Short-Form
Scripting Masterclass

Views ≠ money.
Who sees it beats how many.

"You can get 1M views and make $0, or 1,000 views and make $1M."

• The format you pick pre-selects the audience the algorithm feeds you.
• Pick formats that pull qualified, older, decision-maker viewers — not broke, younger scrollers.
• Script for the buyer, not the view count. Every rule ahead serves this.

The 3-second gate Convergent

Instagram's #1 read: does the viewer survive the first 3 seconds? That's the skip rate.

• You need ≥50% held past the first 3s or IG stops distributing entirely.
• 80% of your time and energy belongs to the hook.

Three hooks fire in frame 1 Convergent

The brain pre-categorizes the video as worth-it or waste from three things fired inside ~1 second.

When all three pull the same direction, the first 3 seconds do more work than the rest of the video combined.

Sends & shares are now #1 Convergent

Mosseri confirmed: sends carry more weight than likes, comments, or saves. Comments are now dead last.

Test every script: "Would someone actually send this to a friend?"

Don't judge a post on day one Convergent

The old 1–2 hour spike-or-die window is gone. IG now runs a YouTube-style multi-day cycle, re-testing against new pools.

• Review at 48–72 hours minimum, never day 1.
• Signal density > volume: post 1×/day for 30–90 days so the algo learns exactly who you're for.

You don't find winners.
You manufacture them. Convergent

• 1B/mo creator shifted 40 hrs/wk from one polished video → 100 imperfect ones → 10 hit 10M+.
• Script every variation of an idea (talking / no-talking / B-roll) — you can't know the winner until they're all posted.

Three archetypes carry the most evidence

Treat these as your defaults. The other 12 families are rotations.

The 5–8 word rule Convergent

• Under 4 words rarely lands — except outliers like "Unpopular opinion."
• Reading level ≤5th grade. "When you confuse, you lose."
• Niche jargon buyers actually use is fine ("waterproofing"). Agency jargon is banned — no "Selling-Spin Method."

Stack the hook across surfaces

Layer 2–3 surfaces that all point at the same promise — fishing with multiple lures. Layers help only when they reinforce, never compete.

Keep audio as reinforcement only — the opening must still work on mute.

The text hook can beat the spoken one Convergent

"The text hook is 10× more valuable than the spoken hook — text on screen is where your eyes go first."

• 99 of 100 people read the title text, then decide whether to keep watching.
• Write the on-screen title as a standalone line for the eye — not a transcript of sentence one.
• Green-screen over your own headline gets 3–5× more viewers than a plain talking head.

Goldilocks the curiosity gap Convergent

Too wide = no reason to care. Too narrow = answer visible, they self-solve. Aim for the middle.

Blur, crop, or delay the reveal. The coffee reel: "This coffee should not taste good" + blueberries blurred so viewers can't self-solve.

12 fill-in hooks to memorize

Compress each to ≤12 words. Niche + pain readable in 3 seconds.

Hook mistakes to avoid

Half-second screenshot test: show frame 1 to a "drunk grandma" — does she get the niche + tension on mute? A contractor should grok the trade + the problem with sound off.

Every winning short rides one spine Convergent

• Every talking-head source converges here. What changes is what fills each slot, not the spine.
• One clear payoff — not payoff-payoff-payoff. The viewer must know what to take away.

Mitchell's 7 Laws & vein theory

One video = one "vein of thought": one topic, one problem, one line hook→action. Cut anything that repeats or detours — 30–50% of words in most content are non-essential.

Raw Dog Reel: plan backwards

The £11,102/week engine. Record forwards, plan backwards.

• Hook ≠ insight. Insight resolves; the hook is the disruptive tease of it.
• Talk through the idea, not about it. Filter: "if anyone else could say it, don't."
• Don't script — bullet points only. Lower the bar; perfectionism is the killer.

The silent-movie / POV reel $1M/mo

B-roll + on-screen text, no voiceover, dramatizing a relatable conversation. The open loop opens at the hook, closes at the CTA — that drives completion.

• Hook = the opening line of a conversation that hints at good news or names a clear problem — cropped to leave a loop.
• Build = text slides amplify the pain, trace it to a common enemy (the corporate job, the failed business).
• Bridge = introduce the vehicle so a third-grader gets it → show timeline to first win → payoff.
• Takeaway → CTA = follow / DM / comment a keyword. Save rate is the headline metric.

Pacing rules that hold viewers

Posting-limit folklore is wrong (Hormozi org): one account posts 100×/day for 9B views/mo. Each post goes to a different ~1% slice. "You're just not doing enough."

Dopamine = anticipation, not reward

Dopamine fires on expected reward, not delivery. The first 2s are a neurochemical event.

