How to Write a Video Script That Converts
A video producer's guide to writing a script that performs: one message, two columns, a hook that earns the first seconds, and the right runtime.
A video script is the plan for everything a viewer will see and hear, written before anyone turns on a camera or opens the timeline. The scripts I trust fit on a page and can be read from memory, because the writer settled one question first: what is this video for? That single decision does more for the finished piece than any clever line. Most scripts I have seen stall for the same reason. Three stakeholders wanted the video to carry four messages, the script tried to serve all of them, and the viewer left in the first ten seconds.
This guide is how I write and pressure-test a marketing video script, from the one-sentence brief to the word-count math that keeps it on time. It works for a product demo, a founder story, a social ad, or a training module. The wording changes; the process does not.
Decide the one message before you write a word
Before any wording, write the one sentence you want a viewer to repeat after the video ends. One video, one job. If you cannot say it in a sentence, the script will wander and so will the audience.
I treat this as a short brief, not a formality. Name the exact viewer (a revenue leader comparing tools, not “marketers”) and the single action you want. Show the tone with two or three reference videos rather than an adjective like “friendly.” List the hard constraints up front, like the legal lines that must appear and any claim you are not allowed to make. A brief that says “show the product” is not a brief. On one SaaS demo, that instruction sent us through four re-records because nobody had decided whether the video was for a cold prospect or an existing customer. Deciding costs an hour. Guessing costs a reshoot. If you want a template for this step, our guide on how to write a creative brief covers what to lock before writing.
Write two columns: what they see and what they hear
Write the script in two columns, visuals on one side and audio on the other. This is the AV script format, and it is the single habit that separates a script from a blog post read aloud. It forces you to decide what the audience sees at every line, which is where video earns its keep.
The rule that comes with two columns: let the picture carry what the picture can carry. If the screen already shows a cursor dragging a task card into a new column, the voiceover does not say “she drags the card.” It says why that saved her twenty minutes. Words explain the reason; the image shows the action. When I edit, the lines I cut first are the ones narrating something already on screen. For more on planning the visual half, these visual storytelling examples show the technique in finished work.
Earn the first five seconds
Write the opening last and rewrite it most. It is the only line guaranteed an audience, because most of the people who start a marketing video never reach the end. Wistia’s look at millions of videos found people watch only about half of a short marketing video, and the steepest fall comes right at the start.
I write three to five different openings for every script and pick the one that survives being read cold. A sharp number works (“Teams lose a full day a week to hunting for files”). So does a named problem your viewer already feels, or a question they would answer yes to without thinking. What does not work is a slow build. Save the logo and the mission statement; open on the viewer’s world. If the hook is the part you find hardest, we wrote a full breakdown on how to create video hooks that stop the scroll.
Build the body on five beats
Once the hook lands, the body of a marketing script runs on five beats in order: hook, problem, solution, proof, and call to action. It is a cousin of the AIDA framework (attention, interest, desire, action) and of the classic three-act structure, and it holds up because it moves a viewer from a feeling to a reason to a next step with no gap.
Map your one message onto each beat before you write prose. Problem: show the cost of doing nothing. Solution: introduce the product as the answer to that exact problem, not a feature tour. Proof: one number, one short demo, or one customer sentence, never a wall of them. Then the ending. Keep each beat earning its place. If a line does not push toward the next beat, it comes out.
| Beat | Runtime | Job |
|---|---|---|
| 1. Hook | first 3 to 5 seconds | Name the viewer's world and stop the scroll |
| 2. Problem | next 15 to 20% of runtime | Show the cost of leaving things as they are |
| 3. Solution | middle 30 to 40% | Introduce the product as the answer to that exact problem |
| 4. Proof | following 20 to 25% | One number, one short demo, or one customer line |
| 5. Call to action | last 5 to 10 seconds | A single next step matched to the viewer's intent |
Structure built from this guide, not a dataset. The percentages are rough runtime shares to plan against, not fixed rules.
