Narrated product film
Let the voice lead, but make every line earn a picture. Pair each spoken claim with one visual subject that proves or clarifies it.
Script first
Write short, speakable lines. One capability or claim per line usually makes one beat. Trace factual claims to the current product. When a feature depends on a mode or setting, name that condition in the line and show it on screen. Read the script aloud before building the timeline.
Mark the intended voice, pace, pronunciation, and pauses. Record or synthesize the voice only through a service the user has authorized and for which the voice rights are clear. Keep line and word timings so a revised read can retime captions and picture without guessing.
Pair words and images
Make a beat table with the spoken line, the real UI or visual proof, the entry and exit, and any on-screen text. Use subtitles or highlighted words where they improve comprehension; do not stack the full script over a busy interface. A simple recurring layout can make different features feel like one film.
Timing should follow the voice. Let actions happen with the words that name them, and let the result settle before moving on. Keep transitions motivated by the product or story, not by the availability of an effect.
Sound and verification
Mix music below intelligible speech and use sound effects only where they support a visible action. Use owned or licensed audio. Check the export by listening and watching separately: spoken words match the script, captions match the audio, each visual matches its line, no UI control is hidden, and the closing is long enough to read. Reframe and inspect each requested ratio.