App preview video vs screenshots: what actually converts

Every other surface went to video years ago. Store listings are the last place where good teams still ship five static frames, and the teams who do add video often break their own search result doing it.

What video is actually worth

Aggregated 2026 benchmarks from Meta ad accounts (agency round-ups, not Meta's own data, so directional):

Those numbers do not transfer. In a feed you are interrupting someone, and video wins because it stops a scroll. On a product page the user already tapped and is deciding whether to install, which is why store lifts land in single and low double digits, and why a bad video can lose you installs where a bad ad is merely ignored.

SplitMetrics reports about +16% from adding an App Store preview, StoreMaven puts it at +20% to +35%. A strong static set is worth a similar amount on its own, so video is additive to good screenshots, not a replacement. By category, AppFollow's directional ranges:

Several testing shops also report video hurting conversion in finance and productivity by 3 to 7 percent. The rule: video wins when your app is something to watch, and loses when it is something to use.

The two stores ship video differently

On the App Store video is a discovery asset that plays before anyone has chosen you. On Play it only exists after the tap. That one difference should decide the budget split, and it is the opposite of how most teams allocate.

Play: YouTube channels per language

Under Grow users, Store presence, YouTube Videos you can attach channels or playlists to the listing, with a language set on each, one language per channel. A listing in four languages can carry four video surfaces instead of one. Very few do. The rules:

• Public videos, though playlists can be unlisted.
• Monetization off, embeddable, owned by your app or game.
• No Shorts, no live videos.
• Games: uploaded within 90 days to show on the listing, 21 days for the Games tab, 180 days for the Apps tab.
• Needs Premium growth tools eligibility and the Manage store presence permission.
• One playlist for an app, several playlists or channels for a game.

Play Console then reports viewers, click-through and installs over 28 days. A live example: Buzzin: TV Party Game Show. Play also accepts a referrer parameter on store links, so tag them before sharing rather than guessing which channel drove an install.

The App Store landscape trap

Apple lets a preview be portrait or landscape and warns about neither consequence. On the product page a landscape preview does not lead the gallery, it moves into a separate A Closer Look section:

In search results it takes the full width of the row and the screenshots are not shown beside it. In the wild it gets worse: the row renders with no creative at all. Below, the same app sits between two competitors showing three frames each, and shows nothing but an icon and a line of text.

• Portrait app, portrait preview. It sits alongside your first screenshots instead of replacing them, so you keep both surfaces.
• Genuinely landscape product (a TV game, a racing game, a video editor): either letterbox the footage into a portrait frame so the row keeps the gallery, or accept a bare row and let the icon, title and subtitle carry it.
• Never mix orientationsin one device's set. A portrait UI in a landscape frame is a documented App Review rejection reason, and the exact sizes each store expects are worth checking before you re-export.
• Verify on a clean device after every release. The layout differs for people who already have the app, so your own phone tells you very little.

The same app on the other store, for comparison: Buzzin on the App Store.

The craft rules that decide the outcome

• Your poster frame is a screenshot. Autoplay is conditional, so when it does not fire the whole video collapses to that one still. Caption it, make it legible at thumbnail size, and never ship a black screen with a wordmark.
• The first three seconds work on mute. Show the product in second one, carry the message in on-screen text, front-load the best moment.
• 15 to 20 seconds even though Apple allows 30. Shorter loops get seen more than once.
• Localize it. Video is per localization in both stores; burned-in English text on a German listing is the same mistake as an untranslated caption.
• Keep it current. A preview showing a UI from two redesigns ago is worse than no video.

Before you spend the budget

1. If your first two frames are raw captures with no captions, fix those first. That is a bigger and cheaper win, and ordering and captioning them costs an afternoon.
2. If a still frame communicates the same thing, the video adds load time and risk without adding information.
3. If budget is tight, iOS first: there video buys a discovery surface, on Play it only pays off after the tap.

Then test it rather than trusting the averages above. Product Page Optimization on the App Store and store listing experiments on Google Play are free. Run video against no video, and poster frames against each other separately, because those are two different experiments. Give each two weeks, change one variable, and record what the listing looked like when the numbers moved. That last part is what quietly breaks, and keeping a per-language, per-store record of what shipped and when is a large part of what we built AppBoard to do.

Do the cheap check first, though. Search for your own app on a phone that has never installed it and look at what the row renders. If it comes back as a line of text between two competitors showing three frames each, no amount of production value fixes that.

Frequently asked questions

Store rules verified against Apple and Google documentation in August 2026. Conversion figures come from A/B testing vendors (SplitMetrics, StoreMaven, AppFollow) and are directional, not peer-reviewed.