Most iOS teams treat the screenshot gallery as a formality: export five shots from the simulator, drop them into App Store Connect, ship. Then they wonder why paid traffic converts at 22% while a competitor with a worse product sits at 38%.
We build and ship product pages for a living at Iris, and the pattern is boringly consistent. App store screenshots that convert are not prettier. They are structurally different. They follow layout rules that respect how the App Store actually renders the gallery, how long a visitor looks at it (roughly 5 to 7 seconds), and how much of the frame survives on a small screen.
Below is the teardown: 8 layout rules, exact sizing specs for 2026, the mistakes that waste gallery space, and a testing loop you can run this quarter.
Why the screenshot gallery carries more weight than the icon
On the App Store, a visitor lands on your page in one of two states:
- Search results view: your icon, title, subtitle, rating and the first slice of your gallery are visible without a tap. Many users decide here.
- Product page view: after a tap, the gallery sits directly under the title block, above the description. The description is collapsed. Almost nobody expands it.
That means your screenshots are doing the job of a landing page hero, a feature list and a trust section combined. Apple’s own Product Page Optimization tooling exists precisely because these assets move conversion rate by double digits.
Rule of thumb from our tests: a well-structured set typically lifts install conversion by 15% to 40% against a raw, uncaptioned simulator export. That lift compounds on every acquisition channel you pay for.
Rule 1: The first-three-screenshot rule
In portrait orientation, iOS search results show roughly three screenshots side by side before the user has to swipe. In landscape, it shows one. Everything after that costs an interaction, and most people never make it.
So treat the first three as a self-contained pitch that answers three questions in order:
- Screenshot 1: What is this? The single strongest value proposition, in plain language. Not a feature. The outcome.
- Screenshot 2: How does it work? The core interaction or the moment of magic. One screen, one idea.
- Screenshot 3: Why should I trust it? Social proof, a rating badge, a press mention, a user count, or a recognisable integration.
Screenshots 4 through 10 are for people who are already interested. Put secondary features, edge cases, pricing clarity and platform coverage (Watch, iPad, widgets) there.
What we see in high-converting sets
- Fitness and habit apps lead with a result screen (progress, streak, transformation), not an onboarding screen.
- Finance apps lead with a dashboard with real-looking numbers, then a security or trust panel by shot 3.
- AI tools lead with a before and after split inside a single frame, which compresses two shots into one.
- Games often break the rule intentionally and lead with pure art, because the buying decision is aesthetic, not functional.
Rule 2: Caption hierarchy, and the 5-word ceiling
The caption is the part of the screenshot people actually read. The UI underneath is scanned, not read. So the caption carries the message.
A working hierarchy for each frame:
| Layer | Content | Relative size |
|---|---|---|
| Headline caption | 3 to 5 words, benefit-led | 100% (largest element on the frame) |
| Support line (optional) | 6 to 10 words, clarifies the how | 45% to 55% of headline |
| Badge or proof chip | Rating, award, “No ads”, “Offline” | 30% of headline |
| App UI | Cropped screen, zoomed to the relevant region | Visual anchor, not the message |
Concrete numbers we use: headline set between 90 and 120 px on a 1290 px wide canvas, bold or semibold weight, line height around 1.1, maximum two lines. If your headline needs three lines, the copy is wrong, not the layout.
Caption placement
Top-aligned captions outperform bottom-aligned captions in most of our tests, for one mechanical reason: in the search results carousel, the bottom of each thumbnail is where the eye exits. Reading order goes top to bottom, and the top third is the only region guaranteed to be legible at thumbnail scale.
Exception: games and visual apps where the artwork is the pitch. There, a bottom caption bar over a darkened gradient keeps the art unobstructed.
Copy patterns that work
- Verb first: “Scan any receipt in 2 seconds”
- Outcome, not mechanism: “Sleep through the night” beats “Adaptive white noise engine”
- Numbers where credible: “12,000 recipes offline”
- Objection handling by shot 4 or 5: “No account needed”, “Your data stays on device”
Rule 3: Device framing, and when to drop it
Device frames (an iPhone bezel around your UI) are the default in nearly every screenshot generator. They are not always the right call.
Use a device frame when:
- You want to signal “this is a native iOS app” (useful for utilities, productivity, finance).
- You are showing multi-device support: iPhone plus Watch, or iPhone plus iPad.
- Your UI is dense and the bezel gives it visual breathing room.
Drop the frame when:
- Your app is visual and the screen content is the product (photo editors, games, video tools). A bezel shrinks your artwork by 20% to 30% for zero benefit.
