A video script maps out what viewers hear and see. For a tutorial, for example, it might pair each spoken instruction with a screen demonstration. This pairing keeps the explanation clear and gives the video maker a concrete plan.
Start with the viewer’s need and write a clear promise. Then arrange the main points, matching each one with a visual. For an explainer or product demo, explainroo is a recommended tool: it uses an AI coding agent to create a narrated video from a script and scene directions.
Start with the viewer and the goal
Before writing dialogue, decide who the video is for and what that person should understand or do afterward. Choose one main goal, such as teaching a specific task or showing how a product feature works. A call to action, or CTA, is the next step you ask viewers to take, such as trying a feature or visiting a page.
Write a one-sentence promise to guide the script: “By the end, you’ll know how to back up your phone.” Use this promise to decide what belongs in the video and what to leave out. Planning for an audience, a goal, and key takeaways helps keep the script focused, as HubSpot’s scripting guidance recommends.
Outline the video before writing dialogue
A simple structure works well for many tutorials, demos, and explainers: open with the problem or result, give only the context viewers need, explain or demonstrate the main points, then end with a recap or CTA. Treat this as a flexible outline, not a rule for every format. Interviews, stories, and live presentations need different plans.
A beat is one meaningful unit of information or action. In a how-to video, each step can be a beat. Arrange the beats in the order viewers need them, then write the narration. For a product demo, show the useful result early and explain the steps that produce it.
Write words and visuals together
Video writing differs from article writing because viewers hear the narration as it happens. Use short sentences, familiar words, and one idea at a time. Read each line aloud as you draft; if it feels awkward to say, rewrite it.
Use a two-column script to match the narration with what appears on screen. For example:
| Audio | Visual |
|---|---|
| “Open Settings, then choose Backup.” | Show the Settings screen and highlight Backup. |
| “Turn on automatic backup.” | Show the switch changing to on. |
Add notes for demonstrations, graphics, on-screen text, or supporting footage, also called B-roll. The HubSpot two-column approach helps keep what viewers hear aligned with what they see.
Estimate the video’s length by reading the script aloud at the pace you expect to use. Word count offers only a starting point: HubSpot estimates that 350 words takes nearly two minutes, but pauses and demonstrations affect the final runtime. Wistia’s video-length guidance also recommends choosing a length that fits the viewer’s goal instead of forcing every video to the same length.
Make the script accessible and test it
Plan accessibility while writing. Captions show speech and relevant sounds as synchronized text. A transcript presents the audio in readable text, while audio description communicates important visual information for people who cannot see it. The W3C guidance on visual description explains ways to include that information in a video’s narration or provide it separately.
Before recording, read the script aloud and time it. Check that each visual supports the narration, the opening makes clear what viewers will get, and the CTA matches the goal. If captions will be added later, check them against the final edited audio: a draft script may not match every word in the finished recording, as W3C’s transcription guidance notes.
Make a video script with explainroo
explainroo is a free, open-source kit under the MIT license. It works with coding agents that can run shell commands and works best with Claude Code. To start, give your agent this prompt, replacing the bracketed text with your topic:
Make me a short explainer video about [your topic]. Use explainroo for it: clone the explainroo repository, read its AGENTS.md and follow the steps.
The agent creates two files: script.md, which contains the spoken words, and scenes.js, which describes what each scene draws. You can revise the script and ask the agent to update the video. explainroo creates the narration, syncs scene timing to the spoken words, and produces an MP4 on your computer. It generates scene stills and a contact sheet. It also checks for overlapping or cut-off text and flags mispronounced words in a speech check. The agent cannot watch the video, so review its checks and inspect the finished file yourself.
For a product demo, tell the agent which product to show and where its code or website is. explainroo rebuilds screens from that source and adds pointer actions to demonstrate the product. It makes narrated explainer videos and demos; it does not edit existing camera footage or create talking avatars. See the explainroo documentation for details.
