How to Write a YouTube Script Step by Step

Most YouTube scripts do not fail because of bad writing. They fail because of bad planning. The creator opens a document and starts typing, and the result is a transcription of a thought process, which is exactly as meandering as thought processes tend to be. That draft starts slow because the creator was warming up, loses momentum in the middle because nothing structural gives the viewer a reason to stay, and ends abruptly because there was nothing left to say.

The fix is not "be a better writer." The fix is engineering, and four mechanisms do most of the work:

1. The Hook Trident. A three-part opening: the spike (a specific claim or result), the proof (one piece of evidence), and the stakes (why it matters to this viewer).
2. The SPEC outline. A four-part plan, Setup, Points, Evidence and Conclusion, that fixes scope and order before you draft a word.
3. The 300-word section rule. A drafting constraint: no section runs past 300 words without a structural break. It is a working convention, not a measured threshold, and it is the easiest of the four to enforce on a draft.
4. The adversative transition. A rule for the joins between sections: they move the argument forward rather than announcing the next item on a list.

They are presented in the order they touch a draft. The hook is planned first, the outline is built around what the hook promises, the draft runs under the section-length rule, and the transitions are where the structure becomes audible to a listener. Where a number in this guide is measured, the source is linked. Where a number is a working convention, the text says so.

The generic write-a-script process, the part every competitor publishes, is compressed into one summary at the end, so you can see where the four mechanisms sit inside a normal workflow.

The order is counterintuitive: plan the hook before you draft, then write the actual hook text after the body is finished. If you write the hook first, you make promises before you know what the video delivers, and you end up claiming "5 game-changing tips" when only three survived drafting. If you plan it first and write it last, the hook can promise exactly what the video delivers: the strongest moment, the most surprising finding, the most useful tip.

The three prongs.

Prong 1: The Spike. A specific, unexpected claim or result.

• "I spent $3,000 testing every lighting setup on Amazon."
• "There is a YouTube setting that is off by default that doubles your suggested impressions."
• "I have been making videos for 4 years and just discovered I was doing [fundamental thing] wrong."

Prong 2: The Proof. One piece of evidence that makes the spike credible.

• "Here is the data." / "Let me show you the side-by-side." / "My analytics before and after."
• This does not need to be comprehensive. It needs to signal that the video has substance, not just claims.

Prong 3: The Stakes. Why this matters to the viewer specifically.

• "If you are filming in a home studio, this changes your entire setup."
• "This is the difference between a video that gets 500 views and one that gets 50,000."
• "Most creators do not know about this because [reason], and it is costing them [specific thing]."

Template: "[Spike: specific claim with number or result]. [Proof: one piece of evidence]. [Stakes: why this matters to you]."

What the measurement says about timing. The three prongs run about a sentence each, so the planned opening is roughly three sentences long. Our 349-video hook study found that hits and flops delivered their first concrete reason to keep watching at almost the same point (median word 14 for overperformers against word 15 for flops), so an early payoff on its own separates nothing. What separated the groups was whether the opening engaged the title's promise: 87 percent of overperformer openings against 78 percent of flops.

How to use it. Write three hook options. Run each through script analysis and keep the one with the highest hook score that still sounds natural in your voice. The published bands put a strong score at 80 and above (prepublish-fe/lib/score-bands.ts), so treat anything below that as a draft opening rather than a finished one. For more hook patterns, see our hook examples guide and first 30 seconds guide.

Before drafting, build a structural outline on the SPEC framework. The proportions below are planning shares, not measurements.

S: Setup. The context the viewer needs: the problem you are solving or the question you are answering. This is not the hook. It is the short orientation that follows the hook.

P: Points. The main arguments, tips, steps or sections, in order. For each one, write a single sentence describing the value it delivers.

Critical rule: every point must pass the "so what?" test. Ask why it matters to the viewer. If you cannot answer immediately and specifically, the point is too obvious, too vague, or belongs in a different video.

E: Evidence. For each point, note what supports it: a statistic from a source you can cite, a personal experience, a demonstration, an example, or a comparison. Points without evidence read as opinions. Points with evidence read as expertise.

