How store search and browse work
Both stores surface apps in two ways. Search returns apps for the words a person types, and most installs on both stores arrive this way for apps that are not already famous. Browse covers everything else: editorial features, charts, category pages, and the personalised placements the stores build from what a person has installed before. Search is the part a marketer can work on directly; browse is earned by quality, by newsworthy releases and by the engagement the app already has.
The stores document which fields search reads, and they differ. Apple's documentation describes the app name, subtitle and the keyword field in App Store Connect as what search matches against, with the description written for people rather than for the index. Google's Play Console documentation describes the title, short description and full description as text that affects how the app is discovered, which means the long description carries weight on Play that it does not carry on the App Store. Beyond the words, both stores say that an app's install and engagement history, its ratings and its freshness influence where it appears, and neither publishes the weights. When fewer people find the app at all, the diagnosis of organic app installs falling separates dropped search words from lost outside demand, a spent launch, ratings, broken routes and counting changes.
The comparison with web SEOImproving visibility in unpaid search results. is useful mainly for what it lacks. There are no backlinks, no site architecture, no rich results to win and no crawl to manage. The ranking inputs are a few fields of text and the app's own performance, so a listing that converts visitors into installs tends to climb for the terms it ranks for, and a listing that does not tends to sink. That is why the two halves of ASO have to be worked together: ranking without conversion is fragile, and conversion without ranking has no one to convert.
Keyword research for app stores
Store keyword research answers a narrower question than web research: which words do people type into a store when they want an app like this one, and which of those words can this app plausibly rank for. The sources are different from the web, and the volumes are usually far smaller.
| Source | What it tells you | How to use it |
|---|---|---|
| Store autocomplete | The phrasings the store expects people to continue, in the target country | Type each seed term and record the suggestions; repeat per locale |
| Competitor listings | The terms rival apps have chosen for their names, subtitles and descriptions | Read the top apps for each seed; note shared and distinctive words |
| Reviews of the category | The language people use to describe the job and the frustration | Mine reviews of rivals for nouns and verbs that are not in your listing |
| Console acquisition reports | Which sources and, where reported, which search terms produced listing visitors and installs | Compare visibility with conversion by term; find terms that show but do not convert |
| Apple Search Ads keyword data | Relative popularity of search terms on the App Store, shown for advertising | Use as a popularity signal for prioritising, even without running ads |
| Web keyword tools | Demand for the category on the web, as a proxy for store demand | Treat as a direction, not a volume; store searches are shorter and more generic |
The output is a short list grouped into three tiers: the terms the app must rank for because they describe its job, the terms it can win because competition is thin, and the brand and competitor terms it should not spend indexed fields on. A common error is to fill the fields with the highest-volume generic words, which the app cannot rank for, and leave out the specific phrases it could own. The specific phrases are usually where a new app's first store installs come from.
Title, subtitle and description
The title is the strongest indexed field on both stores and the one people see first in search results. The right pattern is the brand name followed by the clearest short statement of the job, because people who search the brand will find it either way and people who search the job need to see it in the result. Apple's subtitle and Google's short description play the same role: one line under the title that says what the app does and for whom. The craft is the same as writing a page title and meta description for the web, which the title tags and meta descriptions guide covers, with tighter limits and no second chance below the fold.
Apple's keyword field is invisible to users and exists for search; the documentation says to separate terms with commas and not to repeat words already in the name or subtitle, which wastes space. Apple's description is for people, and its promotional text can be changed without a new release, which makes it the right place for seasonal messages and launch notes. On Google Play the full description is indexed, so it should contain the tier-one and tier-two terms in natural sentences, early, without the repetition that reads as stuffing.
Both stores' review guidelines constrain the text. Apple's guidelines, which this guide cites rather than paraphrases at length, cover accuracy of metadata, the use of other apps' names and the claims a listing may make; the Play policies cover the same ground. Text that misrepresents the app or trades on another brand's name risks rejection and, repeatedly, the account. Read the current guidelines before writing, because the cost of a rejected release at launch is far higher than the cost of a dull subtitle.
Screenshots, previews and icon
Screenshots are the conversion asset. On both stores the first few are visible in search results and at the top of the listing, which means they are seen by far more people than the description, and most people decide from them. The pattern that works is a caption-led sequence: each screenshot carries one short statement of a benefit in large type above a framed screen that proves it, and the sequence tells the story of the app's job in order. The first screenshot answers "what is this and why would I want it"; the second and third show the core action; the rest cover the supporting features and the social proofEvidence that other people have chosen, used or approved something, used to reduce a buyer's doubt..
Previews (Apple) and promo videos (Google) add motion. The App Store plays previews in search results and on the listing without sound by default, so the opening moments have to work silently and the first frame has to read as a screenshot. Google Play's promo video is hosted on YouTube and shown on the listing. A preview that shows the app being used, screen by screen, outperforms one that is a brand film, because people are deciding whether the interface will suit them. For games, where the store page converts traffic from creators, trailers and ads alike, the entertainment and gaming marketing guide explains how the listing fits a launch plan.
The icon is the smallest asset and the most visible. It appears in search results, on charts and on the home screen after install, and a glanceable icon lifts click-through (CTRShare of impressions that result in a click.) from the results as much as any other change. Keep it simple, high-contrast and consistent with the screenshots' colour. Every visual asset has size and content rules documented by the store, and both stores reject assets that show devices they do not run on, content the app does not contain, or pricing claims that do not match the listing.
To decide whether a new screenshot set should replace the current one, the app store screenshot test builds the listing experiment around one priority traffic source, with installer retention as a guardrail.
