# The Viral Script System: 155 hooks, 6 timed scripts and an AI script doctor for short videos that get watched, followed and commented

**The direct-response method behind short videos that work, built on Eugene Schwartz and checked against 77 real breakout videos.**

by Dan Shipped (@danshipped) · AI & SaaS — shipped. · Last verified: October 2026

---

Most short videos don't fail because of the edit, the mic or the algorithm. They fail in the first sentence, and again in the last one.

The first sentence talks about the tool instead of the viewer. The last one asks for three things at once ("like, follow, comment, share"). Everything in between explains instead of shows.

This kit fixes both ends and the middle. It's the system I use to write every video for @danshipped, and it comes from two places:

1. **Eugene Schwartz's *Breakthrough Advertising*** (1966), the book many direct-response copywriters treat as their bible. I've translated it from print ads to 30-second videos. No academic tone, just what to do.
2. **77 real breakout videos** from 27 AI and business creators (videos that did 2.5× to 12.8× their creator's usual views), tracked by my own radar between July and October 2026.

## What's inside

| Folder | What it is | Use it when |
|---|---|---|
| `tools/hook-scorer.html` | Offline tool: scores any hook 0–100 with 7 visible rules, tells you what to fix, writes scored rewrites | Before you film. Every time. |
| `skills/script-doctor/` | A Claude skill (+ `PROMPT.md` for ChatGPT/Gemini): paste an idea or draft → diagnosis, hook score, 5 new hooks, a tight script, on-screen text, one CTA | When a script feels "off" and you don't know why |
| `hooks/150_HOOKS.md` | 155 hooks, sorted by where your viewer is, each as a template + a filled AI-niche example | When you're staring at a blank first line |
| `hooks/BREAKOUT_PATTERNS_77.md` | What 77 breakout videos have in common, with the multiplier as evidence | When you want to know what's working *now* |
| `templates/` | 6 script templates with second-by-second timing, each with a fully written example | When you know the idea but not the shape |
| `checklists/` | The 10-question pre-publish check + an outlier tracker (CSV) | Before posting / every Friday |

---

## Start here: your 5-minute win

Pick one. Both take under 5 minutes.

**Option A — score the hook you were about to post (2 minutes)**
1. Double-click `tools/hook-scorer.html`. It opens in your browser. Nothing is uploaded anywhere.
2. Paste the first sentence of your next (or last) video.
3. Read the red rows. Fix the one with the biggest gap.
4. Scroll to **Rewrites**, type your task and audience, and click **Use** on the highest-scoring one. Adjust it to sound like you.
5. Keep anything that scores **75+**.

**Option B — get a full script from an idea (5 minutes)**
1. Open `skills/script-doctor/PROMPT.md`, copy the box into ChatGPT, Claude or Gemini.
2. Fill the 5 lines at the bottom (your niche, platform, goal, the proof you really have, your idea).
3. You get back: which desire your video channels, where your hook must start, a hook score, 5 new hooks, a timed 25–45s script with on-screen text, and one CTA.
4. Film the ★ hook version today.

Using Claude? Install the skill instead. In Claude Code: copy the `skills/script-doctor` folder into `~/.claude/skills/`. In the Claude app: zip the `script-doctor` folder, then upload it as a custom skill (Claude's help center, "Use skills in Claude", has the current steps; code execution must be on in Settings → Capabilities). Then just say "doctor this script" and paste.

That's the win. The rest of this guide is *why* it works, so you can do it without the tools.

---

## Chapter 1 · What a short video actually sells

Every short video is a direct-response ad. It sells **one action**:

1. **The follow.** Your main free "product". It buys you the next video.
2. **The comment keyword.** Your lead magnet. It buys you a DM, then an email.

That's it. Views are not the product. A video with 400,000 views and 12 follows is a failed ad that happened to get seen. A video with 9,000 views and 600 keyword comments is a working one.

So before you write a word, decide: **does this video sell the follow, or the comment?** Never both. Two calls to action split attention, and the viewer ends up doing neither.

Schwartz's whole book rests on three ideas. Each one becomes a decision you make before writing:

| Schwartz idea | The question it answers | Where it shows up in your video |
|---|---|---|
| **Mass desire** | What does the viewer already want? | The promise of the whole video |
| **State of awareness** | How much do they know about their problem and its fix? | *Where* your hook starts |
| **State of sophistication** | How many videos like this have they seen? | *What* your hook says: a claim, a mechanism, or an identity |

The next three chapters take them one at a time.

---

## Chapter 2 · One desire per video

> "Copy cannot create desire for a product. It can only take the hopes, dreams, fears and desires that already exist in the hearts of millions of people, and focus those already existing desires onto a particular product." — Eugene Schwartz, *Breakthrough Advertising*

Plain English: **you don't make people want things. You find what they already want and point it at your video.**

Nobody wakes up wanting "a new AI note-taker". They wake up wanting to stop losing the details of yesterday's client call. The tool is the vehicle. The desire is the engine.

### The rule
**One video = one desire = one promise.** If your script serves two desires ("save time *and* make more money *and* look professional"), the viewer believes none of them.

### How to find the desire (the 3-test)
Schwartz rated desires on three things. Use them to pick the strongest one for each video:

1. **Urgency** — does it hurt *today*? ("My inbox is a mess right now" beats "I'd like to be more organized someday".)
2. **Staying power** — does it come back no matter how often it's satisfied? (Wanting to keep up with AI never ends; there's a new tool tomorrow.)
3. **Scope** — how many people feel it? (All freelancers hate unpaid admin. Few care about one obscure file format.)

### Desire bank for AI, business and creator niches
Pick one per video. Say it in the viewer's words, not yours.

| Desire | In the viewer's words | Typical video |
|---|---|---|
| Don't fall behind | "Everyone's using AI and I'm still doing this by hand." | New tool, new feature, "what changed this week" |
| Do more with less | "I don't have time for this." | Tool demo, automation, before/after |
| Stop paying for it | "Why am I paying $30 a month for this?" | Free alternative, "I cancelled 5 subscriptions" |
| Build my own thing | "I want my own product / income." | Build in public, "shipped in 48h" |
| Get clients / money | "I need more clients this month." | Outreach, proposals, pricing |
| Look competent | "I don't want to look like an amateur." | Templates, polish, "what pros do" |
| Get seen | "Nobody watches my stuff." | Hooks, scripts, growth (this kit) |

**The tool is never the desire.** The tool is the proof and the mechanism. Start from the desire; bring the tool in as the *how*.

---

## Chapter 3 · Awareness: where your hook starts

This is the most useful idea in the book, and the one most creators get wrong.

Schwartz sorted people into five stages by how much they know. **Your hook must start where they are.** Start too far ahead and they don't get it. Start too far behind and they're bored.

| Stage | What they know | Where the hook starts | Example hook |
|---|---|---|---|
| **Unaware** | Nothing. They don't know they have a problem. (Most of the cold feed.) | Their **identity**, their day, a story, an observation | "If you answer client emails every morning, this is for you." |
| **Problem-aware** | They feel the pain, don't know there's a fix | The **problem**, in their words | "Spending 3 hours on slides every week? Stop." |
| **Solution-aware** | They want the result, don't know your way exists | The **desire** + "there's a way" + a hint of how | "There's a way to turn any link into slides in 30 seconds." |
| **Product-aware** | They know the tool or category, not convinced | The **difference**: test, comparison, verdict | "I tested 23 AI note-takers. Only 2 survived." |
| **Most aware** | They know you and want what you offer | The **offer** or the series, directly | "Comment VIRAL and I'll send you all 155 hooks." |

### The wall
A hook written for the most-aware viewer ("Comment VIRAL for my hooks!") means nothing to a stranger. A hook written for the unaware ("If you post every day and still get 200 views…") bores your followers. Schwartz calls the gap between them a psychological wall: the same sentence works on one side and dies on the other.

### The practical rule
- **Growth videos (cold feed): start at Unaware, Problem-aware or Solution-aware.** That's where your next 10,000 followers are.
- **Stories, posts for followers, launch reminders: start at Product-aware or Most aware.**
- **The tool name goes first only when** the viewer already knows it (product-aware), or the tool *is* the news that day. In the breakout data, 36 of 77 titles open with a product or model name, almost all within days of a big launch, when the whole market became product-aware overnight. On any other day, start from the viewer.

The 155 hooks in `hooks/150_HOOKS.md` are sorted by exactly these five stages (A to E).

---

## Chapter 4 · Sophistication: claim, mechanism or identity

Awareness tells you *where* the hook starts. Sophistication tells you *what it can say*.

Sophistication is how many similar promises your viewer has already heard. "5 AI tools that will save you hours" was fresh once. Now it's background noise. Schwartz described five stages, and each needs a different kind of hook:

| Stage | The market | What works | Example |
|---|---|---|---|
| **1** | Nobody has made this promise yet | A **simple, direct claim** | "This turns a call into a proposal." |
| **2** | A few have; you need to beat them | The **claim, made bigger** (a number) | "This turns a 40-minute call into a proposal in 2 minutes." |
| **3** | Everyone makes the claim; enlarging stopped working | A **new mechanism**: *how* it works | "It reads the transcript first, then writes the proposal in the client's words." |
| **4** | Competitors copied your mechanism | The **mechanism, made better** (faster, easier, safer) | "It reads 2,000 AI sources every 15 minutes. I only see the 3 that matter." |
| **5** | People are tired of all of it | **Identity**: speak to who they are | "For people who ship while everyone else scrolls." |

### Where AI and creator niches are now
"AI tools" content on Reels and TikTok is at **stage 4–5**. Viewers have seen thousands of "X AI tools that…" videos. If your hook is a simple claim, they discount it before you finish the sentence. Making it bigger ("50 AI tools!!") makes it worse.

**So in a crowded niche, do one of two things:**
1. **Say the how.** Name the mechanism: "I built a radar that…", "the 3-tab method", "it reads your last 40 emails first". A mechanism is new even when the claim isn't.
2. **Speak to the identity.** "Builders don't wait for the newsletter." "This isn't for AI hobbyists. It's for people who bill clients."

The breakout data agrees: the two formulas with the highest median multiplier were **borrowed authority** (a credible person's *how*, 4.85×) and **identity / who you become** (4.65×). Plain "how to use X" titles had the lowest median (2.75×) and none reached 5×.

### The 10-video test (2 minutes)
Before you write, search your topic on the platform you post on. Watch the first 2 seconds of the top 10 videos. If 7 of them make the same promise, you're at stage 3+: lead with the mechanism or the identity.

---

## Chapter 5 · The 25–45 second structure

Every script in this kit follows the same five blocks. The timing flexes; the order doesn't.

| Time | Block | Its only job | Schwartz technique |
|---|---|---|---|
| **0–2s** | **Hook** | Make them hear the next sentence | Headline + awareness stage |
| **2–6s** | **"This is you"** | A guaranteed "that's me" | Identification: put the viewer in the scene |
| **6–20s** | **Show it** | The result happening on screen | Intensification: the product in action, not described |
| **20–26s** | **Proof / how** | One specific number or one line of mechanism, placed where they'd think "is this real?" | Verification + mechanization |
| **Last 4–6s** | **One CTA** | Follow OR comment the keyword | The offer as the climax |

### Rules for each block

**Hook (0–2s)**
- ≤12 words. One thought. Same promise in 3–6 words of on-screen text.
- The first *frame* shows the result or the problem, never a logo or you saying hi.
- More in Chapter 6.

**This is you (2–6s)**
- One sentence the right viewer can't say no to: "If you [their exact task] every [period], this is for you."
- Use their words. "Answer client emails" beats "manage communications".
- It also pushes the wrong viewers away. Good. They weren't going to follow.

**Show it (6–20s)**
- Screen recording or real footage. A new shot every 2–3 seconds.
- Narrate what's happening, not what the tool "can do". "Paste the call notes. Click once. There's your proposal."
- Reveal the result before the mechanism. If the how is complicated, it comes after they've seen it work.