C: Conclusion. The takeaway, what the viewer should understand or do afterwards, and the "one more thing" that rewards viewers who stayed.

Three planning rules.

• Keep the point count small. Past seven points, numbered formats tend to lose the consistency that makes them satisfying. That is a formatting convention, not a measured threshold. If you have more than seven, cut the weakest or split the video.
• Order the points for momentum, not logic. Logical order is chronological or simple to complex. The order that holds attention starts strong, carries momentum through the middle, and saves the most impactful point for the second-to-last position. The reasoning behind that placement is the peak-end finding covered in the watch time psychology guide.
• Allocate words per section. A word budget shows you which section will run long before you write it, and it is the fastest way to keep the 300-word rule from step 3.

Three copy-paste outlines. These are the starting structures we use. The word figures are starting allocations to adjust, not targets to hit. They were published as a standalone template page, which is now folded in here so one URL owns script structure.

Outline 1: the standard video. Commentary, analysis, opinion, and most educational content. The word figures are starting allocations for a mid-length video, to be scaled to your target.

HOOK (100 words):
[Specific claim or result]: "I tested/discovered/found [specific thing]"
[One piece of evidence]: "Here is the data/proof/comparison"
[Why this matters to YOU]: "If you are [viewer situation], this changes [specific outcome]"

SECTION 1 (250 words):
[Mini-hook]: why this section matters, in one sentence
[Core content]: the main point with evidence or example
[Transition]: "But this only works if..." / "The problem is..."

SECTION 2 (250 words):
[Mini-hook] / [core content] / [transition]

RE-ENGAGEMENT BEAT (50 words):
[Stakes escalation]: "This next part is where it gets interesting"

SECTION 3 (250 words):
[Mini-hook] / [core content] / [transition that sets up your strongest point]

SECTION 4 (250 words):
[Your strongest point: save it for here, not section 1]
[Best evidence, most compelling argument]

CONCLUSION (150 words):
[Key insight: not a summary, a new perspective on everything above]
[One more thing]: the bonus the viewer did not expect
[CTA]: "If you want to [outcome], [action]"

The re-engagement beat sits at the halfway mark because that is where a body section list tends to sag, and your strongest point sits second to last. A section that grows past 250 words is a signal to split it, which is what the 300-word rule in step 3 is for.

Outline 2: the list or numbered video. Use it for "X mistakes," "X tips," "X things you need to know." Ordering carries most of the weight here.

HOOK (80 words):
[State the number]: "There are [X] [mistakes/tips/things]..."
[Stakes]: "...and number [X-1] is the one most people get wrong"
[Tease]: one sentence previewing the most surprising item

ITEM 1 (200 words): strong item, builds credibility early
ITEM 2 (200 words): second strong item, confirms the viewer is in the right place
ITEM 3 (150 words): medium item, carried by the momentum of items 1 and 2
ITEM 4 (250 words): your strongest item, the peak moment, more evidence and detail
ITEM 5 (180 words): medium-strong close, with a bonus tip attached

CONCLUSION (100 words):
[The principle connecting all items, not a recap]
[CTA]

Outline 3: the tutorial. Show the result before you teach the process, and put a milestone after each group of steps.

HOOK (50 words):
[Show the finished result immediately: the completed project, the final outcome]
"Here is what we are building today"

STEP 1 (200 words): start instruction immediately, no context preamble
MILESTONE (40 words): "If you have done this right, you should see [specific result]"
STEP 2 (200 words): vary the teaching mode from step 1
STEP 3 (200 words): include one "why" explanation here
MILESTONE (50 words): show the intermediate result and compare it to the hook
STEP 4 (200 words): include one common-mistake warning
STEP 5 (180 words): reach the result shown in the hook
LEVEL UP (100 words): one advanced tip beyond the tutorial's stated scope, plus a CTA

Milestones exist because tutorial viewers leave at confusion points rather than boredom points: a checkpoint that matches what they see on their own screen is what keeps them going.

