YouTube Script Writing Full Course (Write In Minutes, Not Hours)

Shane Hummus - Content Growth Engine

Abstract

Nearly two hours laid out as a complete scripting system, opening on a claim that reframes the effort question: hours spent on a script do not predict whether the video works. A small set of specific things does, and the rest of the course is that set.

• Every content format is a list underneath, so everything gets built as one and structure stops being a problem.
• Thumbnail, title and intro get written together, or the three end up promising slightly different things.
• The labour gets divided by what a machine cannot supply, and that remainder — screenshots, proof, editor direction, lived anecdotes — is the edge.
• Telling people to take an action rather than asking lifts compliance; the grammar is the lever.
• A share brings a non-viewer onto the platform with no impression spent, which makes its economics unlike any other request.
• Information makes a video valuable and stories make it memorable — two different properties, and most instruction produces only one.

Everything below follows the course in order, from why scripts matter through the listicle default, the holy trifecta and the assistant workflow, into the calls to action one by one, out through research, craft, story structure, anti-fragility and hiring. For anyone building a personal brand the recurring claim is the one about the moat: everything else in marketing can be copied and a lived history cannot.

Chapter summaries

00:00 - Hours spent on a script do not predict whether the video works

Hours spent on a script do not predict whether the video works; a small set of things does.

"So, as somebody who has over a million YouTube subscribers on my main channel, I've"

— 00:00

Time spent is not the variable. A small set of specific things decides whether a video works, which is the claim the next hour and a half is spent enumerating.

02:38 - Scripts matter for three reasons, and value delivery is the first

Scripts matter for three reasons: they guarantee you deliver value, they raise watch time, and they make content sticky and memorable.

"scripts make sure that you actually deliver value. Now, value is kind of this nebulous"

— 02:38

They guarantee value actually gets delivered, they raise watch time, and they make the content stick. Mastery of a subject is worthless if it cannot be communicated so people understand and resonate with it.

04:18 - A script is a guide rather than something read word for word

A script is a guide, not a word-for-word read; most talking-head delivery improvises off it to sound natural.

"there are many different script types out there. So, you might be thinking, what method"

— 04:18

Most talking-head delivery improvises off the document rather than reading it. Treating it as a guide is what keeps the delivery natural — the document is scaffolding rather than the performance.

08:22 - Every content format is a list underneath

Every content format is, at its core, a list — so build all content as a listicle.

"the type of script that you use, there's another way of categorizing scripts, and that"

— 08:22

Every content format is a list underneath, so everything gets built as one. It is a strong claim and a useful default, because it turns structure into a solved problem before the writing starts.

11:34 - The listicle type gets chosen by whose authority can be borrowed

Choose opinion, fact, or hybrid listicles by whose authority you can borrow; hybrid satisfies both audiences.

"right, so now let's talk about the types of listical scripts. First, you've got opinionbased"

— 11:34

Opinion, fact or hybrid — chosen by whose authority can be borrowed. The hybrid satisfies both audiences, which makes it the safe answer when the authority question is unresolved.

14:53 - Every script starts with a title and a single purpose

Every script starts with a title and a purpose, where purpose is the single action you want the viewer to take after watching.

"topic. So, how do you actually format the script? Well, let's talk about the basic"

— 14:53

A title and a purpose, where purpose means the single action the viewer should take afterwards. Naming one action rather than several is what makes the rest of the script decidable.

16:32 - The intro decides whether anybody keeps watching

The intro decides whether people keep watching, and a weak intro sinks an otherwise strong video.

"inspiration video for just about every video that I ever make. So, just as an"

— 16:32

A weak intro sinks an otherwise strong video, which is why an inspiration video sits at the top of the document — aligning the writer, the editor and the thumbnail designer on the same target.

18:20 - Thumbnail, title and intro get written together

Write the thumbnail, title and intro together — the "holy trifecta" — so all three are congruent; the hook is the first line.

"Now, I go into deep detail on exactly how to write the intro with what"

— 18:20

The viewer meets them in that order, and any incongruence between them breaks the click-to-watch transaction. All three get built at once so the vibe matches across them rather than being reconciled afterwards.