Ratings, reviews and replies
The rating is shown beside the app in search results on both stores, so it works on conversion before anyone reads a review. The way to raise it is to ask the right people at the right time. Both stores provide a system rating prompt that can be shown inside the app; Apple's review guidelines direct developers to use the provided prompt rather than a custom one, and both stores prohibit incentivising ratings or filtering who is asked by how happy they look. The lever that remains is timing: ask after a moment of completed value, never at launch, never during an error, and not again for a long while if the person declines.
Reviews are read by prospective users and by the team. Both consoles let the developer reply publicly, and a reply that fixes a problem, explains a change or thanks a person is visible to everyone who reads the review afterwards. A pattern of unanswered complaints is a conversion problem in plain sight. Apple also allows a developer to reset the summary rating when a new version ships, which is worth doing only after a release that genuinely fixes what the old reviews complained about, since a reset to no ratings converts worse than a mediocre score. Google's documentation describes Play ratings as weighted towards recent reviews and shown per country, so improvement shows faster on Play and stale problems fade.
The wider practice of gathering and displaying social proof, including how to respond to negative feedback and what to show on the website, is covered in the reviews and social proof guide; the store-specific part is simply that the rating appears where the decision is made. The tutorial on app rating prompts and review replies turns these rules into a routine with named owners, from choosing the prompt moment to tracking the rating by version.
Localisation and store experiments
A listing in one language serves one market. Both stores let a developer provide localised metadata and assets per storefront, and the stores show the localised version to people whose device language matches. Localising the title, subtitle, keywords, screenshots and description for each market where the app is sold is the single largest visibility gain available to most apps after the first listing is right, because search in each storefront runs on the words people type there. Translation is not enough: the keyword research has to be repeated in each language, with local autocomplete and local rivals, and the screenshots' captions rewritten rather than translated.
Both stores also provide listing tests. Apple's product page optimisation lets a developer run treatments of the icon, screenshots and previews against the original and reports the conversion rate (CVRShare of visits or clicks that convert.) of each; Google Play's store listing experiments do the same for the icon, graphics and descriptions, with options for localised tests. These are the only reliable way to know what moves store conversion, because the counterfactual is otherwise unobservable. The discipline is the usual one for an A/B TestControlled comparison of two variants.: one hypothesis per test, enough traffic to reach a decision, and a decision made on the store's result rather than on taste. The A/B testing guide covers the method; the stores' tools apply it to a page the developer cannot otherwise instrument.
Both stores also allow custom listings beyond the default: Apple's custom product pagesA variant of an app's store page, with its own address and screenshots, built for a specific campaign, audience or feature. and Google's custom store listings let a developer build variants of the page aimed at a particular audience or campaign, with their own URL, which the app install campaigns guide uses to match the ad's promise to the page it lands on. The app marketing hub explains how that pairing fits the wider programme.
Before any of these listing changes is submitted, the app store listing checklist runs through the text, screenshots, privacy details and attribution links in one pass. To keep the listing current release after release, with an owner and a quarterly plan for listing tests, the app marketing operations playbook runs it alongside reviews, install campaigns and messaging. To judge how the listing and ratings compare with the rest of the app's marketing, the app marketing audit scorecard scores store presence next to measurement, onboarding and messaging.
Marlowe Meals rebuilds a listing in two markets
Marlowe Meals, an invented meal-planning app, ranks for its brand name and little else. Research in the UK storefront shows that "weekly meal planner" and "grocery list app" have steady autocomplete presence and thin competition; the app's title becomes "Marlowe Meals: Weekly Meal Planner" and the subtitle "Plan dinners, build the shopping list". Screenshots are rebuilt around three captions: "Plan the week in five minutes", "Shopping list writes itself", "Swap meals with one tap". Over eight weeks, impressions for the two target terms rise from a few hundred a week to around 4,200, and store conversion moves from 14% to 21%. A product page test of a preview video against static screenshots shows no reliable difference, so the team keeps the cheaper static set. A localised German listing, researched separately rather than translated, produces 38% of the app's installs in that market within a quarter, from almost none before.
In a quarterly plan the listing is the Act stage, and RACE for app installs shows store optimisation getting its own owner, budget and conversion objective alongside paid reach.
Store listings also act as landing pages for paid video, and the TikTok setup for app installs shows how to align screenshots with the promise the winning creative makes.
To measure a listing from the store console, see using the conversion rate calculator for app store listings, which works from product page views to installs per traffic source.
The effect of a better listing on paid campaigns is worked through in the cost per install walkthrough for the CPA calculator, where the store page's install rate is the conversion input.
For how much of an install plan store work should carry, app stores for app installs sets the tactics, the KPI set and a budget that puts listing work ahead of store ads.
When fewer store visitors install than before, the diagnosis of a dropped app store conversion rate reads conversion by source first, then checks screenshots, the rating, campaign landing pages and untested changes.
Common mistakes
- Filling the title with the brand alone. People who search the job never see the app.
- Repeating words across indexed fields. Each repeated word is a word not used for another term.
- Chasing generic head terms. A new app cannot rank for the category's broadest word; the specific phrases are where its first installs come from.
- Screenshots that show the interface without saying why. The caption is the message; the screen is the proof.
- Asking for a rating at launch or after an error. The prompt belongs after a moment of completed value.
- Ignoring reviews. Unanswered complaints are read by everyone who looks before installing.
- Translating instead of localising. The research and the captions have to be redone per market.
- Changing the title and subtitle every release. Ranking for a term takes time to earn and is reset by churn.
- Testing on taste rather than with the store's tools. The listing tests are the only counterfactual available.
