How Do You Optimize an App for the App Store and Google Play?

App Store Optimization is two jobs that get reported as one. Ranking decides how many people ever see your listing. The listing decides how many of them install. Work on the wrong one and the effort disappears, because a listing that converts brilliantly to nobody earns nothing, and neither does a top-ranked listing that nobody taps.

Search is why the ranking half matters so much. AppTweak’s 2026 ASO and Apple Ads benchmarks report, built on the top 1,000 US apps and games across both stores through 2025, puts store search at 65% of iOS discovery against browse at 18%, referrers at 12% and ads at 5%. On Google Play search accounts for 58%.

Where do app installs actually come from?

Most app installs come from store search on both platforms, at 65% of discovery on iOS and 58% on Google Play per AppTweak’s 2026 report. Paid ads account for 5% of iOS discovery in the same dataset, which is worth holding next to every conversation about user acquisition budgets.

That split has a practical consequence for sequencing. Keyword and metadata work compounds and costs nothing per install once it is done, while paid acquisition stops the day the budget stops. An app with no organic ranking has no floor under its install volume, so every paid campaign is buying from zero rather than adding to a base.

Browse traffic behaves differently from search traffic and should be read separately. Someone who arrived from an editorial feature or a category chart has no query behind them, so the listing has to do all the explaining. Someone who searched for “sleep tracker” has told you exactly what they want, and the listing only has to confirm it.

Referral traffic, at 12% of iOS discovery in the same AppTweak study, is where creator content and paid social actually land.

A user who watched a creator video and then searched your app name is counted as search or referrer depending on the path they took, which is one reason store analytics and ad platform numbers rarely reconcile. Read both, expect them to disagree, and do not try to make them match.

Which fields does each store index?

Apple and Google index different fields, and this is the single biggest reason a copy-paste listing underperforms on one of the two stores. Apple indexes the app name, the subtitle and a hidden keyword field, and does not index the description for search at all. Google Play indexes the title, the short description and the entire long description.

	App Store	Google Play
Name or title	30 characters, strongest ranking weight	Indexed, 30 characters
Subtitle or short description	30 characters, indexed	80 characters, indexed
Keyword field	100 characters, hidden from users, indexed	none
Long description	4,000 characters, not indexed for search	4,000 characters, fully indexed
Keyword repetition	no benefit, Apple deduplicates	treated as a ranking input

What follows from the table is two different writing jobs. On iOS you have 160 indexed characters in total, so every word has to be a search term someone actually types and nothing can be repeated between the three fields. The description is pure conversion copy and can be written for humans without a single keyword in it.

On Google Play the long description is both. You are writing something a person will read and something Google will index, which means natural keyword placement through the body copy rather than a keyword list. The two stores reward opposite instincts, and the correct output is two listings, not one.

What conversion rate should a store listing hit?

Around 33% on the App Store and around 28% on Google Play. AppTweak’s 2026 benchmarks put average tap-through-to-install at 33.4% on iOS against 27.7% on Play, a gap of 5.7 percentage points, and UXCam’s 2026 mobile app benchmarks land independently at 33.7% and 26.4%.

Two sources agreeing within half a point is unusual enough in this category to be worth saying out loud, and it makes the platform gap credible rather than a sampling artefact. AppTweak attributes most of it to layout: Apple renders the first three screenshots above the fold, while Play surfaces the description and similar apps earlier, so Play visitors see competitors before they see your third screenshot.

Category variance is much wider than the platform gap, which limits what either number is good for. Utilities listings routinely run in the mid thirties while finance and health listings sit far lower, and a rate that would be poor for a casual game can be strong for anything requiring an account and a credit check.

Use the medians to check you are in the right order of magnitude, then compare against your own previous listing.

The one comparison to avoid entirely is your iOS rate against your Android rate. Play’s install base spans far more markets, arrives through a differently ordered page, and converts on different intent, so the difference between the two tells you almost nothing about either.

What moves store listing conversion most?

