How to write a YouTube script (with a template)

A YouTube script is what you'll say on camera, in order, with notes for what's on screen. It isn't an essay: it's written to be heard once, at the speed you talk, by someone who can leave at any second. Most scripts that hold people follow the same shape:

1. Hook. The first few sentences: what the video gives and why to keep watching. It has to match the title and thumbnail that got the click.
2. Setup. Who it's for, what they'll have by the end, and only the context they need to follow. A few sentences, not a minute.
3. The content, in parts. Three to five parts, each making one point. Each opens with a line that says what it is (“Mistake two: the oven isn't hot enough”) and ends on what it means.
4. Payoff. The answer, the result or the reveal the title promised. Don't let the video trail off before it.
5. Call to action. One ask, after the payoff: watch the next video, subscribe, or get the thing you mentioned. One, not five.

To write one:

1. Write the promise in one sentence. If the title is “3 mistakes that make homemade pizza soggy”, the promise is: by the end, you'll know all three and how to fix each one.
2. Outline the parts, a line each. Cut any part that doesn't serve the promise, and put a strong one early.
3. Draft the parts, then the hook and setup, then the payoff and call to action. A hook is easier to write once you know exactly what the video delivers.
4. Read it out loud with a timer. Rewrite whatever you stumble on, and cut until it's as long as the topic needs.

Both work. Which one suits you depends on the video and on how you talk:

Most creators land in between: script the hook, the line that opens each part, the payoff and the call to action word for word, and bullet-point what goes between. The lines that have to be exact are exact, and the rest sounds like you. If you go off script and it's better, keep it: the script is a plan, not a contract.

A reader can go back over a sentence. A viewer hears it once, at your speed. Write for that:

• Short sentences, one idea each. If you'd need a breath in the middle, split it in two.
• Write how you talk. Contractions, plain words, “you”. Say “one should utilize a higher temperature” out loud and you'll never write it again.
• Say where you are. “Mistake two.” “Here's the fix.” “Last one.” Signposts that look clumsy on a page keep a listener with you.
• Repeat the name, not “it”. On a page, “it” points back to something the reader can still see. In a video, it points at something they might have missed. Say “the dough” again.
• Numbers the way you'd say them. “About five hundred degrees”, not “approx. 500°F”, and round wherever exact doesn't matter.
• Cut the throat-clearing. “Hey guys, welcome back, in today's video I'm going to…” only delays the hook. Start with the hook.
• Note what's on screen in brackets: [close-up: the pale base], [text: 500°F]. When the picture shows something, the script doesn't have to say it.
• Read it aloud before you film, and rewrite anywhere you trip.

The National Center for Voice and Speech puts the average rate of speech for US English speakers at about 150 words a minute, and ACX, Audible's audiobook marketplace, says most narrators read about 9,300 words an hour (about 155 a minute). So 150 words a minute is a fair place to start. The slower and brisker columns are a step either side, to show the range: they're our assumptions, not measured averages.

Three things change the count:

• Your own pace. Read a page of the script the way you'd film it, and time it: that number beats any average.
• Time with no talking. A demo, a screen recording, a pause on a result: each adds time and no words. A hands-on video can need far fewer words than the table says.
• The edit. Cutting pauses and retakes makes the finished video shorter than the read-through.

Write for the length the topic needs, not a word target: padding a script out to a length adds exactly the parts people skip. The free teleprompter shows your script's word count and how long it takes to read at your pace, so you know the length before you film.

YouTube says its systems use both average view duration and average percentage viewed as signals for ranking, then likes and post-watch surveys to judge whether viewers enjoyed the video. YouTube Studio shows you where they leave. In a video's Analytics, the audience retention report highlights key moments (generally on videos at least 60 seconds long with at least 100 views):

• Intro: the share of viewers still watching after the first 30 seconds. When it's low, YouTube suggests making the thumbnail and title reflect the video better, or changing the first 30 seconds and trying different styles.
• Top moments: where almost no one left. If they come late in the video, YouTube suggests moving compelling content earlier, because the audience shrinks as a video goes on.
• Spikes: parts people rewatched or shared: a strong part, or one that wasn't clear the first time.
• Dips: where viewers skipped ahead or left.

