| name | shortform-content |
| description | Create shortform videos (TikTok, Reels, Shorts) that win attention and convert, using Kane Kallaway's Lego Brick framework. Use this skill whenever the user wants to plan, script, critique, or improve shortform video content — including requests like "write a script for a short", "why did this reel flop", "how do I hook viewers", "break down this viral video", "plan my shortform strategy", or any mention of hooks, retention, virality, thumb-stoppers, bingeability, or growing on TikTok/Instagram/YouTube Shorts. Also trigger when the user wants to reverse-engineer a winning video or build a repeatable shortform system for a business. |
Shortform Content Creation
Create shortform video that earns attention and converts it into trust, using the system Kane Kallaway detailed in his content masterclass (Open Residency podcast). The core insight: strategy is commoditized — the edge is deconstructing every rep into parts, comparing each part against the best in your niche, and iterating faster than everyone else.
Step 0: Pick your content game
Strategy depends on the business model. Make the user pick one of five games before advising anything else:
- Entertainment media — monetize views (CPM, brand deals). Maximize raw views.
- Education media — same monetization, but viewers can tactically apply what they learn.
- Commoditized consumer product — product needs no explanation (clothing, furniture). Goal is awareness, not education.
- Consumer product needing education — viewer must learn what it is before wanting it (e.g., a greens powder). Short education gap to close.
- B2B / high-trust offers — services, coaching, info products, SaaS. Heavy education, long trust curve.
Games 1–2 need maximum views; games 3–5 need on-target views — dollars per view beats views. Kallaway's warning: the pure media game is arbitraging to zero as AI content floods feeds. Most businesses should play 3–5, and the durable position is hyper-niche education with proof, aiming to own one term nobody owns yet (niche terms are easiest to attach your name to).
The 7 Lego Bricks (core framework)
Every shortform video decomposes into seven bricks:
- Topic — what the video is about.
- Angle — what's interesting about the topic: shocking (entertainment) or useful (education).
- Hook structure — the visual hook, text hook, and spoken hook (see below).
- Story structure — how the story is laid out / video type: case study, breakdown, listicle, personal narrative, tutorial.
- Visual format — how visuals sit on screen: split screen, full screen, POV/vlog, circle overlay, in-the-world.
- Key visuals — what fills the format: A-roll, B-roll, graphics, screen recordings.
- Audio — music and sound effects.
The process:
- Find the videos and creators winning in your category.
- Explode their best videos into these seven bricks.
- Hold constant every brick that's already world-class in your version.
- Improve the bricks where you fall short of the benchmark.
- Restack into your own new formation and post.
Topic and angle change every video; hook structure, story structure, and visual format can be held constant and replicated from proven winners. Beginners should copy proven bricks to shrink the sphere of unknown unknowns; only invent new formats after mastering the fundamentals. Differentiation comes from being first-mover on one brick while keeping the other six proven.
The hook framework
A hook has three components that must align:
- Visual hook — what's shown on screen.
- Text hook — on-screen text: captions (word-for-word) plus "title text" (a larger headline not spoken verbatim, e.g., "Stop making hooks like this").
- Spoken hook — what's said.
If they don't match, the viewer gets confused, and confusion leads to churn. The opposite of confusion is comprehension — every element exists to keep comprehension high.
The hook does three jobs in order:
- Stun (stop the scroll). Visuals register before words — they act like a stun gun. Optimize for high motion, high color, high contrast with the rest of the feed. The brain is wired to notice motion; a static talking head loses to anything that moves. Benchmark against your niche's maturity: saturated niches (tech/AI) demand dialed visuals; immature niches (e.g., dentistry) pop with far less.
- Set context in the first sentence. The brain processes top-down: broad context → how → examples → takeaway. Dropping viewers mid-funnel with no context confuses them. Maximize speed to value — compress how fast you get to (or tease) value.
- Create contrast. Contrast = the gap between what viewers expect and what you show them. Two types:
- Direct: state the common belief, then flip it ("Most people think X. The data says the opposite."). Stronger, because leaning them to +100 then flipping to −100 creates a 200-point arc.
- Implied: a contrarian statement or question with no stated common belief. Weaker — the viewer must supply the expectation themselves.
Spoken-word rules: fewest possible words, active voice ("the dog jumped," not "the jump was made by the dog"). But the real KPI is absorption rate — can the viewer absorb 100% of what you said and be certain what it means? Cut fluff, but add words back if cutting drops absorption below 100%.
Scripting: the question chain
The easiest way to validate a script:
- The first line (context) should open exactly one question in the viewer's head. If you can't name the one question it opens — or it opens several of equal strength — rewrite it.
- The next line answers that question and opens the next one. Repeat to the end.
