How to make an explainer video

01An explainer is not a demo

A demo shows a product doing something. An explainer makes a case: here is the problem, here is why the obvious fix does not work, here is the idea. The screen is illustration rather than subject, which changes almost every decision that follows.

The most common failure is recording an explainer as though it were a demo, ending up with two minutes of an interface being operated while a voice talks about something more abstract than what is on screen.

• Write the script before you open the recorder, not after
• Decide what each sentence needs on screen, and cut the sentences that need nothing
• Sixty to ninety seconds for a landing page, up to three minutes for a considered purchase
• End on the idea, not on a button

02Write it, then read it out loud

An explainer is the one screen recording format where a written script is genuinely required, because the argument has to hold without the crutch of demonstrating something. Read it aloud before recording. Sentences that work on a page and not in a mouth are found here and nowhere else.

Time the read. A ninety second script is around 220 words at a comfortable pace, and scripts routinely come in at twice that on the first draft.

03Gather the screen material first

Record the screen sections separately from the narration rather than trying to do both live. Recording is controlled from the tray, and paused segments are stitched back together automatically, so a single recording with pauses between the pieces is easier to manage than six files.

The source picker takes a display, a window or a region drawn with the Area tab, which is what you want when the material is one diagram or one panel rather than a full application.

04Fixed-point zoom, not cursor tracking

This is the setting that matters most for explainers. Automatic zoom follows the cursor, which is correct for demos and wrong here, because in an explainer nobody is driving the interface. The frame needs to be where the argument is.

Any individual zoom can be switched off cursor-following and pointed at a fixed spot, set with a crosshair on a mini preview canvas. Combined with the intensity control from 1.0x to 4.0x and a slow zoom speed, that gives a deliberate move onto a specific part of a diagram, timed to the sentence that needs it.

05Open with a face, then get out of the way

A face at the opening establishes who is making the argument, which an explainer needs more than a demo does. The webcam sits in one of eight positions at 5 to 40 percent of frame height with an adjustable corner radius, and a digital zoom of up to 3.0x crops into the camera feed itself.

After the opening, the frame belongs to what you are explaining. A permanent bubble in the corner of an explainer competes with the thing the viewer is being asked to understand.

06Captions carry the argument

Explainers are watched on mute more than any other format, because they are embedded in landing pages and social feeds that autoplay silently. If the argument only exists in the audio, most viewers never receive it.

Word level captions with progressive highlight, at an opacity you set, keep a scripted narration followable without sound. Transcription runs on your own machine with a choice of model sizes and is a Pro tier feature.

07A scripted read needs more audio work, not less

Reading a script produces a flatter delivery than talking through a product, and it exposes sibilance and room tone that conversational energy usually covers. The six voice processing controls handle it: cleanup, adaptive noise suppression, bass boost, voice clarity, voice boost and a de-esser around 7 kHz, applied at export rather than baked in.

Automatic silence trimming is less useful here than elsewhere, since a scripted read has few long pauses, and the pauses it does have are usually deliberate.

08Export square and vertical from the same project

Explainers earn their cost by being reused: the landing page, the social post, the sales deck, the onboarding email. Six aspect ratios including 1:1 and 9:16 mean those come out of one project rather than one recording each.

The Pro tier exports to 4K at up to 120 fps with GPU accelerated rendering. For a landing page embed, 1080p is usually the better trade, since the file has to load before the argument can start.