What that means for the script:

• Deliver on the title in the first 30 seconds. Show the result, state the answer you're about to prove, or start the story at its most interesting point. For openers that work, see short video hooks or try the free hook generator: a long video's first line follows the same rules as a Short's.
• Say what's coming. “Three mistakes, and the third is the one most people never notice” gives people a reason to stay to the end.
• Put strong parts early. Fewer people are left by the end, so don't save all your best material for it.
• Re-hook at each part. The line that opens a part should make the next minute sound worth it.
• Cut detours. Every tangent is a likely dip.
• Script the last 20 seconds. End screens can only go in the last 5 to 20 seconds of a video at least 25 seconds long, so write a closing line that points at the video you'll show there, after the payoff.

The title and thumbnail make the promise; the script keeps it. For the other half, see how to make a YouTube thumbnail; the free YouTube thumbnail maker starts from a frame of your video and adds a bold title.

Copy this, fill in your own lines and cut what you don't need. The example column is for a made-up video, “3 mistakes that make homemade pizza soggy”, about 8 minutes long. The times are a guide for a video that length, not rules.

Under each part, note what's on screen and whether the part could stand alone as a Short. The parts are your chapters too: list their start times in the description, from 0:00, at least 3 in order, each at least 10 seconds long, and YouTube shows them on the progress bar. The free YouTube timestamp generator makes that list while the finished video plays.

Reading to a lens is a skill, and a few habits make it much easier:

• Keep the words near the lens. The farther the text is from the camera, the more your eyes visibly drift. Put it right under the lens, keep the lines short so your eyes don't sweep side to side, and sit a little farther back, where small eye movements don't show.
• Set the pace to how you talk, not how you read. You read silently faster than you speak. Start the scroll at your speaking pace and adjust it over the first takes.
• Mark pauses and stress. A line break where you breathe, a slash before a big line, capitals on the word you'd lean on. You're writing a performance, not a document.
• Rehearse twice, out loud. The first read finds the tongue-twisters; the second makes the words yours.
• Film one part per take, with a beat of silence before and after: clean edges make the edit, and any Shorts, easier.
• Learn the hook. If you know the first two lines by heart, you can say them straight to the lens, where eye contact matters most.

The free teleprompter does this in your browser. Paste the script, press Start, and after a 3-2-1 countdown the text scrolls at your words-per-minute pace, big, with a reading line; change the speed and text size as it runs, mirror it for teleprompter glass, or go full screen. “Pause when I pause” listens to your microphone's level on your device (no speech recognition) and only scrolls while you're talking, and camera mode records you to a file you download: an MP4 where the browser can record one, otherwise WebM.

A Short can run up to 3 minutes, but most are far shorter, and the script works the same way at a smaller scale. At 150 words a minute, 60 seconds is about 150 words (130 to 170 across the paces above), and 30 seconds about 75. In a Short:

• Skip the setup. The hook is the first line and the first frame.
• Make one point. A Short with three points is a long video in a hurry.
• End on the payoff. The last line is the answer or the result, not “thanks for watching”.

How long a Short should actually run is in how long a short video should be.

Planning the long video so it cuts into Shorts. If the long video is built from parts that stand on their own, you don't need separate Shorts scripts. Write each part so a stranger could start there:

• Open each part with a line that makes sense alone. “Your oven isn't hot enough, and here's how to tell” works on its own. “So, like I said earlier” doesn't.
• Don't refer back. Name the thing again instead of saying “the trick from before”.
• End each part on its point. A sentence that lands is a clip's ending.
• Keep each part's core to 30 to 60 seconds of talking, even if the whole part runs longer with the demo.
• Mark the candidates in the script, so you know which moments to cut once the video is up.

Once the video is up, YouTube Studio's Clips and Shorts tool makes a Short from a part you select in the transcript or on the timeline, and links the full video as the Short's related video (unless the Short is private, unlisted or made for kids). YouTube says you won't be penalized for using your own content in a Short.

By hand, each Short still means finding the moment, trimming it, reframing it to 9:16 and captioning every word. Clyper does those steps from the YouTube link: it reads the transcript, picks moments that make sense on their own, cuts each where a sentence ends, scores each clip's hook, frames whoever's talking and captions every word, up to one clip for every 2 minutes of video. A video built from self-contained parts gives it clean moments to find. More on YouTube to Shorts, and if the channel is new, see how to start a YouTube channel.