How to Write a YouTube Script That Keeps People Watching

How to write a YouTube script

Write a YouTube script by starting with the promise of the video, not the intro. Decide the one thing a viewer will be able to do or understand by the end, write the hook that states that promise in the first 15 seconds, then outline three to five sections that deliver it in order. Draft the whole thing at roughly 130 to 150 words per minute of finished video, read it out loud, and cut every sentence that does not move the viewer toward the promise.

How long should a YouTube script be?

Multiply your target video length by 130 to 150 words per minute. An 8 minute video is roughly 1,000 to 1,200 scripted words. A 12 minute video is closer to 1,600 to 1,800. Subtract time for b-roll, demos and silence, because those minutes need no words. Always draft long and cut, never draft short and pad.

The words-per-minute figure is the most useful number in scripting and almost nobody applies it. Most people write what feels like the right amount of material, then discover in the edit that their "10 minute video" is 19 minutes. Count the words. It takes ten seconds and removes the guesswork.

Two adjustments. If your video is demo-heavy, where you're talking over a screen recording and pausing while things load, drop the word count by 20 to 30 percent and let the visuals carry the time. If you're a fast talker, and most people who watch their own footage discover they are, use the top of the range.

What is the best structure for a YouTube script?

Hook, promise, delivery, close. Open with 10 to 15 seconds that state what the viewer gets. Confirm it is worth staying for. Deliver the substance in three to five clearly separated sections, each of which could stand alone as a clip. Close with one specific next action. Skip the channel intro, the greeting, and the "before we start" section entirely.

Here is the same structure with target lengths, using an 8 minute video as the example.

The part people get wrong is section order. There's a strong instinct to build up: context first, then background, then the good part at minute six. That works in a lecture where the audience has committed to staying. On YouTube it's why retention falls off a cliff at 0:40. Put the strongest material first and let the context arrive underneath it, in the sentences where you explain why the strong thing is true.

The other reason to keep sections self-contained: each one becomes a clip. A section that references "what I said earlier" can't be posted vertically without a rewrite. Write them as if each will be watched alone, because some of them will be.

How do you write a hook for a YouTube video?

Say the most specific true thing you have in the first sentence. Name the outcome, the number, or the disagreement. Cut every word before it: no greeting, no channel name, no "in this video." A working test is whether the first sentence would still make sense as the video's title, and whether someone who disagreed would want to argue with it.

Weak hooks share a pattern: they describe the video instead of starting it. "Today we're going to be talking about email deliverability" is a table of contents read aloud. Compare: "Your emails are landing in spam because of a DNS record you set up in 2021 and never checked." Same video. The second one has a person in it who's about to be either right or wrong, and that's what holds a viewer.

Three hook shapes that reliably work, all of which need real material behind them:

• The contested claim. "Most product demos lose the deal in the first two minutes, and it's not because of the product." The viewer either agrees and wants the proof, or disagrees and wants to catch you being wrong. Both watch.
• The concrete result. "This is the exact 6 step process we used to cut render times from 40 minutes to under 3." Specific, checkable, and it sets the promise.
• The open loop with a deadline. "There's one setting in here that will break your whole export, and I'll show you where in about four minutes." Only use this when the payoff lands, and land it early.

Write the hook last. Writing it first is how you end up with a hook that promises a video you did not actually make, which is the fastest way to teach an audience that your thumbnails lie.

Should you script your YouTube videos word for word?

Script word for word when precision matters: tutorials with exact steps, anything with legal or technical claims, and the first 30 seconds of every video regardless of format. Use bullets for interviews, reactions, and personality-led content where stiffness costs more than a fumbled sentence. Most creators land on a hybrid: fully written hook and close, bulleted middle.

The argument against full scripts is that they sound read, and they usually do, but that's a delivery problem rather than a scripting one. Three fixes. Write in your speaking voice: contractions, fragments, sentences short enough to say in one breath. Read the draft out loud and rewrite anything you stumbled on, because a stumble is almost always the script's fault. And don't memorize the script, learn it, then say it in your own words from the bullet version.

The 30 second rule is worth defending though. Even hosts who improvise a full hour should write the opening verbatim. It's the part you get exactly one attempt at with a new viewer, and improvised openings drift into throat-clearing every time.

Research the script before you write it

The gap between a script that performs and one that dies is usually not writing quality. It's topic fit. You wrote the video you wanted to make instead of the one your audience was already trying to find.

Two cheap sources fix this. YouTube's search bar autocomplete tells you the exact phrasing people use, and those phrases belong in your title, hook and section headings, in their words rather than yours. Second, the comments on the top three videos in your topic are a list of everything those videos failed to answer. Before scripting anything with real production budget behind it, it's worth 20 minutes to see what people are already saying about the topic across social and the wider web, because the objections and the vocabulary you find there are the objection-handling section and the hook, pre-written by your audience.

Take the recurring question from the comments and make it a section. Take the recurring complaint and make it the hook. Least glamorous part of scripting, highest return.

The script is also your title, description and clips

A finished script has already done most of the work for the rest of the launch, which most teams don't act on. Your promise line is the raw material for the title, and YouTube titles max out at 100 characters. If the promise doesn't fit in 100 characters, it isn't sharp enough yet. Turning that line into something that survives the mobile grid is its own craft, and how to write a YouTube title covers the 60-character rule and the native A/B test most creators never switch on.

The description follows the same logic. You get up to 5,000 characters, but only roughly the first 157 show on desktop before the "Show more" cut, so those first two lines have to restate the promise and nothing else. The rest becomes timestamps (your section list, already written), links and the fuller summary. Pulling that out of the script beats the one-line description you type at 11pm because the render finished late. A YouTube description generator does that pass from the script or transcript directly, and the numbers behind the box are worth knowing if you have never checked: how long a YouTube description should be covers what actually gets read.

Then the clips. Each section you wrote as a standalone unit is a vertical cut, and each needs its own hook because it will be seen by people with no context. That's the same rewriting job, five times over, in a different format. A youtube script generator handles the first draft from your source material, whether that's an outline, a blog post you already published, or a transcript from a video you recorded off the cuff, and the TikTok script version reshapes each section for vertical. You keep the judgment work: the angle, the examples, the actual opinion. The retyping is what gets compressed.

The mistakes that show up in every weak script

• The warm-up. "Hey guys, welcome back to the channel, hope you're having a great week." Forty seconds of nothing. Cut it and start at the first real sentence.
• Writing for the eye instead of the ear. Long subordinate clauses read fine and speak terribly. If you can't say it in one breath, split it.
• Saving the best part for the end. Nobody's there at the end. Front-load.
• Four calls to action. Subscribe, like, comment, click, join. Pick the one you actually want and ask for it once.
• No objection section. The smart viewer's counterargument doesn't disappear because you ignored it. It shows up in the comments and drags the whole thread there.

A YouTube script template you can use today

Open a doc and fill in these seven lines before you write a full sentence of the script. Promise: what the viewer can do after watching, in one line under 100 characters. Hook: the most specific true sentence you have. Stakes: why this matters now. Sections 1 to 4: one idea each, with the example or demo named. Objection: the thing a smart viewer will push back on, and your answer. Close: one action.

If those seven lines are strong, the script writes itself and comes in near your word target on the first pass. If they're weak, no amount of drafting saves it, and you'll find that out at minute six of the edit instead of minute two of the outline. Once the script exists, the rest of the distribution comes off the same source. The video repurposing workflow covers turning the finished recording into the posts and the newsletter, and the repurposing strategy playbook covers the cadence that keeps one video working for a month.