How to make an explainer video

An explainer video makes one idea understandable in about a minute. The reliable way to build one is five parts in a fixed order: the problem the viewer has, the solution in one sentence, how it works in three or four short steps, one piece of proof, and one ask. Write the narration for those parts first, give every scene one idea and one picture, and stop. Most explainers that fail do so because they try to explain three ideas, not because the pictures were wrong.

The example below the post is the Wren onboarding video, a fictional HR portal explained to a new starter in eight landscape scenes and 50 seconds. It uses the same bones: why you are here, the steps, a recap, who to ask, and a close.

Why five parts, in this order

Two pieces of learning research do most of the work here.

The first is about length. Philip Guo and colleagues studied 6.9 million video watching sessions across four edX courses and found that shorter videos were much more engaging, with median engagement time at most six minutes regardless of how long the video was. Viewers of tutorial videos engaged for only two to three minutes. An explainer that makes its point in under two minutes is working with viewers instead of against them. Wistia’s 2026 report points the same way for marketing video, putting average engagement for videos under a minute at 52 percent.

The second is about order. Richard Mayer and Roxana Moreno’s 2003 review, Nine Ways to Reduce Cognitive Load in Multimedia Learning, describes a pretraining effect: people understand an explanation better when they already know the names and behaviours of the parts involved. That is what the problem and solution scenes are for. By the time the mechanism starts, the viewer knows what the pieces are called and why they should care.

Scene one: the problem

State a problem the viewer already has, in one factual sentence. Not a caricature of a frustrated person, and not a statistic about the industry. The dashboard tour opens with its problem as a question, “Why did sign-ups drop last week?”, and every scene after it answers part of that question.

Name the thing the rest of the video depends on. In the Coppervale Water plan in the table under this post, the problem line is “Most water filters get changed when someone remembers, which is rarely.” The filter is the part; the forgetting is the problem. Both are now in the viewer’s head before any mechanism appears.

Keep the visual simple and drawn. A cartridge fading from white to grey says more than a stock photo of someone frowning at a tap.

Scene two: the solution in one sentence

Name the solution and what it does, in words a viewer could repeat. “Coppervale posts a new filter the week yours runs out.” One claim, no features list. Mayer and Moreno’s coherence effect is the reason: people learn better when interesting but extraneous material is left out. Every extra benefit in this scene competes with the one the viewer needs to carry into the next.

The Wren video does this in its intro: “Welcome. Your first day in Wren, in four steps.” It names the product, the job and the size of the job in nine words, so the viewer knows how much attention to budget.

Scene three: how it works, in short steps

This is the middle of the explainer and the part most likely to overload. Mayer and Moreno’s answer is segmenting: break the explanation into bite size segments with time between them rather than one continuous stream. In a video, a segment is a scene. One step, one scene, one picture.

The Wren video gives each of its four steps seven or eight seconds and a single action: “Complete your profile: photo, phone, emergency contact.” Then “Add your bank details under Pay; they are checked before the first run.” That second line carries its reason, which is what gets a step done today rather than next week.

Two more effects from the same review shape these scenes. Temporal contiguity: people learn better when the animation and the narration describing it happen at the same time, so the picture should change as the line is spoken, not after. Signalling: cues that show where to look reduce wasted effort, which is why the expense claim walkthrough spotlights the control each step is about and dims the rest.

Guo’s study adds a note on style: tablet drawing tutorials were more engaging than slides, and the authors recommend motion and continuous visual flow. Drawn steps that build on screen tend to hold attention better than static bullet slides.

Scene four: proof

Proof answers the doubt a viewer actually has after the steps. For Coppervale that doubt is effort, so the proof is “Swapping takes two turns: old one out, new one in.” For a product with measured results, it is one number with its sample, the way the Replyline launch video prints “across 32 beta support teams” under its 41 percent.

Internal explainers prove themselves differently. The Wren video’s proof that onboarding is finishable is its checklist: “By Friday: profile, bank details, policies, meet your buddy.” Four ticks turn a vague first week into something a starter can complete.

If the proof is a chart, say the number in the narration. The W3C’s accessibility guidance notes that charts, graphs and on screen text need description for people who cannot see them, and that description is easier when it is part of the script from the start.

Scene five: the ask

End with one action. For a customer explainer, it is the first step of using the thing: “Enter your filter model to see your first delivery date.” For an internal one, it can be as small as the Wren close, “See you Monday,” after a drawn three card scene tells the starter who to ask.

Do not add a second ask, a social link or a recap of the whole video. The explainer has done its job if the viewer knows what the thing is, why it matters and what to do next.

The table below plans Coppervale’s explainer as seven scenes, with the narration line, the visual and the research principle each scene leans on. Swap in your own lines and keep the column on the right; it is the part that stops the video drifting into a feature tour.

Common mistakes

Mechanism before motive. Starting with how it works leaves the viewer holding steps with no reason to remember them.

Full sentences on screen while the voice reads them. Mayer and Moreno’s redundancy effect found identical narration and on screen text at once hurt learning. Use short labels and let captions do the accessibility job.

Eight steps in one explainer. Teach three, link a tutorial for the rest.

Proof that answers a question nobody asked. Awards and logos rarely address the viewer’s real doubt.

Stock faces. Drawn diagrams explain ideas better, and an explainer does not need a presenter; see what a faceless video is.

A tutorial in disguise. If the viewer has to follow along click by click, write a tutorial instead; the structure for that is in how to write a tutorial video script.

Build it

The AI explainer video maker builds from a brief of up to 8,000 characters plus up to 8 images and 5 documents, and documents are summarised rather than transcribed, so put the key steps in the brief itself. Choose Promotional for an explainer that ends on a sales ask, or Tutorial, from the Starter plan, for one that teaches; tutorial steps run 5 to 9 seconds with 10 to 25 words of narration and never get the energetic voice. Scenes can carry 10 chart kinds and 6 diagram kinds, including a funnel, a cycle and Venn diagrams. Captions are on by default and the format is landscape or vertical. The steps are in make a promo video with an AI voiceover.