Quick answer
The four icon mistakes that cost real installs are adding your own rounded corners, shipping an icon with transparency, putting your app name in the icon, and designing at 1024 pixels without checking how it reads at around 60. Google Play now renders store icons at a 30 percent corner radius, so anything close to the edge gets visually cropped. Apple requires a fully opaque icon with no alpha channel, and a transparent one is the single most common icon rejection.
Short Answer
> Quick answer: Ship a fully opaque icon with square corners and let the system do the masking. iOS wants 1024 by 1024 as a PNG with no alpha channel, and Google Play wants 512 by 512 under a megabyte, also non transparent. Play now renders store icons at a 30 percent corner radius, which is close to circular, so keep anything important well inside the edges.
Then check it at roughly 60 pixels. That's the size people actually see, and detail that looks good at 1024 disappears completely at the size that decides whether they tap.
Then check it at roughly 60 pixels. That's the size people actually see, and detail that looks good at 1024 disappears completely at the size that decides whether they tap.
Rounding Your Own Corners
This is the most common self inflicted problem and it's entirely avoidable.
Both platforms mask your icon for you. Supply a square image and the system applies its own shape, which means an icon that already has rounded corners baked in ends up with two radii fighting each other. The result is a visible white or transparent notch at each corner that looks like a rendering bug.
Drop shadows have the same problem. A shadow drawn into the artwork gets clipped by the mask and leaves a dark smear along one edge, and since the platforms apply their own elevation treatment, yours was never adding anything.
Play's 2026 change makes this worse rather than better. At a 30 percent corner radius the mask is aggressive enough that a pre rounded icon loses noticeably more than it used to, and plenty of icons designed before that change now look wrong in the store without anybody having touched them.
The fix takes minutes. Export square, full bleed to the edges, and let each platform shape it. Apple's app icon guidelines state the same thing if you need to settle an argument about it.
Both platforms mask your icon for you. Supply a square image and the system applies its own shape, which means an icon that already has rounded corners baked in ends up with two radii fighting each other. The result is a visible white or transparent notch at each corner that looks like a rendering bug.
Drop shadows have the same problem. A shadow drawn into the artwork gets clipped by the mask and leaves a dark smear along one edge, and since the platforms apply their own elevation treatment, yours was never adding anything.
Play's 2026 change makes this worse rather than better. At a 30 percent corner radius the mask is aggressive enough that a pre rounded icon loses noticeably more than it used to, and plenty of icons designed before that change now look wrong in the store without anybody having touched them.
The fix takes minutes. Export square, full bleed to the edges, and let each platform shape it. Apple's app icon guidelines state the same thing if you need to settle an argument about it.
The Safe Zone Nobody Draws
A 30 percent radius takes a serious bite out of every corner, and most icons were not designed with that in mind.
Anything positioned near an edge is at risk. A logo that fills the canvas edge to edge, a border running around the outside, a wordmark sitting along the bottom, all of them get clipped or crowded once the mask lands. The artwork isn't wrong, it's just the wrong shape for where it's displayed.
Treating the icon as a circle during design solves it. Keep the focal element centred and well within an imaginary circle inscribed in the square, and it will survive any mask either platform applies now or later. That's a stricter rule than necessary today and it ages better.
Background is the other half. A flat or simply graded background that runs genuinely edge to edge means the mask only ever cuts background, never subject, and the icon reads identically whatever shape is imposed on it.
Checking this is quick if you mock it up rather than imagine it. Dropping the icon into a realistic device view in the editor shows the masked result immediately, which is faster than reasoning about radii.
Anything positioned near an edge is at risk. A logo that fills the canvas edge to edge, a border running around the outside, a wordmark sitting along the bottom, all of them get clipped or crowded once the mask lands. The artwork isn't wrong, it's just the wrong shape for where it's displayed.
Treating the icon as a circle during design solves it. Keep the focal element centred and well within an imaginary circle inscribed in the square, and it will survive any mask either platform applies now or later. That's a stricter rule than necessary today and it ages better.
Background is the other half. A flat or simply graded background that runs genuinely edge to edge means the mask only ever cuts background, never subject, and the icon reads identically whatever shape is imposed on it.
Checking this is quick if you mock it up rather than imagine it. Dropping the icon into a realistic device view in the editor shows the masked result immediately, which is faster than reasoning about radii.
Transparency Is the Top Rejection
An alpha channel in an App Store icon is the single most common reason icons get kicked back, and it usually happens by accident.
