App Store screenshots are the ad, not the documentation
Most listings use raw device screenshots in feature order. The first two images do nearly all the work, and most apps waste them on a login screen.
Store screenshots are usually produced at the end of a project, by whoever has the simulator open, in the order the features were built.
They are the highest-leverage marketing asset the product has. Most people decide from the first two images without scrolling.
The first image has one job
It answers what is this and who is it for, in about a second, to someone not reading carefully.
That is almost never a raw screenshot. A screenshot shows an interface; it does not say what the interface is for. The first image needs a short line of text stating the benefit, with enough of the interface visible to make it concrete.
- Say what it does, not what it has. Plan your week, not Calendar view.
- Six words or fewer. It is read at thumbnail size.
- Show the app doing the thing, not the app's login screen.
The login screen appears more often than any other first image, and it is worth understanding why. It is the first screen the app shows, so it is the first screen the person capturing screenshots sees. It is chosen by the capture process rather than by anyone deciding what to say.
Order by value, not by menu
The instinct is to present features in navigation order. Present them in order of how much someone cares, which usually means the thing the app is genuinely best at comes first even if it is buried three taps deep.
A useful exercise: ask the founder which single feature makes someone choose this app over the obvious alternative, then check where that feature appears in the listing. It is frequently fourth, behind two screens of setup and an account page.
Screenshots are not a tour. They are an argument.
Legibility beats fidelity
A pixel-perfect screenshot of a dense screen is illegible at listing size. Zoom into the part that matters, increase the type, and cut the chrome. Nobody is evaluating your navigation bar.
Use realistic content, not lorem ipsum and not implausibly perfect data. Names of normal length, a list with a few messy entries. Real content reads as a real product.
The test we use is a thumbnail test. Export the image, scale it to about the width of a thumb, and look at it for one second. If you cannot tell what the app does, the image is decoration. This catches more problems than any amount of discussion at full size, because full size is not how anybody sees it.
Write the caption before the screenshot
The fastest way to produce a good set is to write the five sentences first, as a list, with no images at all. Each sentence is a claim about what the product does for someone.
Read the list back on its own. If it reads like a coherent pitch, the screenshots are a matter of illustrating five things you have already decided. If it reads like a menu, no amount of design work on the images will fix it, because the problem is the argument rather than the pictures.
This also stops the common failure where five captions are five nouns. Five verbs almost always outperform them, and you cannot notice that while looking at devices.
Localise the text, not just the device
Teams that sell in several markets usually generate localised screenshots by running the same capture with the app language changed. That translates the interface inside the frame and leaves the caption, which is the part doing the persuading, in English.
It is worth checking which markets actually produce installs before spending on this. For many apps two or three languages cover most of the audience, and doing those three properly beats doing twelve mechanically.
Keep them current
Screenshots showing an interface the app no longer has are a common rejection cause, and a worse conversion problem: people arrive expecting one product and find another.
We put screenshot review on the release checklist for exactly this reason. It is the item most often skipped and most often wrong.
A platform redesign makes this sharply worse, because every screenshot ages on the same day without anybody touching the app. If the operating system has changed its appearance and your listing has not, the images say the product is unmaintained, which is the one message a listing must never send.
If you change one thing
Replace the first image. It carries more weight than the other nine put together, it is the one most likely to be a login screen or a splash, and changing it costs an afternoon.
Everything else on this list is worth doing. None of it is worth doing before that.