- You are cropping a zoomed detail. A bezel around a cropped region reads as a mistake.
- You want maximum text legibility at thumbnail scale.
A middle path we use often: bleed the device off the bottom edge of the canvas. You keep the native signal from the bezel’s top corners while recovering vertical space for the caption. Angle the device slightly (6 to 10 degrees) only if the set is consistent across all frames. Random tilt on one frame looks accidental.
One hard rule from App Review: the UI inside the frame must be your actual app. Heavily mocked-up screens showing features you do not ship get flagged as inaccurate metadata. This is the single most common rejection reason we see when teams lean on generic templates. Related reading: AppLaunchpad: Free App Store Screenshot Generator.
Rule 4: Portrait vs landscape is a strategic choice, not a taste one
Portrait shows three tiles in search results. Landscape shows one. That is a 3:1 information advantage before anyone taps.
| Orientation | Tiles visible in search | Best for |
|---|---|---|
| Portrait | About 3 | Utilities, social, finance, health, most apps |
| Landscape | About 1 | Landscape-only games, video editors, racing and console-style titles |
If your app supports both orientations, ship portrait unless the landscape experience is the product. And do not mix orientations inside a single set: the gallery will render inconsistently and the swipe rhythm breaks.
Rule 5: Get the sizes right, once
Apple has simplified requirements considerably. You no longer need to export six iPhone sizes. As of the current App Store Connect requirements, you upload one iPhone set and one iPad set (if you support iPad), and Apple scales down for smaller devices.
| Device family | Portrait (px) | Landscape (px) | Notes |
|---|---|---|---|
| iPhone 6.9″ (required) | 1320 x 2868 or 1290 x 2796 | 2868 x 1320 or 2796 x 1290 | This is the master set. Design here first. |
| iPhone 6.5″ (optional) | 1242 x 2688 or 1284 x 2778 | 2688 x 1242 or 2778 x 1284 | Only if you want tighter control on older devices. |
| iPad 13″ (required for iPad apps) | 2064 x 2752 or 2048 x 2732 | 2752 x 2064 or 2732 x 2048 | Do not simply upscale the iPhone set. |
Other constraints worth memorising:
- Up to 10 screenshots per device family, per localization.
- Formats: PNG or JPEG, RGB, flattened, no transparency, no rounded corners baked in.
- App Preview videos (up to three) appear before your screenshots and autoplay muted. If you ship one, your first screenshot becomes the second thing people see. Plan accordingly.
Safe zones
Keep meaningful content at least 80 px from every edge on the 1290 px canvas. In search results, tiles are cropped slightly and neighbouring frames are partially visible. Text that sits near an edge gets clipped or crowded.
Rule 6: Design the set, not the screenshots
A gallery is a sequence. Users swipe. The strongest sets we tear down share three traits:
- A consistent grid. Caption block always in the same vertical position, device always at the same scale. When the layout is stable, the eye reads only what changes: the words.
- A colour rhythm. Two or three background treatments alternating, or a gradient that shifts progressively across the set. Ten different backgrounds looks like ten different apps.
- Optional panoramic continuity. A background image or shape that runs across frames 1 to 3, so the set reads as a single wide banner in the carousel. This works beautifully in portrait and is a strong differentiator, but it must survive being viewed out of order.
Where panoramas fail: if screenshot 2 makes no sense without screenshot 1, and a user lands on your page from a deep link that starts mid-carousel, you have lost the message. Every frame should still stand alone.
Rule 7: Localize the layout, not just the language
Running your English captions through a translator and re-exporting is the cheapest localization possible, and it is where most sets break. appscreens.com published something useful on the subject.
Practical issues we hit constantly:
- German and Finnish run 25% to 35% longer than English. A two-line headline becomes four lines and collides with the device frame.
- Japanese, Korean and Chinese are shorter but need different font weights and line heights to stay legible at thumbnail scale. Latin-optimised typefaces often lack proper CJK glyphs and fall back to something ugly.
- Arabic and Hebrew are right-to-left. The device should flip, the caption alignment should flip, and any arrow or directional element should mirror.
- The in-app UI must be localized too. An English UI under a Spanish caption tells the visitor the app is not really translated. That reads as a downgrade even when it is not true.
- Cultural specifics: currency symbols, date formats, names in avatars, and sample data. A screenshot showing “$” in a Japanese listing costs you credibility.
Prioritise by revenue, not by locale count. For most apps, English, then the top three to five markets by installs, then everything else. Apple lets you localize screenshots per storefront language, and partial localization is fine: untranslated locales fall back to your primary language.
