Why screen recordings need automatic zoom
If you record your screen at native resolution and your viewer is on a phone, the buttons in your video are around 6 pixels tall. They cannot see what you clicked. This is a math problem more than a craft problem, and automatic zoom is the cheapest fix.
The math
A 14-inch MacBook Pro renders at 3024×1964. You record at native, then export to a 1920×1080 master. The master gets watched on:
- A 27-inch desk monitor at 2560×1440. Scale factor ~0.75. The button is around 22 pixels.
- A 13-inch laptop at 1920×1080. Scale factor ~1. The button is around 16 pixels.
- An iPhone in a feed, played inline. Scale factor ~0.4. The button is around 6 pixels.
how tall your button is to an iPhone viewer watching an unzoomed 1080p screen recording inline in a feed
16 pixels is the lower bound of "comfortable to read". Six is invisible. Your video is doing its job for the person at a desk and failing for everyone else.
The manual fix is expensive
In Final Cut or Premiere, the fix is a Ken Burns move: keyframe a zoom into the button, hold for the click, keyframe back out. For a 90-second demo with 14 clicks, that's 14 keyframe sets. Each one takes 30–60 seconds to position and retime. You spent 10 minutes on zooms before you even cut the video.
This is why most homemade product demos either don't zoom at all (and lose the phone viewer) or zoom inconsistently (which is worse — viewers notice the asymmetry).
"Most homemade demos either don't zoom at all, or zoom inconsistently. Inconsistent is worse — viewers notice the asymmetry."
What automatic zoom actually does
The recorder watches the mouse. Every click is an event with a screen coordinate. When it sees one, it animates a virtual camera from the previous position to that coordinate, holds while the click resolves, then eases back out. The result is a Ken Burns move on every click, with a consistent rhythm, with zero work in the editor.
The good implementations let you do three things after the fact:
- Edit the zoom — change its target, its strength, its duration.
- Delete the zoom on individual clicks you didn't want emphasised.
- Add a zoom on a non-click moment (cursor parking on a chart, a number changing).
Auto-zoom is a starting point, not a contract. The recorder should give you both — automatic everywhere by default, manual control where it matters.
Where automatic zoom is wrong
It's wrong when the click isn't the point. You're explaining a chart; you click the legend to toggle a series, but what you want the viewer to look at is the chart, not the legend. Auto-zoom will push in on the legend. Delete that zoom; add one on the chart manually.
It's also wrong on long drags — selecting a range, dragging an object across the canvas. A zoom into the start of the drag and a hold doesn't match the motion. For those, either kill the auto-zoom for that click or let the recorder follow the drag (some do, most don't).
Why this matters for indie products specifically
If you ship for the App Store, the Mac, the web, or a creator tool — the first moment a prospective customer sees your product is usually a 30–60 second video on a phone. The single biggest thing you can do to make that video work on a phone is push the camera in on what's being clicked.
Manual is fine if you have 10 minutes per demo. Automatic is the right default if you ship demos weekly. More on the rest of the production checklist here.
The recorders that ship it
Screen.studio popularised this on Mac. Cursorful does a lighter version. CursorFlow ships it on every click by default, with timeline-level editing so you can adjust any individual zoom. A short side-by-side of those last two is in this post.
If your recorder doesn't have automatic zoom — and most of the cloud ones don't, in 2026 — the fix is to manually keyframe in your editor or switch recorders. There isn't a third path. The viewer on the phone is real, and they can't see your button.