Design tools export PNGs with alpha by default, so an icon that looks completely solid on screen can still carry a transparent channel in the file. The rejection arrives after submission, which is a slow way to discover a five second export setting.
Google Play has the same requirement for the store listing icon. Non transparent 512 by 512 PNG, under a megabyte, and a transparent one either fails validation or renders against a background you didn't choose.
Flattening onto an opaque background before export removes the whole category of problem. Not white specifically, just something solid, chosen deliberately rather than inherited from whatever your canvas happened to be.
One caveat worth knowing: this rule is about the store icon rather than every asset in your app. Adaptive icon layers on Android work differently, and mixing up which asset a requirement applies to is a normal way to get confused here.
Design tools export PNGs with alpha by default, so an icon that looks completely solid on screen can still carry a transparent channel in the file. The rejection arrives after submission, which is a slow way to discover a five second export setting.
Google Play has the same requirement for the store listing icon. Non transparent 512 by 512 PNG, under a megabyte, and a transparent one either fails validation or renders against a background you didn't choose.
Flattening onto an opaque background before export removes the whole category of problem. Not white specifically, just something solid, chosen deliberately rather than inherited from whatever your canvas happened to be.
One caveat worth knowing: this rule is about the store icon rather than every asset in your app. Adaptive icon layers on Android work differently, and mixing up which asset a requirement applies to is a normal way to get confused here.
Text in an Icon
Putting your app name in the icon feels like free branding and costs you legibility instead.
At around 60 pixels, which is roughly the size an icon occupies on a phone, text collapses into a grey smudge. Anything smaller than a couple of bold characters simply isn't readable, and the space it occupies would have been better spent making the symbol larger.
The name already appears directly beneath the icon in every context that matters, on the home screen, in search results and on the product page. Repeating it inside the artwork is spending your most valuable pixels on information the user already has.
Look at any top chart and the pattern is consistent: one clear symbol, one or two colours, no text. That isn't a stylistic fashion, it's what survives being shrunk.
A single letter can work if it's the whole design and it's set large enough to read instantly. That's a monogram rather than a wordmark, and the distinction is whether it still functions at thumbnail size.
At around 60 pixels, which is roughly the size an icon occupies on a phone, text collapses into a grey smudge. Anything smaller than a couple of bold characters simply isn't readable, and the space it occupies would have been better spent making the symbol larger.
The name already appears directly beneath the icon in every context that matters, on the home screen, in search results and on the product page. Repeating it inside the artwork is spending your most valuable pixels on information the user already has.
Look at any top chart and the pattern is consistent: one clear symbol, one or two colours, no text. That isn't a stylistic fashion, it's what survives being shrunk.
A single letter can work if it's the whole design and it's set large enough to read instantly. That's a monogram rather than a wordmark, and the distinction is whether it still functions at thumbnail size.
Designing at the Wrong Size
Almost every icon is designed at 1024 pixels and almost none are seen anywhere near that size.
Working large is fine, judging large is not. Fine gradients, thin outlines, small internal details and subtle textures all read beautifully on a monitor and vanish at 60 pixels, which is where the actual decision gets made. A designer can spend a day on detail that no user will ever perceive.
Keep a scaled preview open while you work. Sixty pixels beside your full size canvas gives you continuous feedback, and it tends to push the design toward bolder shapes and stronger contrast without you having to consciously decide to simplify.
Context matters as well as size. An icon on a busy home screen sits among dozens of competitors, and one in a search result sits in a list, and both are more crowded than the empty artboard where it was designed. Our device frames put it in a realistic setting quickly.
I've seen this one mistake account for more weak icons than every specification issue combined, and it never shows up as a rejection because nothing about it is technically wrong.
Working large is fine, judging large is not. Fine gradients, thin outlines, small internal details and subtle textures all read beautifully on a monitor and vanish at 60 pixels, which is where the actual decision gets made. A designer can spend a day on detail that no user will ever perceive.
Keep a scaled preview open while you work. Sixty pixels beside your full size canvas gives you continuous feedback, and it tends to push the design toward bolder shapes and stronger contrast without you having to consciously decide to simplify.
Context matters as well as size. An icon on a busy home screen sits among dozens of competitors, and one in a search result sits in a list, and both are more crowded than the empty artboard where it was designed. Our device frames put it in a realistic setting quickly.
I've seen this one mistake account for more weak icons than every specification issue combined, and it never shows up as a rejection because nothing about it is technically wrong.
When It Is Worth Changing an Icon
Redesigning an icon carries a cost most teams underestimate, so it deserves a real reason rather than a refresh instinct.