**Proof / how (20–26s)**
- **One** proof. A number, a price, a time, a before/after, a screenshot. One believed claim beats ten half-believed ones.
- Place it exactly where they'd doubt, right after the result appears. Proof too early is a boring statistic; too late, they've already left.
- If you don't have a real number, measure one before posting. Never invent it.

**One CTA (last 4–6s)**
- Follow *or* keyword. Chapters 7 and 8 have the exact formulas.

### Word budget
Spoken pace is roughly 2.5–3 words per second (time your own voice once; it varies).

| Length | Words |
|---|---|
| 25s | ~65–75 |
| 35s | ~90–100 |
| 45s | ~115–130 |

If your script is 180 words, you don't have a 45-second video. You have two videos, or one with filler.

### The 6 formats
All six templates in `templates/` use this structure with different timing:

| Template | Length | Best for | Sells |
|---|---|---|---|
| `01_TOOL_DEMO.md` | 30s | A visible result on screen | Keyword |
| `02_MYTH_BUSTER.md` | 35s | Wrong advice in your niche | Follow or keyword |
| `03_BEFORE_AFTER.md` | 25s | Same thing, ugly → good | Keyword |
| `04_I_TESTED_X.md` | 40s | Comparing options honestly | Follow or keyword |
| `05_STORY.md` | 45s | A real moment with one lesson | Follow |
| `06_LIST_THAT_ISNT_A_LIST.md` | 35s | "3 levels of…" without being another list | Keyword |

---

## Chapter 6 · Writing hooks that hold

> The purpose of the headline is to make them read the next sentence. Not to sell. (Schwartz's point, paraphrased.)

A hook doesn't need to be clever. It needs to open a gap the viewer wants closed.

### The 7 rules (they're also the scorer's rules)

1. **≤12 words.** About 4 seconds out loud. 75 of the 77 breakout titles were 12 words or fewer; the median was 9.
2. **A real number or a picture.** 47 of 77 breakout titles had a number. "3 hours" beats "a lot of time". "$30 a month" beats "expensive".
3. **Talk to the viewer.** "You", "your", a named audience ("freelancers"), or a direct instruction ("Stop…").
4. **Open a gap.** A pointer ("This is why…"), a question, a contrast ("…isn't the edit"), or an open ending (":"). Don't answer it inside the hook.
5. **No hype adjectives.** Insane, game-changer, unlock, mind-blowing. Schwartz warned that adjectives the reader *notices* lower belief. Replace them with the evidence: not "this AI is insane" → "it answered 40 support emails in 2 minutes."
6. **Start from the desire or the problem.** First words about *them*. Never "Hey guys", "So", "In this video".
7. **One thought.** If you need two "and"s, it's two hooks.

### The hook stack (do this for every video)
1. Write the **desire** in one line.
2. Decide the **awareness stage** (Chapter 3).
3. Pull **3 hooks of different types** from that stage in `hooks/150_HOOKS.md`.
4. Fill them, score them, keep the best.
5. Make the **first frame** match it (show the result or the pain).
6. Write the **on-screen text**: the same promise in 3–6 words.

### What the 77 breakout videos add
Full analysis in `hooks/BREAKOUT_PATTERNS_77.md`. The short version:

| Formula | Median multiplier | Short-form version |
|---|---|---|
| Borrowed authority ("X reveals his workflow, just copy him") | **4.85×** | "How senior engineers review AI code. Just copy this." |
| Identity / who you become | **4.65×** | "How to become the freelancer who never chases invoices." |
| Ranged list ("from useful to almost impossible") | 3.05×, but the **#1 breakout at 12.8×** | "7 AI agent uses, from easy to almost impossible." |
| Hands-on effort ("I spent 100 hours…") | 2.95×, top one 8.4× | "I spent 100 hours testing AI editors. Here's the verdict." |
| Plain how-to / guide | **2.75× (lowest), 0 of 8 reached 5×** | Never alone. Add a who, a number or a range. |
| Stakes / warning | 3.25×, **0 of 8 reached 5×** | Only as a door to a solution you show right away. |

Honest caveats: these are YouTube titles, not Reels; I only have the winners (no flops to compare); groups are small. Treat them as strong hints. The patterns are what transfer, not the words.

