Most used
Everything else
Eight decisions, in the order you actually make them.
Updated: 2026-09-10
One video, one question. Not 'photosynthesis' but 'why do leaves need light at all'. The test is whether you can say the question in a single sentence without the word 'and'. If you need 'and', you have two videos. This decision governs everything downstream, because a script that wanders produces boards that wander, and a viewer who cannot tell what is being answered leaves at the twenty-second mark.
Two routes. Write it yourself and paste it, and every word is used exactly as typed. Or give a topic and let the writer draft narration sized to the length you picked. Paste your own when the wording matters, when facts must be exact, or when the video has to sound like you. Let it write when you want a first draft to react to. The one mistake to avoid is pasting a full script into the 'write it for me' box: there it gets condensed to the chosen length rather than kept.
Length is a word count, not a feeling. Speech runs at roughly one hundred and fifty words a minute, so a two-minute video is about three hundred words. Decide the length first and write to it. Videos that overrun do not get trimmed evenly; the ending gets rushed, which is where the point usually lives.
The style changes how the video feels more than any other setting. Classic marker for teaching, chalkboard for classroom material, infographic for revision and data, Sharpie for something bolder. Keep one style across a series so the videos read as a set. The frame is separate: landscape for YouTube and slides, portrait for Reels, Shorts and TikTok. Choose the frame before rendering, because a landscape video cropped to portrait loses the sides of every drawing.
Two modes. One board per paragraph is the default: each section of narration gets its own drawing, and the board changes when the topic changes. The guided tour is one single picture for the whole video, with each object appearing the moment the narration names it and staying on the board. Use the tour for anything with parts that belong together: a layout, an anatomy, a process, a lineup. Use one board per paragraph for anything that moves through separate ideas.
Director notes are hard constraints on the picture, not on the content. Name an aesthetic, list three to five concrete ingredients such as hex colours, a recurring character or a prop rule, and state one thing that must never happen. 'Warm, engaging visuals' does nothing. 'Ochre and slate blue only, one recurring woman in a teal jacket, never draw text inside speech bubbles' does everything. Eighty to three hundred characters is the working range.
When the video comes back, do not watch it as a viewer. Check three things. First, the opening: does the first sentence state the question, or does it warm up? Second, the board splits: does the picture change when the idea changes, or does one drawing carry two unrelated points? Third, the timing: does each element appear as the narration names it, or slightly after? If the first is wrong, fix the script's opening line. If the second is wrong, split the paragraph. The third is automatic and rarely needs anything.
Export the MP4 and take the subtitle file with it. Most explainer video views on social start muted, so burned-in subtitles are not an accessibility extra, they are the difference between a view and a scroll. For YouTube, upload the separate subtitle file rather than burning them in, so viewers can turn them off and the text stays indexable.
No. Every decision above is made in text: a question, a script, a length, a style. Nothing is placed by hand on a timeline.
The decisions above take a few minutes. Rendering a one-minute video takes a few minutes more; a seven-minute video takes around ten.
Yes. Paste it under 'I have a script' and nothing is rewritten. The narration is your text, and the boards are built from it.
Abstraction without an object. 'Efficiency improved' has nothing to draw; 'the same machine now runs eleven parts an hour instead of seven' has a machine and two numbers.
Yes. The same script renders with narration in fourteen languages, and the drawing timing is recomputed for each one.
The explanation you already give most often out loud. You know it works, and you will spot immediately whether the video got it right.