Existing users find your app by shape and colour rather than by reading the label, and a changed icon means a period where they cannot find it. That friction is small per person and large in aggregate, and it shows up as a dip in opens that gets blamed on other things.
Good reasons to change anyway include an icon that fails at small sizes, one that is now visually indistinguishable from a larger competitor, and one that is being clipped badly by the current Play mask. Those are functional problems that cost installs every day they remain.
Bad reasons include a general brand refresh with no measurement attached, or a new designer preferring a different style. I've watched an icon change wipe out a quarter of a percent of daily opens for a fortnight with nothing to show for it afterwards, and nobody had defined what success would look like beforehand.
If you do change it, keep one anchor constant, usually the dominant colour or the silhouette. Changing everything at once is what breaks recognition, while evolving one dimension at a time lets people follow you across the change without noticing they are doing it.
Existing users find your app by shape and colour rather than by reading the label, and a changed icon means a period where they cannot find it. That friction is small per person and large in aggregate, and it shows up as a dip in opens that gets blamed on other things.
Good reasons to change anyway include an icon that fails at small sizes, one that is now visually indistinguishable from a larger competitor, and one that is being clipped badly by the current Play mask. Those are functional problems that cost installs every day they remain.
Bad reasons include a general brand refresh with no measurement attached, or a new designer preferring a different style. I've watched an icon change wipe out a quarter of a percent of daily opens for a fortnight with nothing to show for it afterwards, and nobody had defined what success would look like beforehand.
If you do change it, keep one anchor constant, usually the dominant colour or the silhouette. Changing everything at once is what breaks recognition, while evolving one dimension at a time lets people follow you across the change without noticing they are doing it.
A Short Checklist Before You Ship
Six checks, and none of them takes long.
Export square with no rounded corners and no drop shadow, then confirm the file is fully opaque with no alpha channel. Those two cover the rejections and the rendering bugs between them.
Confirm the dimensions per platform, 1024 by 1024 for iOS and 512 by 512 under a megabyte for Play, and check the Play version against a 30 percent corner radius specifically rather than assuming the old shape.
View it at 60 pixels next to five competitor icons from your category. If yours isn't immediately distinguishable in that lineup, the problem is the design rather than the specification, and no amount of correct export settings fixes it.
Then look at it on both a light and a dark background, since the surrounding wallpaper is out of your control and an icon that only works on one of the two is half broken. Getting through that list puts you ahead of a surprising share of shipped apps.
Export square with no rounded corners and no drop shadow, then confirm the file is fully opaque with no alpha channel. Those two cover the rejections and the rendering bugs between them.
Confirm the dimensions per platform, 1024 by 1024 for iOS and 512 by 512 under a megabyte for Play, and check the Play version against a 30 percent corner radius specifically rather than assuming the old shape.
View it at 60 pixels next to five competitor icons from your category. If yours isn't immediately distinguishable in that lineup, the problem is the design rather than the specification, and no amount of correct export settings fixes it.
Then look at it on both a light and a dark background, since the surrounding wallpaper is out of your control and an icon that only works on one of the two is half broken. Getting through that list puts you ahead of a surprising share of shipped apps.
Frequently asked questions
Should I add rounded corners to my app icon?
No. Both platforms apply their own mask, so a pre rounded icon ends up with two radii fighting and a visible notch at each corner. Export square and full bleed.
What corner radius does Google Play use for icons now?
Play renders store icons at a 30 percent corner radius, which is close to circular. Anything near the edge gets clipped or crowded, so keep the focal element well inside an imaginary circle.
Why was my app icon rejected by Apple?
Most often because it contains transparency. iOS store icons must be fully opaque with no alpha channel, and design tools export PNGs with alpha by default even when the artwork looks solid.
What size should my app icon be?
1024 by 1024 pixels for iOS and 512 by 512 under a megabyte for Google Play, both as non transparent PNGs. Judge the design at around 60 pixels, since that's the size people actually see.
Should my app name be in the icon?
No. Text collapses into a smudge at real display sizes and the name already appears directly beneath the icon everywhere it matters. A single large letter can work as a monogram, a wordmark cannot.
How do I check my icon looks right before submitting?
View it at 60 pixels beside competitor icons from your category, on both light and dark backgrounds. Dropping it into a realistic device view in [the editor](/create) shows the masked result straight away.
Build the mockup in your browser.
Drop a screenshot into a real device frame and export at the exact store size — free, no signup.