### Don't copy hooks. Copy structures.
Schwartz warned about the "echo ad": copy a competitor's headline and change one word, and you remind people of *them*. Take the shape ("I tested [N] [things] on [real task]. Only [N] survived."), never the sentence.

---

## Chapter 7 · Proof and belief (why no-hype wins)

Schwartz had a rule that's worth more than any hook formula: **a claim believed 100% sells far more than ten claims half-believed.**

On short-form, belief comes from four things:

1. **Specific numbers in the right place.** "Saves time" → ignored. "Mondays used to be 3 hours of slides. Now: 3 minutes." → believed, *if* you show it.
2. **Deadly sincerity: admit a limit.** "It's ugly and in beta. It's also the fastest way I've found to…" The flaw makes the strength believable. This is why "I tested 23, only 2 survived" works: you're visibly throwing things away.
3. **Camouflage: don't look like an ad.** A real screen recording with your natural voice often beats a polished edit. Native beats produced.
4. **Never fight what they already believe.** If they think most AI tools are hype, agree: "You're right, most are. This one isn't, and here's the test." Build the bridge from their side.

### The no-hype rewrite drill
For every adjective in your script, ask: *what's the evidence?* Then say the evidence instead.

| Hype | Evidence |
|---|---|
| "This tool is insane" | "It wrote 30 product descriptions before my coffee was ready" |
| "A game-changer for freelancers" | "My proposals went from 2 hours to 15 minutes" |
| "Unlock your productivity" | "I got my Fridays back" |
| "The ultimate AI setup" | "Three tools. $40 a month. That's the whole stack." |

(Use your real numbers. These are shapes.)

---

## Chapter 8 · Selling the follow

The follow is a product. Sell it like one: a **promise**, a **mechanism**, an **identity**, and **proof**.

| Part | What it means | Example (Dan Shipped) |
|---|---|---|
| Promise (one) | What they get by following | "One AI tool a day, before everyone else." |
| Mechanism | Why you can keep that promise | "My AI watches the whole AI world, 24/7." |
| Identity | Who follows you | "If you'd rather ship than scroll." |
| Proof | Visible consistency | Numbered drops ("Drop #014"), a series name, real results |

### Follow CTA formulas
- **Teaser:** "Follow — tomorrow: [NEXT VIDEO'S PROMISE]." (Announce the next video without finishing it. The open loop is the reason to follow.)
- **Series:** "This is part 2 of [SERIES]. Part 3 drops tomorrow."
- **Identity:** "Follow if you're the kind of [ROLE] who [BEHAVIOR]."
- **Mechanism:** "I test [N] tools a week so you don't have to. Follow for the ones that survive."

### What makes people follow (beyond the CTA)
- **A recognizable format.** Same opening style, same series name, same layout. People follow a show, not a video.
- **A promise you repeat in new scenes.** Schwartz: you can't repeat, but you can reinforce. Same central idea, a new example every time.
- **Your profile backs the promise.** Bio = the promise in one line. Pinned videos = your 3 best examples of it.

---

## Chapter 9 · Selling the comment keyword

The viewer who reached your CTA just watched 30 seconds. They're warm. Your job now is to make one tiny action feel obvious.

### The 5-part keyword CTA
1. **Name the asset, with a number.** "All 155 hooks" beats "my hooks".
2. **Make it absurdly easy.** "Just comment VIRAL."
3. **Give a reason it's free.** A free thing with no reason feels like a trap. "Free, because I'd rather you follow the next drop."
4. **Compare to the expensive option (optional).** "What I'd charge a client for a script audit, in your DMs."
5. **One keyword. One CTA.** Never "follow AND comment".

**Full example:**
> "Comment VIRAL and I'll send you all 155 hooks plus the script doctor. Free, I just want you here for the next one."

### Choosing the keyword
- One word, all caps on screen, easy to spell, impossible to mistype.
- Related to the asset (VIRAL, HOOKS, BRIEF, VOICE). Not your brand name.
- The same everywhere: in the video, the caption, the pinned comment and your DM automation.
- One keyword per video. If you have two lead magnets, make two videos.