Write for the ear, not the page
Read every line out loud before you call the script done. Words that look fine on paper trip the tongue on camera, and the viewer hears the difference even if they cannot name it. On anything longer than thirty seconds we run a table read, writer and on-camera talent in a room, going through it two or three times until the awkward corners smooth out.
Writing for the ear means contractions and short sentences, one idea to a line. Swap “utilize” for “use” and “in order to” for “to.” If a sentence needs a breath in the middle, break it in two. A useful checkpoint: a person reads roughly two words per second, so a line that takes three breaths to say will feel long on screen. HubSpot’s video team treats that table read as the step that catches what silent editing misses.
Match the runtime to the message
Decide the length before you write, then hold the script to it with word count. The math is plain and it keeps the edit on time: finished voiceover runs about 125 to 150 words per minute, so a 30-second ad is roughly 65 to 80 words and a two-minute explainer is around 300 to 350. HubSpot’s team clocked a 350-word script at close to two minutes, and Vidyard’s script timer will do the sum for you if you would rather not count by hand.
Length is not a free variable. Wistia’s look at millions of videos shows engagement holding fairly steady up to about five minutes, then sliding once a video runs past it. Longer is not automatically worse, but every extra sentence has to earn the attention it spends. If you are writing an explainer and unsure where to land, we went through how long an explainer video should be with the numbers by format.
End where the viewer actually is
Write the call to action to match where the viewer sits in their decision, not where you wish they were. A cold viewer who just met your brand will ignore “book a demo,” because you have asked for a commitment they are nowhere near making. The ending has to fit the temperature of the person watching.
For an awareness video, ask for something small: follow for more, or watch the next one. For a consideration-stage video, point to the deeper asset: see the full demo, or compare the options. For a decision-stage video, be direct: talk to the team, or see how it works with your own data. One ending, one action. Two calls to action split the viewer’s attention and you lose both.
Cut the script to the platform before you shoot
Write the script for the platform it will live on before the shoot, not after. A YouTube script and a TikTok script are different pieces of writing, not the same words resized. TikTok wants the hook inside the first two seconds and rewards a spoken, native tone. LinkedIn tolerates a slower, more considered open. A vertical 9:16 frame changes where on-screen text can sit and how tight the shots need to be.
Note the platform decisions in the script itself: the aspect ratio, and where captions and text overlays sit (most feed video is watched on mute, so the words have to work with the sound off). Write for one master version and resize later, and you inherit awkward crops and buried hooks. Planning format up front is part of the same discipline as the explainer video script, where structure and format get decided together.
One hour on the script saves a week in the edit
A good script is the least expensive fix in the whole production. An hour spent deciding the one message and reading the lines aloud saves days of re-records and a re-edit nobody planned for. Write the two columns and protect the opening seconds, then hold the runtime to the word count. Do that, and most of a strong video is finished before the camera rolls.
If the writing is the part your team keeps stalling on, Moonb works as an embedded creative team that handles the scripting and the production alongside your marketers, with a Creative Director who owns the one-message call from the brief through the final cut. Either way, the script comes first.
Frequently asked questions
Count on about 125 to 150 words of finished voiceover per minute. A 30-second ad runs roughly 65 to 80 words, a 60-second social spot around 130 to 150, and a two-minute explainer near 300 to 350. Write to the word count, not the page count, since a script heavy on silent on-screen visuals will run longer than the words alone suggest.
Not word for word. For interviews I write a question list plus the two or three points each answer has to land, then let the person speak in their own words. A fully scripted talking head usually reads stiff on camera. Script the structure and the must-say lines, and leave room for the delivery to sound like a real human.
A script is the words plus shot notes: what is said, and in the second column, what is seen. A storyboard turns those visual notes into drawn frames, panel by panel. You write the script first, then storyboard it before the shoot. A plain talking-head video can skip the storyboard; anything with animation or lots of shot changes is worth boarding.