Example: "Google just dropped Gemini 3 and it's a crazy model" → opens what can it do? → answer it → opens how do I use it? → answer → opens what does it cost? → answer. The viewer's brain is constantly guessing what comes next; you either confirm the guess or contrast against it — that zigzag holds attention.
Finding the angle — the shock score. For each fact, take, or implication, ask: out of 100 people, how many already believed this? The fewer, the higher the shock score. Great storytelling is finding things most people don't believe but can be convinced of quickly — the "aha" is the contrast. Chain these: establish the common belief, break it, let the new reality become the next common belief, break again.
Two dopamine hits. A dopamine hit in educational content = something non-obvious the viewer can actually use. Land two hits before the viewer would churn and they're locked in; one hit might hold them; zero and they bounce.
Payoff. Clickbait is fine — bait-and-switch is not. Bait the click, then deliver the bait. A story without a payoff feels like a bait-and-switch and burns trust.
Structure ceiling. A 60-second short holds roughly three story loops/sections: the "what is this" open, the how-it-works breakdown, and the implication.
Editing rules
- Alignment first. The biggest editing mistake is visuals that don't match the words at that moment. Wrong visual at the wrong time = confusion = churn.
- Comprehension only decays. It starts at 100% and every word, frame, and cut either holds it or drops it. Viewers churn when it decays past their threshold. Tightest alignment matters most in the first 5–10 seconds. If someone leaves while fully understanding what the video is about, fine — it wasn't for them.
- The eyes-closed test. Play the finished video with eyes closed. Bored → pacing too slow, chop tighter. Can't absorb it → too fast, add breathing room.
- Visual selection. For each sentence ask: what's the best possible way to visualize this? Sometimes it's just you on screen; use a visual only when it adds absorption.
- Audit sentence-by-sentence. For each sentence: "Am I clear what this sentence means and what I should take from it?" Any unclear sentence is a drop-off point.
- Retention benchmark. On YouTube Shorts, ≥70% of viewers retained at 3 seconds predicts a winner; below 50% means the hook failed. But data is confirmatory and lagging — the principles above are the leading indicators.
Bingeability and content minutes
Content minutes is the conversion model: buying requires trust, and trust requires a certain number of watched minutes ("blocks"). A cheap product might take a handful of blocks; a high-ticket offer takes hundreds. Only watched minutes count, so engineer bingeability — chain-reactions where one video leads to the next in a single session:
- Number your series ("Day 4 of 10," "Lesson 12 of 40"). The brain is a puzzle machine — missing pieces pull viewers back through the catalog. Stack episodes before launching so you don't break the promise.
- Design covers/thumbnails for shorts. Two hidden reasons: (1) profile browsers use covers to decide what to binge next; (2) virality runs on shares, shares happen in DMs, and a shared video sits in the chat showing only its cover — a better cover converts more of those DM opens.
- A long-form video accrues trust ~20x faster per rep than a short (10 watched minutes vs. ~30 seconds), so shorts are for discovery; deeper formats are for trust.
The iteration loop (where the results actually come from)
Volume negates luck, but only with learning attached:
- One rep at a time when starting. Batching lengthens the feedback cycle — you repeat rep 1's mistakes through rep 5. Batch only once your fundamentals are solid.
- Post-mortem every rep with the Lego bricks. Which bricks matched the niche's best execution, which didn't? The skill is looking at your rep vs. a winner and naming the specific difference ("hook too slow, visual didn't grab"), then applying it to the next rep.
- Study your own winners first. The best source of new ideas is your own content that already worked — remake it. No winners yet? Remake the niche's winners with your topic.
- Expect the wall. Everyone is bad for the first dozens of reps because of unknown unknowns. Most quit when they realize how many reps mastery takes. Egoless comparison against the benchmark, applied every rep, is the whole game.
- Get outside eyes. Too many variables drive performance to self-diagnose without emotion. Feedback from someone at or above your level beats staring at analytics.
The production workflow
The repeatable loop: idea → research → script → record → edit → post.
- Idea pipes. Continuously scan defined sources for your niche (other creators, Twitter/X, Reddit, YouTube, newsletters). Shortlist ideas, develop each with a proposed angle and candidate visuals, then pitch-filter: is the topic relevant to the audience you want? Interesting? Time-sensitive? Big enough TAM?
- Ideas are a bottomless pit — it's the first thing to delegate (a "content operator"). Editing is bounded and delegates easily; scripting is the hardest to hand off.
- Topic consistency feeds the algorithm. Platforms sample each video to a small test pool built from your existing audience, then scale concentric circles if engagement beats your baseline. Scattered topics poison the sample pool and flatline every video. Pick one topic lane and hammer it; if you must cover multiple lanes, use separate channels.
Review checklist
When critiquing a short, check in order:
