaching 30s mark
• Instagram Reels: 60–90s for max views

Step 3: Write the hook (3 channels)

Specify simultaneously: what the viewer SEES (visual direction), READS (text overlay, 3–7 words), and HEARS (first spoken words, 5–10 words). All three reinforce one message.

Step 4: Write the full annotated script

Use the 6-column format with header block. Populate every column for every time segment. Visual changes every 2–4s. Pattern interrupts every 5–8s. Voiceover at 140–170 WPM for educational content.

See script-format.md for column definitions and annotation vocabulary.

Step 5: Add CTA + ending

Every CTA must be dual-delivered: spoken + text overlay simultaneously. Place a mid-roll CTA at ~50% (not all viewers reach the end) and an end CTA in final 3–5s. Structure content to be genuinely save-worthy (lists, frameworks, steps) — save-baiting only works if the content earns it. For TikTok: design a loop-friendly ending (no "in conclusion"). Match CTA to platform norms.

See platform-guide.md for CTA strategy, comment-driving formats, and platform-specific norms.

Step 6: Write posting metadata

• Caption/post copy for the platform
• Hashtags: 3–5, platform-appropriate
• Cross-posting notes if multi-platform (what to change)

Step 7: Self-check

Before presenting:

• Hook lands in under 3 seconds? All 3 channels (visual + text + audio) specified?
• One core idea only? Single clear takeaway?
• Duration matches framework sweet spot AND platform?
• Visual change every 2–4 seconds? Pattern interrupt every 5–8 seconds?
• Every column populated for every time segment?
• Text overlays reinforce key spoken points? 3–7 words each, 1.5s+ on screen?
• Voiceover word count matches duration? (~75w/30s, ~110w/45s, ~150w/60s)
• All text within safe zones? (900×1400px centered in 1080×1920)
• CTA embedded naturally, not appended? Matches platform norm?
• No greetings, no preamble, no slow setup, no "in conclusion"?
• Loop-friendly ending for TikTok?
• Reading level at 5th grade or below?

Output Format

Present each script as:

**Title:** [title]
**Platform:** [platform(s)]
**Framework:** [framework used]
**Duration:** [Xs]
**Style:** [production style]

---

[6-column annotated script table]

---

**Caption:** [platform post copy]
**Hashtags:** [3–5 hashtags]
**Cross-post notes:** [adjustments for other platforms, if applicable]

**Word count (voiceover):** [N words]
**Reading level:** [grade level — target 5th grade or below]
**Pattern interrupts:** [N interrupts across Xs — one every ~Ys]

Anti-Patterns (never do these)

• Start with greetings, logos, intros, or "Hey guys welcome back"
• Bury the interesting point past 3 seconds
• Leave any column empty — sparse scripts produce bad videos
• Use static talking head for 30+ seconds without visual change
• Signal the ending ("In conclusion," "To wrap up") — kills loops
• Ask for likes/follows before delivering value
• Repurpose long-form by simple clipping without new hooks + restructuring
• Over-produce (cinematic > authentic on social platforms)
• Place text in unsafe zones (bottom 320px TikTok, bottom 300px Shorts/Reels)

References

• frameworks.md — 8 script frameworks with timing breakdowns
• hooks.md — 3-channel hook system, proven phrases, hook killers
• script-format.md — 6-column format, visual vocabulary, example scripts
• pacing-retention.md — Cut frequency, information density, loop structure
• platform-guide.md — TikTok vs Shorts vs Reels: algorithms, CTAs, cross-posting
• anti-patterns.md — 13 mistakes ranked by impact + benchmarks