19:35 - The main keyword goes where it will actually be read

Place the main keyword where it is read: near the start of the title, at the top of the description (repeated), and within the first minute of the intro.

"keyword or keywords. And this is the main keyword or keywords that I'm trying to"

— 19:35

Near the start of the title, at the top of the description and repeated there, and inside the first minute. The keyword goes where it will actually be read rather than everywhere.

20:18 - The body gets built in three steps in thirty minutes

Build the body with a three-step, thirty-minute formula: outline the major parts, add the main points per item, then fill in enriching detail.

"of course, you've got the body, right? So, after the intro, you're going to go"

— 20:18

Focusing only on outline, points and enrichment strips out the slow parts of scripting. The enrichment layer is where the stories, anecdotes and proof screenshots go — which is why it is last rather than woven through.

21:46 - An assistant is currently the best writer for this kind of content

An AI assistant is currently the best writer for content, though a chatbot wired to a microblogging platform wins for real-time news.

"integrating AI. And this is going to make your life significantly easier. Now, when it"

— 21:46

A chatbot wired to a microblogging platform wins for real-time news instead. The distinction is by job rather than by preference.

23:29 - A reusable knowledge base gets set up per channel

Set up a reusable project knowledge base you can shift between per channel.

"So to get started, set up projects."

— 23:29

Stored context gets reused across every script instead of being re-supplied each time. A second, business-facing channel has its own project loaded with its own knowledge files — and the projects feature needs the paid tier.

24:06 - Prompting runs in context, goal, prompt order

Prompting runs in context, goal and prompt order, so who is asking and what is being made comes first, then the objective and tone, then the command itself.

"we get into that, I want to talk about what's known as CGP or context"

— 24:06

Front-loading the context yields accurate, relevant output, and the command itself uses action verbs and can point at a same-format template. Context is role, niche, exact title and style; goal is the objective, a retention target and a well-known creator to echo; the prompt is the instruction to create.

28:30 - The outline comes out of the model before the draft does

Have the model produce an outline first for a better final product, and use others' work only as inspiration or template, never as a copy.

"add in itself. Next, you can also tell it to create an outline. So instead"

— 28:30

Asking for an outline first produces a better final product, and other people's work gets used as inspiration or template rather than copied. The outline step is cheap and it changes what the draft can be.

33:24 - The labour gets divided by what the machine cannot do

Divide the labour: let the AI do everything it can, and add only what it cannot — real screenshots, before/after proof, editor direction, lived anecdotes; that is the edge.

"out of 10. And by using this you can see that you know intro to"

— 33:24

Doing only the parts the machine cannot do is what separates this from posting raw output. A top video was roughly eighty per cent written by the model, with the writer adding items, screenshots and direction for the rest — about eighty, not exact.

37:37 - The like is a minimum viable action

The like is a minimum viable action — repeatable on every video — and engaging with it makes viewers likelier to do everything else you ask.

"your script, you also need to add in calls to action. And calls to action"

— 37:37

Repeatable on every video, which is what makes it minimum viable. Engaging with it makes viewers more likely to do everything else that gets asked, so it functions as a gateway rather than as a metric.

39:31 - A comment request raises duration and builds a returning forum

A comment CTA raises view duration, adds social proof and, on controversial videos, turns the comments into a returning mini-forum.

"some YouTube experts out there will tell you it's a waste of time to ask"

— 39:31

People read and re-read active comment sections, and a debate pulls them back repeatedly. It works especially well where the topic is a little controversial, and best where there is a genuine reason to comment at all.

42:29 - The instruction is a command rather than a request

Tell people to take the action, do not ask — commanding a CTA lifts compliance.

"way, one thing you might notice is when I do the calls to action, I"

— 42:29

A share brings a non-viewer onto the platform with no impression spent, which makes it the single most powerful action available. The economics of it are unlike every other request.

42:56 - The subscribe request is strongest on a new channel

The subscribe CTA is strongest when a channel is new, and should be placed several minutes in to select for long-attention viewers.

"Now, another good one is to tell them to subscribe and ring the notification bell."

— 42:56

Telling people to take the action rather than asking lifts compliance. It is a one-word change with a measurable effect — the grammar is the lever.

44:30 - Every video ends by sending viewers somewhere else

End every video by sending viewers to another video, and give zero warning that the video is ending.

"doing a call to action. Now, another really good call to action is to tell"

— 44:30

Strongest when a channel is new, and placed several minutes in so it selects for viewers with long attention. Placement is doing as much work as the request itself.

47:08 - The link is how the money actually gets made

The link CTA is how the money is made: telling people to click lifts clicks about fivefold; capture emails with a lead magnet if you won't sell directly; place the link no earlier than three minutes, add a QR code and a pinned comment, keep one main link above the fold.

"finally, the ultimate call to action, because this is how you actually make money on"

— 47:08

Every video ends by sending viewers to another one, with zero warning that it is ending. The warning is what gives people permission to leave, which is why it gets removed.

52:47 - Research runs across forums, books, newsletters and niche podcasts

Research script content across community forums and subreddits, books, social and newsletters, niche forums, AI tools, and niche podcasts — which are often a year or more ahead of trends.

"bit of a controversial topic. Now that we know how to actually create a script,"

— 52:45

Telling people to click lifts clicks roughly fivefold. Capture emails with a lead magnet if you will not sell directly, place the link no earlier than three minutes, add a code and a pinned comment, and keep one main link above the fold.

56:50 - Two craft moves are specificity and writing the editor's cues in

Two craft pro-tips: be specific and deep with real data from authoritative sources, and write A-roll/B-roll cues into the script to direct the editor.

"talk about seven pro tips for writing better YouTube scripts. And the first one is"

— 56:50

Community forums, books, social and newsletters, niche forums, assistants, and niche podcasts — which are often a year or more ahead of trends. The last one is the least used and the most valuable.

59:34 - Genuinely new value or an old idea said newly

Bring genuinely new value, or say an old idea in a new way — a fresh analogy counts as intellectual property.

"property. So this especially applies to you if you're creating educational content. So you want"

— 59:34

Be specific and deep with real data from authoritative sources, and write the footage cues into the script to direct the editor. Repetition gets minimised and deliberate flare gets added as a pattern interrupt.

1:00:32 - The perceived investment behind the video gets signalled

Signal the perceived investment behind the video — the research effort, the knowledge acquired, the complexity you simplified.

"That's intellectual property. Then overall, you just want to make sure that you're getting a"

— 1:00:32

Genuinely new value, or an old idea said in a new way — and a fresh analogy counts as intellectual property. That is a generous definition of new and it is a workable one.

1:02:16 - Information makes a video valuable and stories make it memorable

Great information makes a video valuable, but stories make it memorable — because brains think in stories.

"information makes your video valuable, but what makes it memorable are stories. So, here's how"

— 1:02:16

The research effort, the knowledge acquired, the complexity that got simplified. Signalling the investment behind a video changes how its content is received, independently of the content.

1:06:15 - A personal story is the moat, because everything else can be copied

Your story is your moat — everything else in marketing can be copied, your lived history cannot; it is your brand and your differentiator.

"talk about how you felt at the time, right? Talk about the emotions that were"

— 1:06:14

Great information makes a video valuable and stories make it memorable, because brains think in stories. Memorable and valuable are different properties and most instruction only produces one.

1:11:50 - A reusable story structure covers three kinds of story

Use a reusable story structure — normal life, a challenge, a mentor or a thing that helps, a resolution, an improved life — and tell one of three kinds: about yourself, about someone you know, about a notable figure.

"Now, let's talk about a basic storytelling structure because depending on what story you're trying"

— 1:11:50

Everything else in marketing can be copied and a lived history cannot. That makes the personal story the brand and the differentiator simultaneously, which is why it appears in almost every piece of advice in this pillar.

1:17:51 - The script has to be anti-fragile

Make the script anti-fragile — built to withstand edits, cuts and last-minute changes without the video collapsing.

