How to Write a Video Script That Keeps People Watching, With a 60-Second Template

Hook, beats, B-roll and a payoff: one structure for YouTube, TikTok and Shorts, shown on a finished example

An establishing image, the kind of visual a script's first beat asks for. AI-generated, made in Ziztu.

The fastest way to write a video script is to write the last line first, then the hook, then the middle. A script that knows where it ends is easy to cut. A script that does not will run long and lose the viewer.

Disclosure: this story was written with the assistance of an AI writing program. I build Ziztu, which has the script tools I link to, so weigh that.

I started Ziztu after my own attempts to direct high-quality anime and micro dramas with other AI tools kept falling short, and I think the lesson holds for any video: the script decides more than the tool does. This is the structure I would use for a short video, and it works for YouTube, TikTok and Shorts.

Step 1: Answer five questions before line one

Fill in a small card first:

• Title: what the video is called.
• Goal: what the viewer can do or say after watching, in one sentence.
• Audience: who it is for.
• Length and shape: about 40 seconds vertical, or 5 minutes wide.
• Call to action: the one thing you want them to do next.

If the goal needs two sentences, you have two videos.

Step 2: Know how many words you have

Spoken video is short on words. Use these as a guide, the same figures Ziztu’s YouTube script generator uses:

• About 1 minute: around 150 words.
• About 3 minutes: around 420 words.
• About 5 minutes: around 700 words.

Write to the number, then trim. Scripts that read well on a page are often too long when spoken.

Step 3: Write the hook

The first ten seconds should open a question or a surprise and promise what the viewer will get. Drop “hey guys” and “in this video I’m going to”. Three hook shapes that work:

• The contradiction: “The cheapest option is rarely the cheapest.”
• The open loop: “There’s one setting that fixes this, and most people turn it off.”
• The result first: show the finished thing, then say how you got there.

Step 4: Break the middle into beats

A beat is one idea with one thing on screen. For a 60-second video, plan four to six beats of about ten seconds each. Use three columns, one row per beat:

• Visual: what the viewer sees (B-roll, a screen, a person, a diagram).
• Voiceover: the words, written as you would say them.
• On screen: the text caption, in five words or fewer.

If you cannot name the visual for a beat, the beat is a lecture. Cut it or find the picture.

A finished 60-second script

Topic: how to make cold brew coffee at home. Roughly 150 words.

Get Xnirlin’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

0:00 to 0:05. Visual: an iced coffee glass, ice cracking. Voiceover: Cold brew costs a few dollars at a cafe and far less at home. On screen: Far less at home.

0:05 to 0:15. Visual: coarse grounds poured into a jar. Voiceover: Start with coarse grounds. Fine ones turn bitter and clog the filter. On screen: Step 1: coarse grind.

0:15 to 0:25. Visual: water poured over the grounds. Voiceover: Mix one part coffee with about eight parts cold water. Stir until nothing floats. On screen: Step 2: one to eight.

0:25 to 0:35. Visual: the jar in a fridge, a clock turning. Voiceover: Leave it in the fridge for twelve to eighteen hours. Longer is stronger, not better. On screen: Step 3: wait.

0:35 to 0:48. Visual: pouring through a strainer into a bottle. Voiceover: Strain it twice. The second pass takes out the fine stuff that makes it cloudy. On screen: Step 4: strain twice.

0:48 to 1:00. Visual: the glass again, milk swirling in. Voiceover: Pour it over ice, add milk or water, and you are done. Follow for one coffee trick a week. On screen: Follow for more.

Notice: every beat has a visual, every voiceover line is under 15 words, and the last line does one thing. The recipe ratios are one common version, so adjust to your taste.

Step 5: Plan the B-roll

Next to each beat, write what you will film or find. If you are making a faceless video with stock or generated clips, say so in the visual column, and keep each shot short so that you only need a few seconds of footage. A single clip per beat is a good limit.

Three short clips, each one a beat that could sit in a script. AI-generated, made in Ziztu.

Step 6: Edit the script down

Read it aloud with a timer. Cut any sentence that repeats the screen. Cut the first sentence if the second one is the real start. Then check the last line: it should be the call to action, not a summary.

Check it before you shoot

• Does the goal fit in one sentence?
• Does the hook promise something in ten seconds?
• Does every beat have a visual?
• Is the word count close to the length?
• Does the last line do one thing?

Tools for the draft

Ziztu’s YouTube script generator writes three title ideas, a hook, sections with what to say and what to show, and an outro that pays off the hook. For short vertical videos, the TikTok script generator writes hooks and shot ideas. Both give 3 scripts a day with no account, and 20 with a free one. The video script template has the card and the beat table above, ready to fill in. The numbers in any generated script are examples, so check any real figure before you record.

Write the first beat

Start free on Ziztu: draft your script in the YouTube script generator, no card. Type your idea, pick a length, and rewrite the result in your own words. (These links open getziztu.com, outside Medium.)

Images in this story were created with an AI image and video tool (Ziztu).