App Store Optimization for Indie Games (2026)

App store optimization is the work of getting a mobile game found in App Store and Google Play search, and chosen once a searcher sees it. On an indie budget it costs time, not money: fill the few fields each store actually indexes, build a listing that converts, and keep the quality signals clean. It is typing and testing, in that order.

The stakes are one search box. Each store now holds roughly 2.5 million apps, and about 300,000 of Apple's are games (42matters, Business of Apps). Apple's own figure says roughly 65 percent of App Store downloads happen directly after a search (AppTweak). A player typing "offline puzzle game" is the cheapest visitor you will ever get, because they arrive already wanting what you made.

This piece covers the listing itself. The other half of mobile shipping, from developer accounts and fees to review timelines, is our Android and iOS publishing guide.

The short answer#

The whole job fits in one table. The two stores index different fields, so the same game needs different work on each.

Task	App Store	Google Play
Fields indexed for search	Name (30 chars), subtitle (30), hidden keyword field (100)	Title (30 chars), short description (80), full description (4000)
Field that does not index	Description (4000 chars)	None; there is no hidden keyword field
Conversion assets	Icon, 1 to 10 screenshots per device, up to 3 preview videos	Icon, 2 to 8 screenshots per device, optional video
Built-in testing	Product page optimization, up to 3 visual treatments	Store listing experiments, up to 3 variants, text and graphics
Quality signals	Star rating, review prompt capped by the system	Star rating, crash and ANR rates affect discoverability
Country targeting	Localized metadata per language	Up to 50 custom store listings

Source for limits: Apple product page requirements, Google Play graphics requirements.

How each store reads your listing#

Apple compresses. Search reads the name, the subtitle and the 100-character keyword field, and ignores the description entirely. Your whole keyword budget on iOS is 160 characters, and Apple's rules for spending it are explicit: never repeat a word across fields, since a duplicate wastes a character slot; use singular forms, since the store handles plurals; and keep competitor names and trademarks out, because they are grounds for rejection. The name is the field Apple weighs most.

Google writes. There is no keyword field; the algorithm reads the title, short description and full description as ordinary text, and ranking responds to phrases repeated naturally in real sentences. The trade for that freedom is policy: since the 2021 metadata update, Play titles may not carry promotional words like "best", "free" or "top", and the store enforces it (App Radar).

Same game, two different afternoons.

Keyword research for zero dollars#

You are not guessing what people type. You can watch it.

1. Autocomplete in each store. Type a seed word like "puzzle" into the App Store or Play search box and write down every suggestion. Those strings are real queries, ranked roughly by volume; the method costs nothing and needs no account.
2. The Apple Search Ads keyword planner. A free account shows every keyword a popularity score from 5 to 100 (Apple). You never have to run an ad. Scores of 5 to 20 are the long tail an unknown game can realistically rank for; 60 and up is chart territory where marketing budgets live.
3. A teardown of five competitors. Open the listings of games ranking for your seed terms and copy out their name, subtitle and short description. Words appearing in most of them are the market's vocabulary; phrases appearing in none of them are your opening.
4. Play Console search data, once live. Store performance shows the actual queries that surfaced your listing, next to impressions and installs. Apple gives you the channel but not the terms, so on iOS you iterate from conversion data and the ads planner instead.

For games specifically, the head term is lost before you start. Nobody outranks "puzzle game" with a first release. The winnable queries carry modifiers: "offline puzzle games", "puzzle games no wifi", "relaxing physics game". Each modifier is a smaller, named audience, which is exactly what a small game needs.

Writing the metadata: a worked example#

Take a hypothetical physics puzzle called Sling Bot and write both listings.

Apple, 160 characters total:

Field	Limit	Content	Chars
Name	30	Sling Bot: Physics Puzzles	26
Subtitle	30	Offline brain teasers, no wifi	30
Keywords	100	aim,bounce,relax,logic,space,ball,sort,time,zen,tricky,fun	58

Notice what is absent from the keyword field: every word already used above it. "Physics", "puzzles", "offline", "brain" and "wifi" are spent, so the 100 characters buy only new terms. Apple also lets you edit the 170-character promotional text at any time without a review, which makes it the right home for patch notes and event announcements.