"making your script anti-fragile. Now where did I get this word anti-fragile from? Well, I"

— 1:17:51

Normal life, a challenge, a mentor or a thing that helps, a resolution, an improved life. One structure, reused, telling one of three kinds: about yourself, about somebody you know, about a notable figure.

1:19:23 - More content gets created than the script needs

Create more content than you need — over-produce items so weak ones can be dropped without breaking the script.

"First of all, create more content than you need. This is one of the most"

— 1:19:23

Built to withstand edits, cuts and last-minute changes without collapsing. The script is designed for the conditions it will actually meet rather than for the conditions in which it was written.

1:22:08 - The structure is modular, like blocks connected to a theme

Use a modular structure — self-contained sections that connect to the theme like Lego blocks — with flexible transitions and no rigid numbering.

"secret is use modular structure. Make your content in such a way where it basically"

— 1:22:08

Over-produce the items so the weak ones can be dropped without breaking anything. Redundancy at the writing stage is what buys freedom at the editing stage.

1:23:20 - Optionality gets built into the shoot itself

Build optionality into the shoot: write several intros and pick one while filming, give live instructions to the editor, re-record, and keep redundant audio.

"Now, additionally, let's say you are very picky about your introductions. One thing that we"

— 1:23:20

Self-contained sections connected to a theme like blocks, with flexible transitions and no rigid numbering. Numbering is what makes a cut expensive, so it goes.

1:28:52 - Hiring a writer who loves the platform beats hiring an expert

The recommended route is to hire a writer who loves the platform and binge-watches it, then teach them the niche.

"the pros of this one really quick. Uh you'll probably make the best content with"

— 1:28:52

Several intros written and one picked while filming, live instructions to the editor, re-records, and redundant audio kept. Optionality gets built into the shoot rather than hoped for afterwards.

1:35:01 - Team roles scale toward only recording

Hiring on-camera talent to remove yourself carries a large risk: you build a presenter's brand and they can leave and become a competitor.

"is a very very good sign when that happens. But there is another system which"

— 1:35:01

Hiring a copywriter who is also a subject-matter expert makes the best content and is rare and costly. The recommended route is hiring somebody who loves the platform and binge-watches it, then teaching them the niche — and team roles scale from solo to a creative director to only recording.

1:37:53 - The whole system runs as one ten-step build

The whole system runs as one ten-step live build: title, knowledge base, Context-Goal-Prompt, a list with extras, an outline, the draft, verification with real data and screenshots, personal stories and testimonials with an embedded CTA, review, then repeat per section.

"and produce a script effectively that will make you more money from YouTube while working"

— 1:37:53

Hiring on-camera talent to remove yourself builds a presenter's brand, and they can leave and become a competitor. The whole system then runs as one ten-step build: title, knowledge base, prompt order, list, outline, draft, verification, stories and testimonials with the call to action embedded, review, repeat.

Personal Branding Lessons

A full system with a lot of moving parts. The moves below are the load-bearing ones.

Build every piece of content as a list

Every format is a list underneath, so everything gets built as one. It turns structure into a solved problem before the writing starts, which removes the stage where most scripts stall — and the choice between opinion, fact and hybrid then follows from whose authority can be borrowed. 08:22

Write the thumbnail, title and intro together

Written separately, the three end up promising slightly different things and the viewer notices in the first five seconds. Written together they are congruent by construction, with the hook as the first line of the intro. 18:20

Prompt in context, goal, prompt order

Who is asking and what is being made, then the objective and tone, then the command. The order is the technique. Getting it backwards produces generic output from a setup that contained everything needed to avoid it. 24:06

Add only what the machine cannot produce

Let the assistant do everything it can, then spend your time on the remainder: real screenshots, before-and-after proof, editor direction, lived anecdotes. That remainder is the edge, and it is the only part worth the hours. 33:24

Command the action rather than asking for it

Telling people to do something rather than asking lifts compliance. It is a one-word change with a measurable effect, which makes it the cheapest improvement available anywhere in the process. 42:29

End without warning and send them to another video

The warning that a video is ending is what gives people permission to leave. Removing it and cutting straight to the next video keeps the attention that would otherwise disperse at exactly the moment it was about to. 44:30

Over-produce so weak items can be dropped

Create more items than the finished piece needs, so that cutting a weak one breaks nothing. Redundancy at the writing stage buys freedom at the editing stage — which is also why the structure stays modular and unnumbered. 1:19:23

Hire a writer who loves the platform and teach them the niche

A copywriter who is also a subject-matter expert produces the best work and is rare and expensive. Somebody who loves the platform and binge-watches it can be taught the subject — which is the available version of the same hire. 1:28:52

Questions

Each answer ends at the moment in the recording where it is given.

