Decide who the video is for first

Edtech has several audiences at once: students, teachers, parents and the administrators who buy. They care about different things. A teacher wants to know whether it saves time. A student wants to know whether it is easy and worth their attention. A school buyer wants to know whether it fits into what they already use.

Pick one. A video aimed at everyone usually lands with no one. If you need all four, make four short cuts from the same visual style.

Teach one idea per video

The best educational explainers do what good teachers do: they pick one idea and make it clear. For a product video, that idea is one outcome, such as "build a quiz in two minutes" or "see which students need help at a glance." Resist the urge to tour every feature.

A simple structure that works:

1. A relatable situation (10 seconds)
2. The product solving it step by step (40 to 60 seconds)
3. The result and one next step (10 seconds)

Let motion do the explaining

Learning is visual. Animated diagrams, progress bars filling, a score changing or a lesson unfolding on screen all show the idea faster than narration can describe it. Use motion to guide the eye to what changes, and keep everything else still so the viewer is not overloaded.

Typography matters more than people expect. Large, plain, high-contrast text works for young learners and busy teachers alike. Avoid decorative fonts for anything the viewer must read.

Match the tone to the learner

A product for children needs bright colours and friendly characters. A product for university students or professionals needs a cleaner, more adult look. A product for teachers should feel efficient and trustworthy rather than playful. Pick the tone from the audience, not from your own taste.

Accessibility is part of the product

Education products have a duty to be accessible, and the video should show that care:

• Add accurate captions
• Keep colour contrast strong
• Avoid flashing effects
• Keep the narration pace comfortable
• Consider translations if you sell internationally

These choices help everyone, including people watching without sound on a shared device.

Where the video should live

An edtech video can do a different job in each place:

• On the landing page, it explains the product quickly
• In a sales email, it answers a teacher's first question
• Inside onboarding, it gets new users to their first success
• In a school's internal channel, it helps a buyer champion the product to colleagues

Decide this before production, because it changes the length, the opening and the call to action.

Common mistakes

• Packing in every feature
• Using classroom clichés that feel dated
• Voice-over that talks down to the viewer
• Skipping captions
• No clear next step at the end

Final thought

A good edtech video behaves like a good lesson: one clear idea, a calm pace and something the viewer can do right away. If you are planning one, start by writing the single sentence you want the viewer to remember, and build everything else around it.

I make product explainers and UI animation for SaaS, AI and fintech teams, and I take on edtech products too. My work and services are at vaghelabhavesh.com.

1. Send one consolidated round

Collect every comment from everyone on your team first, resolve disagreements, then send a single list. A round of feedback is one batch on one version. Five separate messages from five people usually means five different videos.

2. Use timecodes

Write "0:14, the second line is too small on mobile" instead of "the middle feels off". A timecode tells the editor exactly where to look and saves a back-and-forth.

3. Say what is wrong, not how to fix it

"The product is hard to read in this scene" gives the editor room to solve it well. "Make the screen 20% bigger" might fix that one frame and break the layout around it.

4. Separate must-fix from nice-to-have

Mark each note as a must-fix, a preference or a question. It helps the editor plan the round and stops small taste changes from eating the revision budget.

5. Check against the brief

Before you send notes, reread the goal, audience and length you agreed. Many late changes are really changes to the brief, and they are better raised early, as a scope conversation.

6. Review on the device your audience uses

If the video is for Reels or a phone feed, watch it on a phone, with sound off first. Text size and captions are the most common issues found late.

Quick checklist

• One list, one owner, one round
• Timecodes on every note
• Problems described, not solutions prescribed
• Notes labeled must-fix or preference
• Reviewed on the real device

I go into more detail, including how many revision rounds a project should include, here:

• How to give video feedback
• How many revisions should a video project include?

I'm Bhavesh, a freelance video editor and motion designer in India. If you are planning a product video, see how I work.

1. Check the licence. Read the terms of the tool you used. Confirm that commercial use is allowed on your plan, and keep a record of which tool and plan produced each clip.

2. Avoid real people without consent. Do not generate a likeness of a real person, a celebrity or a competitor's staff. If you use an AI presenter, make sure the tool allows it for ads.

3. Label it honestly. Do not present AI footage as a real customer or a real testimonial. Some platforms require a label for AI-generated or altered content, and regulators are paying attention to fake endorsements.

4. Review every frame. AI video can produce warped hands, garbled text, inconsistent logos and odd physics. Watch the full clip at full size and cut or regenerate anything that looks wrong.

5. Never invent product facts. AI can show a product doing something it cannot do. Product footage and claims must match what the product really does.

6. Keep a human in the edit. Use AI for the pieces it does well, such as backgrounds, b-roll and variations, then edit, grade, caption and sound-design by hand so the result feels like your brand.

7. Have a fallback. If a clip fails review, replace it with stock, screen recordings or motion graphics.

Longer guide: Is AI video safe for brands?

Premiere Pro is for assembling. Cutting interviews, pacing a story, syncing audio, adding captions and cutting a long video into clips all happen faster on a timeline built for editing.

After Effects is for building. Animated UI, logo animation, kinetic type, shape-based motion graphics and anything that needs precise control over how things move belong in After Effects.

Most explainer videos use both. I animate the scenes in After Effects, then bring them into Premiere Pro to add voice-over, music, captions and the final cut. Dynamic Link lets me keep the two connected without rendering every change.

A quick rule I follow:

• If the work is mostly choosing and trimming clips, start in Premiere Pro.
• If the work is mostly designing how things move, start in After Effects.
• If it is a short social edit with simple text, Premiere Pro is usually enough.

For clients, the tool matters less than the result. What counts is the final file in the right sizes for each channel, delivered on time.

I wrote a longer comparison here: After Effects vs Premiere Pro.

Before production

• One sentence about what is new and who it is for.
• Final screens. Animation built on a design that later changes has to be redone.
• Script and length: 45 to 90 seconds, with shorter cut-downs for social.
• Assets: logo, colors, fonts, screen recordings or Figma files.

During production

• Approve script, then storyboard, then style frames, before full animation.
• Give feedback in one list from one decision-maker.
• Review with sound on and off, and on a phone.

Before publishing

• Captions in place, and correct sizes for each channel.
• A strong first frame and clear title.
• Embed on the launch page.

Start at least three to four weeks before launch. The longer version is here: Product launch video checklist.

Start with one concrete example. Pick a task the viewer recognises, such as turning a prompt into an image or an agent finishing a web task. Show the input, then the result.

Show the interface, not the architecture. Most viewers don't need a model diagram. They need to see what they click, what they get and how long it takes.

Keep claims honest. Avoid "perfect" or "human-level" unless you can prove it. Say what the tool is not for. Label any illustrative scenes.

Let motion explain the process. Data flowing in, a result coming out. One idea per scene.

Length. 45 to 75 seconds for a homepage, plus a 20 to 30 second cut for social.

I animated explainers for products in this space, including Browserbase and Krater AI.

Full guide: Explainer videos for AI startups.