### The DM flow (what happens after they comment)
1. **Public reply** to their comment ("Sent, check your DMs"). It shows others it's real.
2. **DM opener** that asks for a tiny reply ("Reply OK and I'll send it"). It opens the conversation.
3. **Delivery** with the link, what they'll get in the first 5 minutes, and one question ("What will you use it for?"). Their answer is your next video idea.

Most creators automate steps 1–3 with a DM automation tool connected to Instagram (e.g. ManyChat). Check each platform's current rules on automated messages before you set one up.

---

## Chapter 10 · The outlier method: find what's working in your niche, free

An **outlier** is a video that did far more views than its creator usually gets. Total views tell you who's big. Outliers tell you **which idea** beat the creator's own average, so they're the closest thing to a signal you'll get for free.

**The rule: a video at 3× or more of its creator's usual views is worth studying.** (In my data, half of the 77 flagged breakouts cleared 3×.)

### Step 1 · Build a list of 10–20 creators (one-time, 20 minutes)
Mix sizes: a few big ones, but mostly creators with 5k–200k followers in your exact niche. Small creators' outliers are more useful, because their audience isn't doing the work for them.

### Step 2 · Find their "usual" (YouTube, the easiest)
1. Open the channel. Go to the **Shorts** tab (or **Videos** for long-form).
2. Leave the sort on **Latest**. Write down the views of their last ~15 posts (skip anything younger than a week).
3. Sort the numbers and take the **middle one**. That's their usual (the median). Use the median, not the average, so one viral hit doesn't distort it.
4. Switch the sort to **Popular**. Anything at 3× the usual or more → into your tracker.

### Step 3 · Search YouTube by topic (catches creators you don't know)
1. Search your topic (e.g. "AI proposal", "Claude workflow").
2. Open **Filters**. Set **Type: Shorts** (YouTube added this option in early 2026), **Upload date: This week** or **This month**, and **Prioritize: Popularity**. ("Prioritize" replaced the old "Sort by", and "Popularity" replaced "View count"; it also weighs watch time.)
3. For any video with surprising views, click the channel and do Step 2 for that creator. Big views on a small channel = outlier.

### Step 4 · TikTok (manual)
1. Open the creator's profile. The grid shows a view count on each video thumbnail.
2. Scan the grid for numbers that are 3× the ones around them. That's your outlier. (If the profile offers a sort by popular, use it. Options vary by app version and region.)
3. In search, use the filter menu to limit to recent dates, so you see what's working now, not in 2023.

### Step 5 · Instagram (manual)
1. Open the creator's profile → the **Reels tab**. It shows view counts on each reel (unless the creator has hidden them).
2. Eyeball the median of the last 15, flag anything at 3×+.
3. Bonus: if you have a professional account, Instagram's **Trial Reels** show a reel to non-followers first. Use it to test two hooks on the same video. (Eligibility depends on your account; check in the app.)

### Step 6 · The tracker (every Friday, 15 minutes)
Use `checklists/outlier-tracker.csv` (opens in Excel, Numbers or Google Sheets). For each outlier, write:
- the multiplier,
- the hook or title (word for word),
- the **formula** (use the table in `hooks/BREAKOUT_PATTERNS_77.md`),
- the **awareness stage** it starts from,
- the **desire** it channels,
- **what to steal: the structure only**, and your own version of the hook.

After 4 weeks, sort by formula. The ones that keep showing up are what your niche rewards right now.

### What NOT to do
- Don't copy the words (the echo ad). Take the structure, desire and format.
- Don't chase one-offs. One 10× video can be luck. Three 3× videos with the same formula is a pattern.
- Don't compare across creators by raw views. Always use the multiplier against *their own* usual.

---

## Chapter 11 · The 10-question pre-publish check

Run it in 60 seconds. Two or more "no" → fix before posting. (Printable version with fixes: `checklists/PRE_PUBLISH_10.md`.)