Rule 8: Ship a hypothesis, then test it
Every rule above is a starting point, not a law. Apple’s Product Page Optimization lets you run up to three treatments against your baseline, split traffic, and read the result on install conversion with statistical significance built in.
A sane testing order:
- Screenshot 1 only. Highest leverage. Test two completely different value propositions, not two fonts.
- Caption copy across the first three. Benefit-led vs feature-led.
- Framing style. Device frame vs full-bleed.
- Set order. Move your social proof frame from position 3 to position 1.
Run each test for at least a full week to absorb weekday and weekend traffic differences, and do not stop the moment you see a lead. Also use Custom Product Pages (you can create many, each with its own screenshot set) to match creative to campaign. A set built around “track your macros” for a fitness campaign will out-convert your generic set on that specific traffic every time.
Common mistakes that waste gallery space
| Mistake | Why it costs installs | Fix |
|---|---|---|
| Raw simulator exports, no captions | UI text is unreadable at thumbnail scale, so the frame communicates nothing | Add one large headline per frame |
| Leading with onboarding or login | First impression is friction, not value | Lead with the outcome screen |
| Device shrunk to 50% of the canvas | Empty background eats the space your UI needs | Zoom and crop to the relevant UI region |
| Six lines of caption copy | Nobody reads paragraphs in a carousel | Cap at 5 words plus one support line |
| Mocked features you do not ship | App Review rejection for inaccurate metadata | Screenshot the real build only |
| Ten frames, ten different styles | Reads as unpolished, breaks scanning rhythm | One grid, two or three background variants |
| Low contrast caption on busy UI | Illegible in light mode, in sunlight, at thumbnail size | Solid or gradient caption zone behind text |
| Screenshots never updated after launch | UI drifts, the set stops matching the app | Refresh every major release |
A 90-minute workflow to rebuild your set
- Write ten captions first, in a plain text file, before opening any design tool. Rank them. Keep the best five.
- Capture the real screens at 1320 x 2868 from the largest iPhone simulator with realistic demo data, full battery, clean status bar.
- Build one master frame with the caption block, device position and background locked in.
- Duplicate five times, swap the screen and the caption. Nothing else moves.
- Zoom to 15% on your monitor. That is roughly thumbnail size. If you cannot read the headline of frames 1 to 3, redo them.
- Export iPhone and iPad sets, upload, and set your first PPO test running the same week.
FAQ
How many App Store screenshots should I upload?
You can upload up to 10 per device family. Ship at least five so the gallery does not look thin, and treat the first three as your real pitch. Beyond seven or eight, returns drop sharply.
What size should App Store screenshots be in 2026?
Design your master set at 1320 x 2868 px (iPhone 6.9″ portrait). If you support iPad, also export a 13″ iPad set at 2064 x 2752 px. Apple scales down for smaller devices, so a single iPhone set covers the rest of the lineup. Always double check the current requirements in App Store Connect before a submission.
Do captions really matter, or should I just show the UI?
Captions matter more than the UI in the search results view, because raw interface text is unreadable at thumbnail scale. The exception is visually driven categories such as games and photo apps, where the artwork itself is the pitch.
Should I use a device frame or a full-bleed screenshot?
Use a frame for utility, productivity and finance apps where “native iOS” is a selling point. Go full-bleed for games, photo and video apps where every pixel of artwork counts. If in doubt, run both as a Product Page Optimization test.
Can I get rejected because of my screenshots?
Yes. The most common cause is showing features, screens or content the app does not actually contain, which Apple flags as inaccurate metadata. Placeholder text, competitor logos and misleading award badges also trigger rejections. Screenshot the shipping build.
How often should I refresh my screenshot set?
Update after every major UI change, and revisit the messaging at least twice a year. Seasonal or campaign-specific variants belong in Custom Product Pages rather than replacing your default set.
Do localized screenshots actually increase installs?
In non-English markets, yes, and often significantly. But translation alone is not enough: the layout must absorb longer strings, the in-app UI shown must be localized, and right-to-left languages need a mirrored layout.
Bringing it together
The teams that win the gallery are not the ones with the biggest design budget. They are the ones who accept that the first three frames carry the decision, that captions are the message, and that every layout choice should be testable.
If you want a second pair of eyes on your current set, or you would rather hand the whole thing to a studio that ships these weekly, get in touch with the team at Iris. We will tear down your existing gallery, tell you exactly which frames are wasting space, and rebuild the set around a hypothesis you can measure.