Do App Store preview videos increase downloads? What Apple's own test data says

It depends, and Apple's own example shows why

Apple's Product Page Optimization page walks through an example test from Simply Piano. The hypothesis was that "adding an app preview video as the first asset on the product page will result in higher conversion compared to only screenshots." The test ran for twelve days with one treatment. Apple's summary: "The original product page without an app preview video outperformed the treatment, contrary to the developer's hypothesis." The gap was about three percent, at full statistical confidence.

That is a music learning app with a strong visual hook, run by a company that knows how to make videos, and the video still lost. Apple did not pick that example by accident. It is a reminder that a preview is a change to the first thing a visitor sees, and changes to the first thing people see can go either way.

We make preview videos for a living, so we have every incentive to tell you they always work. They do not always work. They work when they earn the first three seconds.

Where the big numbers come from

Search for this question and you will find ranges like "20 to 40 percent higher conversion" repeated across ASO tool blogs. SplitMetrics' preview guide uses that range and adds that the effect "is particularly strong for complex, highly interactive, or new-to-market apps." The same guide notes something more useful: in their experience, "two short app previews focusing on two different features performed better than a single 30-second video."

None of those figures are lies, but they are averages of successful tests published by people who sell testing tools. Failed tests do not get blog posts. What the numbers do tell you is the size of the prize when a preview is right for the app and well made: tens of percent, not tenths.

Why the same asset helps one app and hurts another

Two things decide it.

The first is whether motion adds information. A drawing app, a game, a fitness app with animated workouts, a video editor: a still cannot show what these do, and a preview can. A calculator, a notes app, a banking app with a clean dashboard: the screenshot already says everything, and a video that shows the same screens moving slowly adds a delay before the visitor gets the point.

The second is what happens before the point. Previews autoplay on the product page and in search results, muted, and they occupy the first asset slot. If the first two seconds are a logo animation or a tagline fading in, the visitor is looking at nothing while the screenshots they would have read are pushed off screen. That is the mechanism behind a video losing to stills. It is rarely the video being bad in a general sense; it is the opening being empty.

A preview that opens on the app's most recognisable screen, already moving, already doing the thing people came for, keeps the advantage of the still and adds the advantage of motion. That is the version that produces the numbers vendors like to quote.

What a preview that converts has in common

Apple's App Previews guidance is short and worth following literally. Previews "play with the sound muted, so consider using copy to add additional context to your footage." Text has to be "legible, easy to understand, and remain on the screen long enough for people to read." And the poster frame should "convey the essence of your app", because it is what people see when autoplay is switched off.

From our side of the edit, the habits that show up in previews that test well:

• The first clip is the app's main screen, in use, with no intro. Our 30 second script template starts there for the same reason.
• One flow per preview. Two 15 second previews about two features beat one 30 second tour, which matches what SplitMetrics reported.
• Text cards are five to seven words and stay up for at least two seconds.
• The poster frame is the same screen the video opens on, so autoplay-off users and autoplay-on users see the same thing.
• Nothing in the first ten seconds requires sound. (More on that in how to make a promo video that works with the sound off.)

How to measure it on your own listing

Apple's test tool is built into App Store Connect. You create up to three treatments, choose the share of traffic that sees the test, and Apple splits the rest evenly: with 40 percent of traffic in a two-treatment test, each treatment gets 20 percent and the original keeps 60. Users see the same version for the whole test. Tests run for up to 90 days or until you stop them, and the results page reports conversion rate, percent improvement and a confidence level. Apple suggests waiting for at least 90 percent confidence before applying a change.

A few practical notes from tests we have watched:

• Test the preview against your current best screenshots, not against an old set. The comparison should be "video plus screenshots" against "screenshots", with everything else identical.
• Give it enough impressions. Small apps often need the full duration to reach confidence, and stopping early because a number looks good is how people ship a loser.
• If the video loses, try the opposite opening before you conclude that video does not work for you. A preview that starts on the payoff screen is a different experiment from one that starts on a logo.
• Google Play has its own store listing experiments with the same logic, and the promo video there is a YouTube link, so the test is cheap to set up.

So, should you make one?

If your app moves, make one, and open on the movement. If your app is a still-friendly utility, make one anyway if a screenshot cannot show the key interaction, and test it. If a good screenshot already answers "what does this do", spend the budget on the screenshots and revisit the video when you have a feature that needs motion to explain.

Either way, run the test. The vendors' averages tell you the upside exists. Only your listing can tell you whether it exists for you.