The Two Jobs of Metadata Metadata serves two distinct functions that are easy to conflate but operationally separate. The first is indexing — ensuring the platform algorithms understand which search queries should surface your app. This is mechanical work: distributing keywords across the fields each platform indexes, respecting character limits, avoiding duplication, and aligning with how the algorithm parses and combines terms. The second function is persuasion. A user who lands on your product page has already passed the relevance filter. Now the listing needs to answer a single question in under ten seconds: what does this app do for me, and why should I install it? This is where description copy, feature enumeration, and benefit framing matter — not for ranking, but for conversion. These two jobs depend on each other. Strong search visibility without conversion-optimized copy yields impressions that do not convert to installs. A compelling product page that ranks poorly for high-intent queries remains invisible to most of the addressable audience. Both must work in concert. ## Platform Divergence: iOS and Android Require Separate Strategies The iOS App Store and Google Play have fundamentally different indexing models, and treating them identically is one of the most common optimization errors we observe. ### iOS: Field-Based Indexing with Hidden Keywords Apple's App Store indexes three primary text fields for search: the Title (30 characters), Subtitle (30 characters), and a hidden Keyword Field (100 characters, comma-separated, no spaces). Apple's algorithm combines terms across these fields within a single locale, which means duplication wastes indexing potential. A keyword that appears in both the title and the keyword field does not yield additional ranking surface area — it simply occupies characters that could have been used for new terms. The Description field is not indexed for search on iOS. Its sole purpose is conversion: explaining features, building trust, and driving the install decision. This distinction is critical. Writing an iOS description optimized for keyword density is wasted effort. Apple also allows multiple language localizations per country. In the United States, for example, you can create separate store listings for English, Spanish (Mexico), Portuguese (Brazil), Russian, and several other languages. Each locale generates its own internal keyword combinations, effectively multiplying your indexable search surface without additional app versions. This is a lever most developers leave untouched. ### Google Play: Full-Text Indexing with Natural Language Processing Google Play indexes the Title (30 characters), Short Description (80 characters), and Full Description (up to 4,000 characters). Unlike iOS, Google's algorithm reads the full description and extracts keywords from natural prose. The platform applies semantic understanding — it recognizes synonyms, related terms, and user intent beyond literal keyword matching. Keyword density matters, but stuffing degrades both readability and ranking. The practical guideline we track: roughly one exact keyword occurrence per 250 characters in the full description, integrated naturally into feature explanations and benefit statements. Additional ranking signals include user reviews (which are fully indexed), the developer name, and the app's package URL. Since 2025, Google has weighted wiki:retention-rate and app stability more heavily than raw install velocity. An app that acquires users quickly but loses them within days will rank lower than one with slower growth but higher Day 7 and Day 30 retention. This shift fundamentally changes how acquisition and product quality interact with organic visibility. ## Keyword Strategy: Research, Selection, and Distribution Effective wiki:keyword-research is not a one-time activity. It is an ongoing cycle: gather candidate terms, evaluate relevance and competition, distribute them across indexed fields, monitor performance, and iterate based on what ranks and what does not. ### Keyword Sources We gather keywords from five primary sources: - Functional vocabulary — the terms users naturally employ when describing the problem your app solves or the task it performs - Competitor metadata — analyzing what top-ranking apps in your category index for, particularly terms you may have overlooked - Autocomplete suggestions — the live vocabulary users type into store search fields, surfaced through platform autocomplete - User reviews — the exact phrasing real users employ when describing features, pain points, or use cases - Seasonal and trending terms — queries that spike during specific periods or in response to cultural or market shifts ### Evaluation Criteria Each candidate keyword is assessed on four dimensions: 1. Relevance — Does the term accurately describe what the app does? Irrelevant keywords yield low-intent traffic that does not convert. 2. Search volume — How many users search for this term monthly? Higher volume means more addressable impressions. 3. Competition — How many strong apps already rank for this keyword? High-volume terms dominated by established brands are harder to crack than moderate-volume terms with weaker competition. 4. User intent — Is the searcher ready to install, or are they still researching options? Long-tail keywords like "remove background from photo" convert better than broad terms like "photo editor" because the user's need is more specific. ### Field Distribution On iOS: - Title — Brand name plus one or two primary keywords that define the app's core function - Subtitle — Unique secondary keywords that complement the title without repeating any terms - Keyword Field — All remaining high-value terms, comma-separated, no duplication with title or subtitle, singular forms only (Apple matches plurals automatically) - Additional locales — Distribute entirely separate keyword sets across non-primary language localizations to maximize internal keyword combinations On Google Play: - Title — Primary keyword integrated naturally with the brand - Short Description — Two to three high-priority keywords woven into a benefit-focused hook - Full Description — Natural prose that explains features and benefits, with strategic keyword placement (roughly one occurrence per 250 characters for top terms) ## Visual Metadata: Icons, Screenshots, and Video Visual assets are not decorative. They are the primary conversion lever once a user reaches your product page. Studies consistently show that wiki:screenshot quality is the single largest determinant of install rate among users who view a listing. ### App Icon The icon appears in search results, top charts, recommendations, and on the user's home screen after