1. **One desire.** Can I say the single desire in one sentence, in the viewer's words?
2. **Right starting point.** Does the hook start where this viewer is?
3. **Crowded topic?** Am I giving the *how* or speaking to *who they are*, not repeating the claim?
4. **Hook does one job.** ≤12 words, one thought, answer not given away?
5. **"That's me" by second 6.** Will the right viewer recognize themselves?
6. **Shown, not explained.** Is the result happening on screen?
7. **One true proof, right place.** A number or screenshot exactly where they'd doubt?
8. **No ad words.** Zero insane / game-changer / unlock?
9. **One CTA.** Follow OR keyword, with a reason it's free?
10. **Only I could post this.** If I swapped my name for another creator's, would something be lost?

---

## Going further

### 1. Run a weekly script session (60 minutes, 5 videos)
| Minutes | Do |
|---|---|
| 0–15 | Update the outlier tracker. Pick 5 formulas worth trying. |
| 15–25 | Write 5 desires (one per video), each with its awareness stage. |
| 25–40 | 3 hooks each from `150_HOOKS.md`. Score them. Keep the best per video. |
| 40–55 | Fill the matching template for each. Run the script doctor on the weakest. |
| 55–60 | Pre-publish check on all 5. |

### 2. Turn one idea into a series
Schwartz's "rule and example": one central idea, a new example every time. Pick your rule ("I test AI tools so you don't have to") and name the series ("Shipped or Skipped"). Every episode is a new example. Series give people a reason to follow: they want the next one.

### 3. Test hooks, not videos
When a video underperforms, the idea is usually fine and the hook isn't. Re-cut the same video with a hook from a *different type* (not a reworded version of the same one) and post it again a few weeks later, or test it via Instagram Trial Reels. Track both in the tracker.

### 4. Read the comments as research
Every keyword DM ends with "what will you use it for?". Their answers are your next desires, in their exact words. Put the best phrases straight into your "this is you" lines.

### 5. Go deeper on Schwartz
The method here is a translation of *Breakthrough Advertising*. If you want the source, the chapters on the states of awareness and sophistication, and on "intensification" (making the reader see the product working), are the ones that map most directly to short-form.

---

## Limits (read this, honestly)

- **No tool can predict views.** The hook scorer checks craft, not outcomes. A 90 hook on a weak idea still loses, and a scrappy hook on a perfect idea can win. Some strong story hooks score 65–75 because the rules reward "you" and numbers. Use the score to catch mistakes, not as a verdict.
- **The breakout data is YouTube, AI/business only, July–October 2026,** with no control group. It shows what appears among winners, not what *causes* winning. Your niche may reward different formulas: that's why the outlier method is in here.
- **The script doctor is only as honest as your input.** It's told never to invent proof and will leave `[PROOF: …]` placeholders. Fill them with real numbers or don't post the claim.
- **Platforms change.** Filters, view counts, Trial Reels and DM rules described here were checked in October 2026 and will drift. If a button moved, look for the same idea nearby.
- **This won't fix a weak offer.** If your lead magnet isn't useful, a perfect CTA just brings more disappointed people faster.
- **Example numbers are placeholders.** Every number in the hook and script examples shows a shape. Replace it with yours.

---

## Want this done automatically?

Right now you do this by hand: find outliers, pick a formula, score hooks, fix the script.

I'm building **Hook Doctor**: a small web app that watches the breakout videos in *your* niche, scores and rewrites your hooks and scripts against what's working this week, and keeps your swipe file updated for you.

**Want early access? Reply YES to my DM** (or join the waitlist from the link in my bio). If enough of you say yes, I build it next.

**Follow @danshipped for the next drop.** I build SaaS alone with AI, in public, and give away the systems I use.

---

*Sources and checks (October 2026): YouTube search filter changes (Type: Shorts, "Prioritize: Popularity", removed "Sort by rating" and "Last hour") as reported by 9to5Google, January 2026. Instagram Reels-tab view counts and Trial Reels as described in Instagram's rollout coverage; eligibility varies by account. Breakout data: Shipyard OS radar, 77 videos from 27 YouTube channels, captured July 5 – October 1, 2026; multiplier = views ÷ the creator's median views. Schwartz quote from* Breakthrough Advertising *(1966).*

Made by Dan Shipped — @danshipped