Google Play, written as copy:

Field	Limit	Content	Chars
Title	30	Sling Bot: Physics Puzzles	26
Short description	80	Aim, bounce and land. 120 hand-made physics puzzles that play offline.	70
Full description	4000	Opens with the pitch, repeats the target phrases three or four times, then lists levels and modes	900 to 1500 works

The full description's first two lines carry the most weight with humans, because that is what shows before "Read more". Write the opening for the reader, then work "physics puzzle game" and "plays offline" into the body naturally. Density past a handful of honest repetitions reads as spam to both the algorithm and the reader.

The assets that convert#

Search results show four things: icon, title, the subtitle or short description, and stars. Everything else on the listing page has one job, converting a tap into an install.

The icon is the only asset that appears on every surface, including settings and the home screen, so it earns the most iteration. Test it at small sizes; detail that reads at 1024px dies at 48px.

Screenshots: Apple takes 1 to 10 per device size, Play takes 2 to 8 per device type. Show gameplay, not menus; a settings screen has never installed a game. The first three screenshots carry most of the attention, so put your clearest mechanic hook in the first one and caption it in plain words.

Video: a Play listing can lead with a YouTube link, and Apple previews autoplay silently until tapped. Either way the first three seconds must be the game itself, because that is all a scrolling player grants you.

Ratings, and the moment you ask#

Both stores give you a system prompt and ration it. Apple shows a game's rating request at most three times per 365-day period per device (Apple); Google Play enforces an unpublished time-based quota on its in-app review dialog (Android docs). You cannot detect whether the prompt appeared on either platform.

That scarcity makes timing the whole strategy. Ask after a cleared level or a won run, never at a fail state or a paywall. And never buy ratings: both stores purge them routinely, and the policy strike outlasts the bump.

Crashes are an ASO problem on Android#

Google publishes bad-behavior thresholds for stability: a user-perceived crash rate above 1.09 percent or an app-not-responding rate above 0.47 percent puts the app in bad-behavior territory, and apps over the line get reduced discoverability in Play search and recommendations (Android vitals). No amount of keyword work outranks a crash loop on low-end phones. For a 2D game the fix is usually boring and bounded, and we cover the common ones in mobile performance optimization.

The loop, one hour per release#

1. Read the data. Play Console shows the queries; App Store Connect shows the channel. Note which terms bring impressions but no installs, because that is a conversion problem, not a keyword problem.
2. Change one thing. A keyword swap, a subtitle rewrite, a new first screenshot. One variable, or you learn nothing.
3. Use the free machinery. Play store listing experiments test up to 3 variants, text and graphics, on real traffic; Apple's product page optimization tests up to 3 visual treatments for up to 90 days (Apple, Play Console).
4. Re-read the data two or three weeks later, then repeat. A listing is a draft that ships in public.

What indie listings get wrong#

The title stuffed until it stops reading as a name. The same word burned twice across Apple fields. "Best" sitting in a Play title where policy forbids it. A competitor's name in the keyword field, waiting for rejection. Five screenshots of the main menu. A rating request at the game-over screen, the single worst moment to ask. A listing written at submission and never touched again, which is the quiet one that gets most games.

None of these are expensive to fix. Most of them show up in the data as impressions that never became installs; the rest show up as a rejection email.

Where this fits#

If the build is done, the account and review mechanics are in the publishing guide, the common rejection reasons for iOS are collected in our App Store submission mistakes guide, and the Android-side fixes live in the top five publishing errors. Shipping both stores from one project is covered in build once for iOS and Android. If you are still picking an engine, the 2D mobile engine comparison is the earlier question. It is also the one we are building Egmatic to answer: a visual editor that takes a 2D game from idea to a store-ready build without code. The editor is pre-alpha; the waitlist is on egmatic.com, and the listing work above will be yours on launch day either way.

Start with one afternoon. Run the autocomplete exercise for your genre, write both metadata sets in a text editor, and count the characters against the table above. The assets and the loop follow; the 160 characters come first.