Planning a 20-second launch video for a SaaS feature: a simple shot list

Most early-stage SaaS teams don't need a polished brand film to announce a feature. A 15–20 second clip that shows one problem and one result usually does more work in a changelog, on LinkedIn or in an onboarding email. Here is the shot list I use before generating or recording anything.

1. Write the one-sentence promise first

"Export a client report in one click" is a promise. "Powerful new reporting" is not. If the sentence is vague, the video will be too.

2. Plan four shots, not ten

• Shot 1 (0–3s): the pain — a cluttered screen, a long manual step.

• Shot 2 (3–8s): the trigger — the button, command or setting.

• Shot 3 (8–15s): the result, held long enough to read.

• Shot 4 (15–20s): product name and where to try it.

3. Decide what is real UI and what is illustration

Screen recordings should stay real. AI-generated footage works better for the mood shots around them (a person at a desk, an abstract transition) than for the product itself — never let a generated clip imply a feature you don't have.

4. Write prompts per shot, not per video

If you use a text-to-video or image-to-video tool for the illustrative shots, give each shot its own prompt with camera movement, lighting and duration. Tools like Veo3 AI let you try several video models (Veo 3, Seedance, Kling and others) on the same prompt, which helps when one model handles motion better and another handles faces better. It is an independent platform, not Google's.

5. Check it muted and at phone size

Most feeds autoplay without sound. If the promise isn't readable on a phone with the sound off, add a caption before you publish.

What does your team use for feature announcements — screen recordings only, or a mix?