App Store Screenshots and Previews That Convert

Once a search result or chart position has earned an impression, the screenshots decide whether it becomes an install. This guide collects the current size and count rules for both stores, the evidence on where attention actually goes, and how to test creatives without guessing.

What each store allows

The two stores differ in counts, aspect ratios and where the images appear, so a single export rarely fits both.

Apple's iPhone requirement has shifted with the hardware line. The current rule is one set for the Dynamic Island medium display, with larger sizes optional and scaled automatically if absent; a separate requirement for the folding "iPhone Duo" arrives in April 2027 for apps built with the iOS 27.1 SDK ². Google, meanwhile, has been adding a three-image hero carousel at the top of the listing, rolled out from July 2025, which pushes screenshots and the short description further down the page ¹².

The first-three rule

Apple states that search results may display up to three screenshots or previews alongside your rating ⁵, and its product-page guidance says those first one-to-three images should capture the app's essence while each later frame covers one feature ¹. The store itself therefore makes the first three frames the de facto ad.

The behavioural data agrees. Storemaven's analysis of more than 500 million sessions (published December 2021) found that 60% of visitors never scroll past the first impression and that half of all installers decide on that first screen alone ⁸. The same firm splits traffic into "decisive" visitors (about 60%) who install or leave without touching the page, and "explorers" (about 40%) who swipe and read, with the decision typically made within three to six seconds ⁹. These are vendor-reported figures from a testing platform, not independent research, but no published study contradicts them.

Captions

Because the gallery is scanned rather than read, captions behave like headlines. The practical pattern across published redesigns is a three-to-four-word claim at the top of each frame in a large font, with the most colourful screen first and the core value proposition second ¹¹. One documented redesign that followed that recipe, removed a weak preview video and shortened the description recorded a 36% rise in product-page-view-to-install conversion a month later, while cautioning that marketing and seasonality were not controlled ¹¹. Keep text short, avoid time-sensitive wording, and on Google Play do not include store badges or ranking claims in the artwork, which Google disallows ⁶.

Dark mode and localisation

Apple asks for at least one dark-mode screenshot if the app supports it ¹. Both stores let you localise images per language, and Google advises localising taglines and overlay text, not just the in-app UI ⁶. Storemaven's 2021 data suggested reusing the same creatives from Google Play on iOS cost 20–30% of installs, and that culturally adapted variants beat straight translations ⁸.

Video previews

On iOS, previews always play before screenshots, autoplay muted, and a preview whose aspect ratio differs from the screenshots is moved into a separate "A Closer Look" section ³. The poster frame defaults to the five-second mark and is what appears when autoplay is off, so it effectively becomes screenshot one ³. Apple's own Simply Piano case study is a useful warning: putting a preview first lost to the original page after a 12-day test, contrary to the developer's expectation ⁴.

Storemaven reported that a good video can raise installs by more than 25% but that a poor one can cut conversion by the same margin, that 98% of App Store viewers watch with sound off, and that roughly 10% of viewers drop every five seconds ^8,9. On Google Play the video is not auto-played in the gallery, so view rates are far lower and the feature graphic's role as cover image matters more than the footage ^6,9.

A/B testing on both stores

Apple recommends applying a treatment only once it is declared better or worse than the baseline with at least 90% confidence ⁴; Google reports a "Draw" when the difference is smaller than the minimum detectable effect you set ⁷. Both companies advise changing one element per treatment so the result can be attributed ^4,7. Published outcomes give a sense of scale: Peak's icon-only test over 44 days found an 8% lift at over 98% confidence for one icon and small losses for two others ⁴; AppTweak reports a 21.5% install gain for a game after creative testing ¹³; SplitMetrics' 2024 benchmark set, drawn from more than 3,500 tests in finance, shopping and utilities, highlights a 75% improvement from a video A/B test and a 19% gain from an icon change, with the full category tables gated behind a form ¹⁰.

How this shows up in the charts

Charts rank on installs, and installs are impressions multiplied by conversion. Two apps with identical search visibility can sit 50 places apart on the free chart purely because one converts twice as well, which is why creative changes tend to appear on AppWinners as a steady climb over one to three weeks rather than the vertical spike a feature or paid burst produces. The signature is easy to spot: the app's category rank improves while its rating, update cadence and keyword footprint stay unchanged, and if you open the product page you will usually find a fresh first frame. The reverse also holds; when a seasonal screenshot set is left in place too long, the drift down the chart is slow enough to be mistaken for competition rather than a stale gallery.