• Stack multiple micro-questions in the opening. The Stan example: a CEO walking through an expensive office = 3–4 dopamine triggers before a single word about the product.
• Dopamine spikes hardest when the content type shifts (funny → emotional) — engineer shifts mid-script so the brain can't settle.
• Never lead with the product/logo — the brain instantly files it as an ad and scrolls before a thought forms.

The 47-second attention clock

Every cut is a micro-reset of the clock — it tells the brain "something new is coming, stay."

• Every clip must deliver new information; cut any frame that doesn't; close every pause.
• Re-hook cadence: one pattern interrupt in the first 2s, one every 30s for long-form.
• Line-level rule: every sentence either educates or re-hooks. Cut lines that do neither.

Sell the identity, not the pain point

"People don't buy products, they buy identities." The brain only cares about what a thing means to who they want to be.

Inside a story, viewers stop arguing with what they hear — skepticism drops, emotion takes over. Skeleton: Hook → Problem → Story → Payoff.

Put a credential in-frame

The brain takes an authority shortcut — a credential in the first 2s is accepted uncritically as a reason to keep watching.

Also fire social proof (view counts, "everyone's doing this") inside the first 3s to trigger conformity.

Design a reward at three points

Likes cluster at 3 moments — a proxy for the retention curve. Don't burn all your energy on the hook.

Iannotti's reel shape: hook (0–3s) → problem agitation (3–10s) → unique insight (10–30s) → CTA (final 5s). Viewers die at ~10s without sharper stakes and ~30s without a genuinely new take.

Best formats for money & buyer quality

Give every format a funnel job Convergent

• Selling to cold traffic in Stories fails — that's a Reels job.
• The carousel script must be tactical/technical enough to pre-sell (the tax math, the profit math).

Write in beats, not prose Convergent

The edit isn't a post-step. Pre-segment the script into a spine before a word is spoken — each beat pre-tagged with its visual treatment.

• Carry a parallel visual column: [icon], [background swap], [cutout speaker], [screen-record demo].
• Flag which words get promoted to kinetic on-screen text — the script names where text lands and what it says.
• Pick the music before filming. BPM sets pace; the song's dips are your cut points. Script beats land on musical cut points.
• Write cleanly caption-able sentences — captions ARE your spoken hook rendered as text.

Simple vs heavy edit: decide before writing

Both sides agree: color grade / LUT is the last step, audio is half the product, and the hook can be a motion device. If content is your revenue engine, you should not be editing at all — outsource it.

Premium = intention, not polish Convergent

Lighting, editing, wardrobe is ~20%. Treating content as direct-response with funnel intent is the 80%.

• Never copy a viral creator wholesale — lead with your ICP-specific value + personality + story.
• Script only from {ICP's problems} ∩ {your story & proof} — Brand-Market Fit.

Comment a keyword → DM → book

The reel starts the conversation; the DM closes. Money lives in 1:1, never in the feed.

• Earn the CTA. Only viewers who complete the piece — feel the pain, feel the transformation — get asked to act. Compliance is far higher.
• Open loop opens at the hook, closes at the CTA.
• Do you even CTA? On pure organic reach, a hard CTA can suppress watch-time. On conversion pieces (carousels, VSLs, retargeting) an explicit CTA is required.

Track the numbers that mean money

Keep 10 lead magnets per client and rotate them to re-engage old leads. Tooling: ManyChat handles 3–4K DMs/day; GHL chat is capped ~500 msgs/30 days.

Contractor-specific plays

Organic → ThruPlay ad flywheel

Turn every reel into a paid asset. Because you keep viewers on the platform, CPM collapses.

• Weekly: pull the top on-ICP reels into one Meta ad set, objective = ThruPlay (~£45/day across 10–15 reels).
• Builds ThruPlay + follower retargeting audiences → bottom-of-funnel drives to the VSL/offer, hitting only people who already know you.
• No funnel yet? Run profile-visit ads — "your Instagram is your funnel."

Build neighbor-forwardable reels

Shares reach non-followers — new homeowners — your growth channel. Design for the viewer's friend, not the viewer. Be the wingman.

Iannotti's 3-prong shareability test — hit 1 = shares, 2 = double, 3 = explode:

Contractor share-bait: "3 questions to ask any [roofer/cleaner] before you hire them." Insider/contractor's-secret angles hit all three prongs. If a viewer can't describe it to a friend in one line, they won't share.

Where the sources disagree — honestly

The 9-point pre-production gate

Run every hook through this before you produce anything.

Now go manufacture some winners.

Package first · hook in frame 1 · one payoff · earn the CTA · script for the buyer.

Synthesized from 28 sources in the St. Pierre brain · 2026-07-24