App Store Optimization in 2026, the guide you hand to an agent.

Both stores rank on three things: relevance of the text they index, the conversion of people who see you, and the quality of the users you already have. Text gets you into the results. Conversion decides the order. Retention decides whether you keep it. That third part is new: since 2025 Apple weighs post install retention and rating recency, and Google's recommendations lean on behaviour after install, so a download spike on its own no longer moves you.

On the App Store the indexed text is the title, the subtitle, the keyword field, and since June 2025 the captions on your screenshots. Custom product pages can be linked to keywords since July 2025 and appear in organic search, with up to 70 per app since October 2025. On top of matching, Apple's models generate App Store Tags from your metadata and use them for browse placement, and the search interprets conversational queries by intent before it matches words.

On Google Play the indexed text is the title, the short description and the long description, with the long description weighing far more than most people assume. Since September 2025 Guided Search turns broad queries into intent buckets before showing apps, and since I/O 2026 Ask Play answers questions on the listing from your description and your website, in English first.

Start with the three apps that rank where you want to rank and list every term they rank for in each of your top five countries. Then add what your own reviews say, in the users' words. From that list keep the terms with enough volume to matter and a difficulty you can beat within weeks. A term with 48,700 monthly searches in the US where you are not ranked is worth more than a term with 61,300 where you already sit at position 12.

Do this per country. The German market searches for Flaggenquiz as one word, the French for quiz drapeaux, and the Japanese for 国旗クイズ. None of these are translations of each other in volume or difficulty. And since Guided Search and Ask Play reward intent, the plan should carry phrases with an intent word, quiz for kids, offline maps for hiking, not only head terms.

Every term goes in exactly once. Title first, because it weighs most. Subtitle next. Keyword field for everything else, comma separated, no spaces, no plurals if the singular is there. Count the characters: 30, 30 and 100 on the App Store. A title at 30 of 30 with your top term in it beats a clever name every time.

Then the parts models read. The first paragraph of the description states plainly what the app is, for whom, and what it contains, because that is what App Store Tags and Ask Play are generated from, and it should match the first paragraph of your website word for word. Then screenshots: the first three carry the promise in six words or fewer each, using the same words as the title, because captions are indexed and the first three screenshots appear directly in search results. People read three screenshots and decide.

With keyword linking and 70 pages per app, the default listing is only the fallback. Group your target terms into intents, at most ten per country, and give each intent a page whose first three screenshots and captions show that intent. The timer page shows the timer. The rain page shows rain. The same page serves the ad click and the organic click, and category benchmarks put the conversion lift of a matched page between five and nine percent.

Change the title alone, and measure impressions and conversion for 14 days before you touch the subtitle. That is the only way to learn what worked. One title change on MapZap brought 2.1 times the impressions in 14 days with nothing else touched, and that number is only knowable because nothing else was touched. App Store Connect's benchmarks, added in March 2026, let you compare each change against your category rather than against last week.

Reviews are the cheapest research you have. When four reviews in a month ask for the same thing, that is a screenshot caption, a subtitle candidate and a content hook. On Google Play the reviews now also become the users are saying paragraph, an AI summary under your rating that every visitor reads. Answer every review, and read them as a list of things the store page failed to say.

Because retention and rating recency feed the ranking, the push prompt, the onboarding and the rating prompt belong in the ASO plan. Ask for a rating after the first win in the app, not at launch. Ask for push permission the same way. Fix the largest drop in the retention curve before buying any install, since a bought user who leaves on day three costs rank as well as money.