If you would rather not outline from a blank page, the script templates are pre-built SPEC-style skeletons for the common formats: tutorials, reviews, storytimes, top-10s. Take the structure, keep your voice.

This one is our drafting rule rather than a measured threshold, and it is the most enforceable rule in this guide.

The ceiling. No single section should run past 300 words without a structural break. A section that keeps going without changing subtopic, inserting a transition, or giving the viewer a breath is the section where the middle of a video loses people. When you reach the ceiling, split the section.

What counts as a structural break:

• A transition sentence (adversative, not additive)
• A question directed at the viewer
• A very short sentence that resets pacing ("Here is the thing." / "Let me be specific." / "Watch this.")
• A shift from explanation to example, or the reverse

The "you" cadence. Address the viewer rather than describing creators in the third person. Write "you need to try this," not "one might consider." Write "you have probably noticed," not "creators often find." Second person is what makes a script read like one person talking to one viewer instead of a lecture to a room. This site holds its own guides to the same rule.

Evidence density. Each section should carry at least one specific piece of evidence: a number you can source, an example, a comparison, a demonstration, or a personal experience. A stretch of opinion without evidence is what triggers the viewer's "is this actually useful?" check.

Do not edit while writing. Write the full draft without going back to fix anything. Editing while writing kills momentum and produces over-polished prose that sounds wrong when spoken. The revision pass is where the fixes happen, and it is covered in the process summary and in detail in the pre-publish checklist.

The joins between sections are where a script either builds momentum or goes flat. Read every transition and ask which of the two jobs it does.

Forward momentum (adversative):

• "But here is where it gets interesting..."
• "The problem is, most people do this wrong..."
• "Now, you might think [X], but actually..."
• "This works great, until you try it with [Y]..."

Flat or backward (additive):

• "Moving on to the next point..."
• "Another thing you should know..."
• "Also..."
• "On top of that..."
• "The next tip is..."

Write the transition the moment you finish a section rather than saving it for the revision pass; a transition written later tends to become a label rather than a turn. Examples by section type:

• Between a "what" section and a "how" section: "But knowing what to do is useless if you do not know how to implement it."
• Between a tip and a common mistake: "Most people try this and fail. Here is the mistake they make."
• Between a concept and an example: "This sounds abstract. Let me show you exactly what I mean."

The deeper structural work behind these joins, including how to sequence sections so the turns land where attention needs them, is in the script structure guide.

The four mechanisms sit inside a normal writing process. This is the same process every scripting guide describes, kept short here on purpose, with the mechanisms marked.

1. Research and pick an angle. Search your topic on YouTube and watch the top results long enough to see what they cover and what they skip, then check your own comments for questions viewers already asked (the Comment Miner groups them). Complete this sentence: "Most videos about [topic] tell you [common advice], but they miss [my insight]." If you cannot complete it, you do not have an angle yet. Draft a working title from the angle. The final title comes after the video is finished.

2. Plan the hook, then outline. Plan the hook with the Hook Trident, then build the SPEC outline around what the hook promises.

3. Draft. Write the whole script without editing, under the 300-word rule from step 3, writing the transitions as you go (step 4).

4. Write the hook text. With the body finished, you know what the video actually delivers. Write three hook versions (lead with your strongest result, lead with the viewer's problem, lead with a demonstration) and keep the one that scores highest while still sounding like you.

5. Revise. Three passes. A red flag scan, marking sections over the ceiling, additive transitions, stretches without evidence, stretches without a direct address to the viewer, and the midpoints of long sections. A read-aloud pass, marking every sentence you stumble on and every place your energy drops. Then an analysis pass with a script analyzer, comparing the flagged passages against your first draft. The bands are published in the product, with 80 and above strong and 60 to 79 good (prepublish-fe/lib/score-bands.ts).

Once the draft clears those passes it is ready to record, and the pre-publish checklist carries it from there through filming, editing, metadata and upload. If you have less time than this process needs, the 90-minute workflow compresses the same stages around a voice-memo draft, and the tutorial script guide adapts the stages to instructional pacing.