Why translating your US keyword field word-for-word loses rankings
Suppose your English app targets budget planner,expense tracker,bill reminder (44 bytes). If you translate those words literally into another language without checking local App Store search data, three things go wrong:
- Different everyday vocabulary: In Türkiye, users searching for an expense or budget app type
bütçe takibi,gelir gider takibi, orfatura hatırlatıcı—not a literal dictionary translation of “planner.” - Hybrid English + Native queries: In Germany, Japan, and Korea, users frequently search for English loanwords (`habit tracker`, `pomodoro`, `pdf scanner`) alongside native terms. Only checking live storefront rankings and `searchPopularity1to100` shows which form local users actually prefer.
- UTF-8 byte expansion: A 98-character English keyword list translated into German, Turkish, Russian, Arabic, or Japanese will almost always exceed Apple’s 100 UTF-8 byte limit (`120–240 bytes`).
How Searchbop fixes Apple’s cross-language suggestion pollution
When you query Apple’s `/v1/suggestions/keywords/query` endpoint for an app in the US, UK, or Türkiye, Apple frequently returns suggestions in completely unrelated scripts (such as Arabic, Cyrillic, Hebrew, or Chinese terms that happen to co-occur across global app metadata) even though you passed `countriesOrRegions: ["TR"]` or `["DE"]`.
To keep your keyword research table clean, Searchbop runs a two-layer filter in [`research.ts`](file:///Users/tarik/Desktop/TengrX-SEO/packages/core/src/aso/research.ts#L42-L52):
| Filter Layer | Where It Runs | What It Drops |
|---|---|---|
| 1. Deterministic Script Filter ( fitsCountryScript(term, lang)) |
Locally in JavaScript before saving or scoring (`countries.ts`) | Immediately drops terms containing non-Latin scripts (`Cyrillic`, `Arabic`, `Hebrew`, `Han`, `Hiragana`, `Hangul`, `Thai`, `Greek`) when researching a Latin-script storefront (e.g. `en`, `tr`, `de`, `es`, `fr`), while permitting Latin + native script in CJK/Arabic/Cyrillic storefronts. |
| 2. AI Semantic & Language Score ( scoreRelevance(llm, app, lang, terms)) |
Batched JSON call to your connected AI model (`0.0–1.0`) | Scores `0` whenever a candidate term is in a foreign language other than the storefront’s native language (or English) or is a competitor’s brand name; keeps only terms scoring `>= 0.35`. |
How many keywords actually fit inside 100 UTF-8 bytes per language?
Because Apple’s Keyword field is capped at 100 UTF-8 bytes, your keyword density changes depending on the storefront’s character encoding:
- English (`en-US`, `en-GB`, `en-AU`, `en-CA`): `1 byte` per ASCII character → ~12 to 15 comma-separated words fit comfortably.
- Turkish, German, Spanish, French, Portuguese, Polish, Vietnamese: Base Latin letters cost `1 byte`, accented letters (`ğ, ş, ı, ö, ü, ç, ä, ß, ñ, é`) cost `2 bytes` → ~10 to 13 words fit.
- Arabic, Hebrew, Russian, Ukrainian, Greek (`ar`, `he`, `ru`, `uk`, `el`): Every letter costs `2 bytes` in UTF-8 → ~7 to 9 words fit (~50 characters total including `1-byte` commas).
- Japanese, Korean, Simplified/Traditional Chinese (`ja`, `ko`, `zh-Hans`, `zh-Hant`): Every CJK character costs `3 bytes` in UTF-8 → ~30 characters total fit inside 100 bytes.
Extracting real local phrasing from native top-10 apps
Instead of guessing translations, add your app to Searchbop for your target country (for example, `DE` for Germany or `JP` for Japan) and use Competitor Title & Listing Extraction. Searchbop queries `itunes.apple.com/search?country=de&lang=de_de`, pulls the native German titles and descriptions of the apps actually ranking in Germany, and extracts the exact German 1-to-3-word phrases local developers use.
Honest boundaries: 91 supported Apple Ads countries
Why 91 countries? Searchbop supports the exact 91 App Store countries and regions where Apple Ads operates (defined in [`packages/core/src/aso/countries.ts`](file:///Users/tarik/Desktop/TengrX-SEO/packages/core/src/aso/countries.ts)), ensuring that store search rankings, language/script rules, and Apple Ads popularity queries all map to the same verified storefront list.
Research context & search queries covered on this page Multi-Country ASO · P1
Cluster & intent: aso-localization (Informational / Commercial · P1) · Target queries: app store localization keywords, international app store optimization, multi country aso, ios storefront keyword research.