The first two or three screenshots, the icon and the rating do most of the work, because most visitors decide without scrolling and without reading the description. Neither store indexes image content, so screenshots are a pure conversion lever with no ranking effect, which makes them the cleanest thing on the page to test.

Most apps never test them, which is the useful part.

AppTweak found that only 35% of top apps and 33% of top games ran two or more screenshot tests in a year on the App Store, and that between 76% and 80% of apps on Google Play have never tested an icon, video or feature graphic at all.

Being in the third that tests repeatedly is a cheaper edge than most of what teams spend on instead.

Events are the other habit that separates the top of the games charts. According to the same AppTweak report, 51% of top games ran more than five in-app events on the App Store, against 38% of apps.

Paid traffic gives you a second surface worth optimising separately. Custom product pages let an Apple Ads campaign land on a listing built for that specific query or creative, and AppTweak found that ad variations based on custom product pages lift conversion 6.6% on average in Apple Ads search results campaigns, rising to 8% for mobile games.

Custom product pages are also the cheapest fix for the mismatch that quietly wastes paid social budget. A creator video sells one specific outcome, the generic store listing sells the whole product, and the person who tapped is now reading about something else.

Matching the landing page to the ad promise is the same discipline as matching a paywall to it, which is part of what we run in an app launch.

How do ratings and post-install behaviour affect ranking?

Ratings work as a gate rather than a dial, and the gate sits higher than most teams assume. AppTweak found that 95% of apps featured by the App Store hold a rating of 4.0 or above, so editorial visibility is effectively closed below that line regardless of how good the listing copy is.

Both stores also now weight what happens after the install, which changes the incentive structure of keyword targeting. Per AppTweak’s breakdown of Google Play ranking factors, Android Vitals suppress Play rankings for crash and ANR rates above Google’s thresholds, and Apple factors uninstall rates and session frequency into App Store ranking.

A listing that wins installs from a badly matched keyword sends users who churn immediately, and the ranking it earned degrades as a result.

The practical version of that rule is to read keyword performance alongside retention rather than on its own. A term that delivers volume and day-one churn is worse than no term, and it will not show up as a problem in an ASO dashboard that stops at the install.

When you ask for the rating matters as much as whether you ask. A prompt fired on second launch reaches users who have not yet got anything out of the product, while one fired after they have finished something reaches users with a reason to be positive. The rating that results feeds both conversion and eligibility for features, so the timing of a single dialog ends up gating editorial visibility.

How often should you update your store listing?

Update your store listing quarterly at minimum, and faster while you are still learning which keywords convert. Keyword rankings move as competitors change their own metadata, and a listing left alone for a year has usually drifted off terms it used to hold.

Build the cycle around one change at a time. Both stores support native A/B testing of creative, tests need a week or more to reach usable volume, and stacking a metadata change on top of a screenshot test destroys the reading of both. Slow sequential tests beat fast confounded ones.

Keep a record of what each round changed and what happened, because store algorithms are undocumented and your own history is the only reliable evidence you will get about them. After four quarters you will have something no published benchmark can give you, which is a model of how your specific category responds.

The listing is also not where this work ends. A store page that converts 35% of taps into installs and then hands those users an onboarding flow that loses most of them has moved the problem rather than solved it, which is covered in why app onboarding is so important.

The short version

App Store Optimization is ranking plus conversion, and search is the channel that matters most, at 65% of iOS discovery and 58% on Google Play per AppTweak’s 2026 benchmarks. Paid ads account for 5% of iOS discovery, so organic metadata work is the floor everything else sits on.

Write two listings, not one. Per each store’s published field rules, Apple indexes the name, subtitle and a hidden 100-character keyword field and ignores the description, while Google Play indexes the title, short description and the full 4,000-character long description, so keyword strategy inverts between them.

Expect around 33.4% tap-through-to-install on iOS and 27.7% on Google Play per AppTweak, and never compare the two to each other. Test the first three screenshots and the icon repeatedly rather than once, since that is what top performers do, keep the rating above 4.0 to stay eligible for features, and read keyword performance against retention because both stores now penalise installs that churn.