How-to tutorial videos with AI: step cards, pacing and scripts
How to make how-to tutorial videos with AI: when to film and when to generate, a step structure, words per step, type sizes, pacing and a full example.
Show the result, then the steps
A good how-to video shows the finished result first, then walks through numbered steps with one action per scene, and ends with a quick check that the viewer did it right. To make one with AI, write the steps as short commands, decide which steps need real footage, and use generated clips, diagrams and step cards for the rest.
People search for how-to videos while they are in the middle of the task, often with their hands full. Clarity beats style. If a viewer has to rewind, the video has failed, however good it looks.
Name the video the way people search for the task: “how to repot a houseplant,” not “Plant care, part 3.” Use the same words in the first line of narration and in the description, so search and the viewer both know they are in the right place.
Decide what to film and what to generate
AI is not the right tool for every part of a tutorial. Sort each step before you start.
- Film or screen-record: software steps, exact hand techniques, anything where the viewer copies precise movements. A screen recording of the real interface beats any generated version.
- Generate: the finished result, the setting, the materials laid out, general actions like “water the plant,” and transitions.
- Draw as graphics: measurements, diagrams, warnings, step numbers and anything with exact text or numbers.
If a generated clip shows a technique slightly wrong, viewers will copy the mistake. When precision matters, film it.
Write the steps
- Start each step with a verb: “Loosen,” “Tip,” “Trim,” “Press.”
- One action per step. If a step has “and then,” split it.
- On-screen text per step: a number and three to seven words.
- Narration per step: one or two sentences, 15 to 30 words, explaining what and why.
- Add a check after tricky steps: “The roots should hold the soil’s shape.”
- Put safety warnings on screen as text, not only in narration.
Most tutorials land between five and nine steps. If you have fifteen, make a series.
Design and pacing guidance
- Step badge: a number at 120 to 160 px in a consistent corner, inside the safe area.
- Step text: 64 to 80 px on a 1080 by 1920 frame, two lines at most.
- Materials list card: 48 to 56 px, one item per line, six items at most per card.
- Hold each step card for at least 2 seconds, and keep the step text on screen while the action plays.
- Use a consistent color for the step badge and a second color only for warnings.
- Change the shot every 3 to 5 seconds, but never cut away before an action finishes.
- Keep a 4.5:1 contrast ratio between step text and its background panel.
Viewers often pause tutorials. Design every step frame so it makes sense as a still image.
Full example: repot a houseplant
A 60-second vertical tutorial. Generated clips are noted with a sample prompt. Graphics are drawn as cards.
- 0–4 s · Result: a healthy plant in a new pot on a sunny shelf. Text: “Repot a houseplant in 6 steps.” Prompt: “Leafy houseplant in a fresh terracotta pot on a sunlit wooden shelf, soft morning light, slow push-in.” Try it
- 4–9 s · Materials card: new pot one size up, fresh potting mix, trowel, water. VO: “You’ll need a pot one size bigger, fresh potting mix, a trowel and water.” Try it
- 9–17 s · Step 1 “Water the day before.” VO: “Water the plant a day ahead. Moist roots slide out more easily.”
- 17–25 s · Step 2 “Tip and slide it out.” Filmed clip of hands tipping the pot. VO: “Turn the pot sideways, support the base of the stems, and ease the plant out.” Try it
- 25–33 s · Step 3 “Loosen circling roots.” VO: “If roots circle the bottom, gently tease them apart with your fingers.”
- 33–41 s · Step 4 “Add mix to the new pot.” Diagram: plant top sits just below the rim. VO: “Add enough mix so the top of the root ball sits just below the rim.”
- 41–49 s · Step 5 “Fill and firm.” VO: “Fill around the sides and press lightly. Don’t pack it hard.”
- 49–55 s · Step 6 “Water until it drains.” VO: “Water slowly until it runs from the drainage hole.”
- 55–60 s · Check card: “Plant stands upright, soil settled.” Text: “Save for repotting day.”
Making it in ClickPop
In Faceless Studio, paste your steps and mark which scenes use your own footage. ClickPop draws the step badges, cards and diagrams in code, so measurements and numbers are exact, and it adds narration, captions and music. Generate result shots and setting shots in Create Video. If your audience speaks several languages, ClickPop’s AI voices cover 30+ languages, so you can reuse the same visuals with a new voice and translated cards.
Mistakes to avoid
- Starting with a long intro instead of the result.
- Generated clips for steps that need exact technique.
- Step text that disappears before the action ends.
- Missing materials in the list, so viewers stop halfway.
- No safety warning on screen for steps involving heat, blades, chemicals or electricity.
Questions
Can AI make tutorial videos?
AI can script, narrate, caption and illustrate a tutorial. Steps that need exact technique or a real software interface should still be filmed or screen-recorded.
How long should a how-to video be?
As short as the task allows. Simple tasks fit in 30 to 90 seconds. Longer tasks work better as a series of short videos.
How much text should be on each step?
A step number and three to seven words on screen. The narration explains the detail.