Open YouTube Shorts for five minutes and you will see the pattern fast. The videos that hold you are not always the most polished. They are the ones that tell you, in the first few seconds, why you should stay. The ones you swipe past are not always bad ideas. They are often good ideas that took too long to get to the point. That is the whole game of a 60-second script: say less, say it faster, and give the viewer a reason to stay every ten seconds or so.
AI can help a lot here, but it cannot save a script with no point. If you ask a chatbot to “write a viral script” with no topic, no audience, and no payoff, you will get a generic script that sounds like every other generic script. The creators who get real value from AI use it differently. They bring the idea, the angle, and the proof — a real fact, a real demo, a real before-and-after — and they let AI help with hooks, structure, tightening, and variations. That division of labour is what this guide is about.
One honest note up front: no script formula guarantees views. YouTube does not publish a recipe for virality, and any tool that promises one is selling hope, not a method. What a good script can do is raise your odds. It can improve retention, which is one of the signals Shorts pays attention to, and it can make your video easier to understand, share, and comment on. That is a worthwhile goal, and it is fully in your control.
First, decide what the viewer gets in 60 seconds
Before you write a single line, answer one question in plain words: when this Short ends, what will the viewer know, feel, or be able to do that they could not do before? If you cannot answer that in one sentence, your script is not ready. “How to grow coriander in a pot” is a promise. “Plants are amazing” is not. “Three mistakes that kill seedlings” is a promise. “Gardening tips” is a topic, not a promise.
That one-sentence promise becomes your hook, your title, and your test. Every line after it should either prove the promise, show it, or build curiosity toward it. If a line is funny but pulls away from the promise, cut it or save it for another video. Sixty seconds is not the place for a slow warm-up. You can be warm, funny, and personal — just do it while you are already delivering value, not before you start.
It also helps to pick one viewer. Are you talking to a beginner who has never grown anything, or to someone who already grows herbs and wants faster results? A script for a beginner needs simpler words and more showing. A script for an experienced viewer can skip the basics and jump to the twist. AI is useful here too: paste your draft and ask it to rewrite the same script for a complete beginner, then for an intermediate viewer, and compare which version is clearer. You will often find lines you assumed were clear that are not.
The 60-second skeleton that most strong Shorts follow
You do not need to copy this skeleton word for word, but it is a useful check. If your script is missing a piece, you will feel it when you edit: the video will sag in the middle, or the ending will feel flat, or viewers will drop off before the payoff.
| Section | Time in a 60s Short | Job of this section | What good looks like | Common slip |
|---|---|---|---|---|
| Hook | 0–3 seconds | Stop the swipe and state the payoff | Bold on-screen text plus a spoken line that matches it | Starting with “Hey guys, welcome back” |
| Promise and setup | 3–10 seconds | Show what will happen and why it matters | One clear setup shot or example | Listing three disclaimers before the demo |
| Payoff beat 1 | 10–25 seconds | Deliver the first real result | Viewer learns or sees something new | Talking about the result instead of showing it |
| Payoff beat 2 / twist | 25–45 seconds | Deepen, contrast, or surprise | A comparison, mistake, or unexpected outcome | Repeating beat 1 with different words |
| Close and loop | 45–60 seconds | Land the point and invite the next action | A recap line, a question, or a loop back to the hook | A long outro that kills retention |
Notice how little time the hook gets. Three seconds is not a lot, but it is enough for one strong line and one strong visual. “Stop watering your coriander like this” with a shot of a wilting plant beats a 10-second intro about your channel every time. Your channel intro can live in your profile, in a pinned comment, or in the last two seconds. It does not need to open the video.
The middle beats matter just as much. A lot of Shorts open well and then lose people around second 20, because the creator keeps talking without showing anything new. Plan a visual change or a new piece of information every 10 to 15 seconds. That could be a new angle, a close-up, a text pop, a result, or a question on screen. You are not trying to be frantic. You are trying to avoid dead air, where the viewer has already understood the point and is waiting for you to move on.
Writing with AI, step by step — without sounding like AI
Here is a workflow that keeps you in charge and uses AI for what it is actually good at: speed and variations.