install. It must be instantly recognizable at the smallest rendering size (29×29 pixels on iOS) and visually distinct from competitors. Effective icons use one or two colors maximum, avoid fine details or text, and communicate the app's function or brand identity at a glance. ### Screenshots Both platforms allow up to 10 screenshots per device size. Use all available slots. The first two or three screenshots are visible without scrolling in most contexts — these must communicate your core value proposition immediately. Lead with outcomes, not features: "Save 2 Hours Every Week" converts better than "Calendar Sync Feature." A recent development: both Apple and Google now index text overlays on screenshots for search relevance. This means screenshot captions serve a dual purpose — they persuade users visually and contribute to keyword indexing. A screenshot showing workout tracking should include a caption like "Track Every Workout Automatically," not generic placeholder text. Screenshots should be localized. Displaying English-language captions to non-English audiences introduces friction and reduces conversion, particularly in markets with low English proficiency. ### Video On iOS, app preview video content autoplays silently in search results. On Google Play, the promo video appears at the top of the listing. In both cases, the first three seconds determine whether a user continues watching or scrolls past. Open with the most compelling feature or the problem the app solves — not a logo animation or splash screen. Keep runtime under 20 seconds even though longer durations are permitted. Add captions, since most users browse with sound off. ## Localization: Language-Specific Keyword Research, Not Translation Localization is the most underutilized growth lever in app store optimization. Fewer than 2% of developers fully localize their listings, yet apps localized in 10 or more languages see an average 30% increase in downloads per locale. The critical error most teams make: treating localization strategy as translation. Direct word-for-word translation of keywords from English to other languages almost always misses the terms users actually search for in those markets. The top search term in English is rarely the top term in Japanese, German, or Spanish. Each locale requires independent keyword research to identify the vocabulary that market uses. Apple's expansion to 50 supported languages in early 2026 — including nine additional Indian languages (Bangla, Gujarati, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, Telugu) plus Slovenian and Urdu — creates new opportunities for developers willing to invest in proper localized metadata optimization. Each new language unlocks a separate keyword space with its own search volume and competitive dynamics. Cultural adaptation goes beyond vocabulary. Tone, messaging emphasis, and visual references must align with local expectations. A promotional message that resonates in the United States may feel overly aggressive in Japan, where softer, benefit-focused language performs better. ## 2025-2026 Algorithm Shifts: What Changed and Why It Matters Two major platform changes in 2025 reshaped how metadata optimization works. ### Custom Product Pages Enter Organic Search (iOS) In July 2025, Apple introduced keyword linking for custom product pages cpp. Previously, CPPs functioned only as landing pages for paid ad campaigns. Now they appear in organic search results when users query specific keywords tied to individual CPPs. This fundamentally changes how iOS metadata strategy works. You can now create up to 70 separate product page variations, each optimized for a distinct keyword cluster or user intent. A fitness app can show running-focused screenshots and copy to users searching for "run tracker," and strength-training-focused assets to users searching for "workout log" — all within organic search, not just paid traffic. The mechanics of keyword distribution across CPPs are still being tested. It remains unclear whether Apple's algorithm prioritizes exact-match keywords over combinations, how overlapping keywords between CPPs are resolved, and whether CPPs compete with the default listing for shared terms. What is clear: this is the most significant expansion of organic optimization surface area on iOS in years. ### Google Play Shifts from Installs to Retention Google's 2025 algorithm update de-emphasized raw install velocity in favor of user retention and engagement. Apps that acquire users quickly but lose them within days now rank lower than apps with slower growth but stronger Day 7 and Day 30 retention. This shift blurs the traditional boundary between acquisition and product quality. Metadata can drive visibility and installs, but if the product experience does not retain users, organic rankings will decline. ASO and product-market fit are no longer independent variables. Google also introduced Guided Search, which organizes results by user intent rather than literal keyword matches. Users increasingly input goals ("find housing," "track calories") rather than specific app names or features. The algorithm interprets intent and categorizes results accordingly. Metadata optimization must now account for the task or problem the user is trying to solve, not just the literal keywords they type. ## Maintenance Cadence: Iteration Over One-Time Setup Metadata optimization is not a launch-day checklist. It is an ongoing discipline. Keyword rankings shift. Competitors update their listings. Seasonal trends emerge. Platform algorithms change. The operational rhythm we track: - After every metadata update — Monitor keyword position changes within 48-72 hours to isolate what worked and what did not - Every 2-4 weeks — Reassess keyword distribution, test new high-opportunity terms, retire underperforming ones - Before seasonal peaks — Adjust metadata to capture trending queries (e.g., tax-related keywords in March-April for finance apps, fitness keywords in January) - After major competitor updates — Analyze what changed in top-ranking competitor listings and whether those changes affected your own rankings Change one variable per update cycle. Testing multiple hypotheses simultaneously (new keywords + new screenshots + new locales) makes it impossible to attribute performance shifts to specific changes. ## Pre-Launch and Ongoing Compliance Checks Both Apple and Google enforce strict content policies and technical requirements. Violations trigger rejections or, worse, listings that go live but underperform due to preventable errors. Key validation points: - Character limits — Verify every metadata field respects platform-specific limits, particularly in non-English languages where the same meaning may require