1. Complete the brief before prompting
Write the audience, offer, problem, desired action, video length and tone before asking an AI tool for scenes. A useful starting line is: “Create a 20–30 second vertical video for [specific audience] about [topic].” Replace every bracket with real information. Avoid asking for a “viral video”; that does not explain what the viewer should understand or do.
2. Choose the right prompt structure
Use an explainer for a confusing topic: open with the problem, give three useful points and finish with one realistic action. Use a demonstration for a product: show the situation before use, the product being used and the practical result. For a service, show one common mistake, a better method and a simple invitation to request details. Keep each scene focused on one message.
3. Turn the response into a production plan
Ask for a numbered scene list containing the visual, spoken line or on-screen caption, approximate duration and required asset. Check whether every proposed shot can be created with your available phone, location, product and people. Shorten long captions and remove repeated scenes. If a claim sounds impressive but cannot be verified, replace it with a clear feature, process or customer-relevant benefit.
4. Complete the quality check
Before publishing, confirm that names, prices, dates and product details are accurate; captions are readable; speech matches the visuals; and the final action is clear. Use only photographs, video, voices, fonts and music you are permitted to use. Watch once without sound and once with sound. Common mistakes include tiny text, abrupt scene changes, unsupported claims and copying the first AI response without editing.