Step 1: Give AI a tight brief, not a vague wish. Instead of “write a Short script about plants,” try: “Write a 60-second Short script for beginners growing coriander in a pot. Hook in 3 seconds. Three mistakes, each shown in under 10 seconds. Friendly Hindi-English mix, short sentences, no jargon. End with a question viewers can answer in comments.” The more specific the brief, the less generic the output. Include your audience, your language mix, your length, and your ending. If you skip those, AI will guess, and its guess will be average by design.
Step 2: Ask for 10 hooks, then pick one and rewrite it in your voice. Hooks are where AI shines, because you can generate a batch fast and judge them like headlines. Ask for different types: a mistake hook (“You are watering it wrong”), a curiosity hook (“This pot grew twice as fast — here is why”), a bold claim hook that you can actually prove, and a question hook. Then take the best one and say it out loud. If it sounds stiff when you speak it, it will sound stiff on camera. Rewrite it until it sounds like something you would actually say to a friend.
Step 3: Draft the beats, then cut a third. Ask AI to turn your hook into a beat-by-beat outline with timestamps, then read it as a viewer. Cross out any line that repeats an earlier point, any line that is setup for setup's sake, and any line that only makes sense if the viewer already knows your channel. Short scripts get better by cutting, not by adding. A 75-second draft that you cut to 58 seconds will almost always feel tighter than a draft that was exactly 60 seconds from the start.
Step 4: Write the on-screen text separately. Viewers often watch Shorts with sound off first, or with sound low. Your on-screen text should tell the story on its own: hook text, step labels, results, and the final question. Ask AI to suggest short on-screen lines — five words or fewer per pop is a good rule — and then check them against your video. If the text and the voiceover say the exact same thing for 60 seconds, one of them is redundant. Let the text highlight, and let the voice explain.
Step 5: Read it aloud twice before you record. This is the step people skip, and it shows. Read the script out loud, time it, and mark where you stumble, where you run out of breath, and where you get bored. Those are edit points, not personality flaws. Fix the line, shorten the sentence, or swap in a visual. A script that reads well on paper but trips you up out loud will cost you retakes and, worse, a flat delivery on camera.
Hook formulas that work — and how to avoid the tired ones
Hooks work when they create a gap the viewer wants to close. “I tried this for 30 days” creates a gap: what happened? “Three mistakes” creates a gap: am I making one? “Do not do this” creates a gap: why not? The formula is not magic. The gap is.
What makes a hook tired is not the formula, it is the lack of proof behind it. If every video opens with “You won't believe what happened,” viewers learn to ignore it. If your hook promises a result, show a glimpse of that result in the first seconds, or at least show the setup clearly enough that the viewer trusts you will deliver. A hook plus a visual promise beats a hook alone.
Try this test on your next five hooks. Cover the rest of the script and ask: would a stranger understand what they will get, and would they care enough to wait 10 seconds? If the answer is no, the hook is either too vague or too focused on you instead of the viewer. “My journey” hooks can work once you have an audience that cares about your journey. For growth, “your result” hooks usually work better: what the viewer gets, what mistake they avoid, what they can try today.
Retention is not a trick. It is pacing, clarity, and payoff.
Creators sometimes treat retention like a set of hacks: faster cuts, louder music, more text pops. Those tools can help, but they cannot fix a script that has nothing to say in the middle. Real retention comes from a simple rhythm: promise, proof, new proof, payoff. When viewers feel progress — they learned one thing, then another, then saw the result — they stay. When they feel stuck on the same point, they leave, no matter how fast the cuts are.
A practical way to check pacing is to watch your edit with the sound off and note the timestamps where you would swipe away. Those timestamps are usually where a beat ran long, a visual repeated, or a payoff was delayed for a joke that did not land. Fix those spots first. Adding a sound effect or a zoom will not save a 12-second stretch where nothing new happens.