Why do some of my videos flop even after I spend hours on the script?

Because hours and results are decoupled — most of that time goes into parts that do not move the outcome. The two ends of one person's own range were hundreds of hours on a script that bombed and under an hour on one that reached millions. The fix is not more effort but checking the few things that matter: the opening, the congruence of title and image, whether real value is delivered, and the structure. 00:00

Do I have to read my script word for word on camera?

No — a script is treated as a guide, not a transcript. Reading verbatim is possible but it takes heavy practice to sound natural, whereas working off bullet points, a flowchart or slides keeps you organised while you say things in your own words. The point is delivery that sounds conversational rather than recited, and even heavily prepared speakers improvise off the page to get there. 04:18

What is the single most important part of a video to get right?

The opening. It decides whether anyone stays, and a strong piece with a weak intro was shown sinking on its own, because early drop-off signals the distribution system that the whole thing is bad. The recommended fix is to build the image, the title and the first line together so all three carry the same feel, since a viewer meets them in that order and any mismatch breaks the decision to watch. 16:32

How should I actually prompt an AI to write my script?

In a fixed order — context, then goal, then the command. Context is who you are, your niche, the exact title and your style; the goal is your objective, a retention target and the tone you want; the prompt is an action-verb instruction that can point at a same-format piece as a template. Front-loading the context is what makes the output accurate rather than generic. Have it produce an outline before the full draft. 24:06

What should the AI write, and what should I add myself?

Let the model do everything it can, and add only what it cannot. In the worked case roughly 80% of a top piece was model-written; the human added real screenshots, before-and-after proof, editor direction and lived anecdotes. That remainder is the edge, because it is exactly what separates your work from raw AI output — and it comes from knowing your audience and your own material better than any tool can. 33:24

How do I end a video so people keep watching my other work?

Send them straight to the next piece and give no warning that this one is ending. A wind-down line — "well, that's my list, hope you enjoyed it" — is a stop signal that triggers exit before the next prompt can land. Instead the last point segues directly into why the next piece is worth clicking, which is how you build a run people binge rather than a single video they leave. 44:30

Why do stories work better than facts, and what should I tell stories about?

Because information makes a piece valuable but stories make it memorable — a vivid one is pictured involuntarily and takes lasting emotional space, while facts fade. Use a simple arc: normal life, a challenge, something that helps, a resolution, a better life. Tell it about yourself, someone you know, or a well-known figure. Your own story is the strongest, because it is the one asset no competitor can copy. 1:02:16

Should I hire a scriptwriter, and what should I look for?

Two routes. A subject-matter expert who can also write produces the best work but is rare and costly, and is only worth it for a business already at scale. The recommended route is to hire a strong writer who genuinely loves the platform and binge-watches it, then teach them the niche — screened with one question about their favourite platform. Expect to train almost anyone; near-none arrive ready to go. 1:27:34

Sources

YouTube Script Writing Full Course (Write In Minutes, Not Hours)

The sections above follow the episode's own order. Each timestamp is the point in the recording where the idea is discussed, so the post reads as a map of the conversation rather than a rearrangement of it.

Quotations come from the episode transcript with spoken filler ("okay", "right", "you know", "um", "actually", "kind of", "sort of") and false starts removed. No words are added: square brackets mark anything inserted for sense, and an ellipsis marks any cut. British spellings are restored, and speech-to-text errors were repaired against context only where the intended word is unambiguous. Where a line appears both in a cold-open teaser and in its place in the conversation, the timestamp cites the second — the point at which it is actually said.

Quotes are stamped with the moment they are said rather than with a speaker's name.