Endings deserve the same care. A 60-second Short should not end with a 10-second “like, share, subscribe, comment, and check out my other videos” block. Pick one action that fits the video. If you asked a question, let the question be the call to action. If the video naturally leads to a next video — “tomorrow: the plant that beat this one” — that teaser can be your close. And if your opening line can loop naturally into your ending, you have a bonus: viewers who watch again, or who do not notice the loop, give you extra watch time without any extra work. Use loops only when they feel natural, though. A forced loop feels like a trick, and viewers notice.
A worked example: from rough idea to tight script
Imagine your idea is “coriander growing.” A rough AI first draft might open with a greeting, explain what coriander is, list general tips, and end with a generic subscribe ask. That is a topic, not a script. A tighter version, after you apply the steps above, might look like this:
Hook (0–3s), on screen: “Stop drowning your coriander.” Voice: “If your coriander keeps dying, you are probably watering it wrong.” Setup (3–10s): show a healthy pot and a struggling pot side by side. “Same seeds, same sunlight. One difference.” Beat 1 (10–25s): show the mistake — daily heavy watering — and the fix — water only when the top soil feels dry. Beat 2 (25–45s): show the second mistake — a pot with no drainage — and the quick fix. Close (45–60s): show the healthy pot again. “Try the dry-top test today. Which mistake were you making — one or two? Tell me below.”
That script is not longer or fancier. It is clearer. It has a viewer, a promise, two concrete beats, and an ending that invites a comment without begging for one. You can reuse the same shape for cooking, tech, fitness, or study Shorts. Change the topic, keep the discipline: one promise, fast proof, one more proof, clean close.
For more tool ideas to pair with this workflow, see Best Local AI Video Models That Run on 8GB VRAM GPUs for local clips you can cut into Shorts, and Best Free AI Video Generators if you need footage while you learn scripting.
What to check before you publish
Run this quick checklist on every script, even when you are in a hurry. Is the hook under 3 seconds and matched by on-screen text? Can a viewer with sound off follow the main point? Is there a new visual or fact every 10 to 15 seconds? Does the ending ask for one thing, not five? Have you read it aloud and timed it? If any answer is no, fix that one thing before you record. Small fixes at script stage are cheap. Fixes after you have shot, edited, and captioned are not.
Also check your facts. AI can invent a statistic, a date, or a “study” that sounds real. If your Short makes a factual claim — a yield, a price, a health point, a record — verify it from a primary source before you publish. A Short that gets corrected in the comments loses trust faster than a Short that was simply boring. For Shorts format rules, lengths, and best practices, check YouTube's own creator guidance before you rely on a third-party summary, since platform guidance changes.
Sources and official pages to check: YouTube Help: Create a Short, YouTube Creators hub, ChatGPT from OpenAI.
Your next Short starts with one sentence
Do not open AI and ask for a full script yet. First, write your one-sentence promise on paper: what will the viewer get in 60 seconds? Then ask AI for 10 hooks for that promise, pick the one you can prove on camera, and build the beats around it. Record a rough version, read it aloud, cut a third, and publish the tighter cut. That habit — promise first, AI second, cutting last — is what turns AI from a generic script machine into a real writing partner.
Try it on your next idea tonight. Write the promise, draft the hooks, and see how much clearer the script feels before you ever press record.
How long should a YouTube Short script be for 60 seconds?
Aim for roughly 130 to 160 spoken words for a natural pace, but timing matters more than word count. Read your script aloud and time it. If you rush to fit 180 words, cut lines instead of speeding up, since rushed delivery hurts clarity and retention.
Can AI write a Short script that actually performs well?
AI can draft hooks, outlines, and variations quickly, but performance still depends on your idea, proof, visuals, and delivery. Use AI for structure and options, then rewrite in your own voice, verify any factual claims, and test the script by reading it aloud before recording.
What makes a good hook for a 60-second Short?
A good hook states a clear payoff in the first 3 seconds and matches it with a strong visual or on-screen text. It creates curiosity by naming a mistake, a result, or a question the viewer wants answered, and it promises something the video actually delivers.
Why do viewers drop off in the middle of a Short?
Viewers usually leave when a beat runs too long without new information or a new visual. Plan a fresh payoff, comparison, close-up, or question every 10 to 15 seconds, cut repeated lines, and show results instead of only talking about them.