FunnelHound / Guides / Keyword localization
App Store keyword translation vs localization: re-research, don't translate
App Store search runs a separate index per storefront: blank localized metadata means you're invisible there, no matter where your app is "available". And the big trap: translated keywords rank you for dictionary words nobody types. Localization is re-research, not translation.
One search index per country
"Available in 175 countries" and "discoverable in 175 countries" are different products. Each storefront indexes its own localized title, subtitle and keyword field; leave a locale blank and local searches simply don't find you. The upside is enormous for small apps: because most developers never localize metadata, keyword competition outside the US is dramatically lower: the same keyword that's a bloodbath on the US storefront can be nearly uncontested in Brazil or Korea.
The translation trap
Machine translation returns the textbook word. People search the way they talk. The classic example: a plant identifier translated for Germany gets "Pflanzen Identifikator": grammatically fine, and nearly nobody types it. Germans search "Pflanzen bestimmen". Same app, different words, an order of magnitude apart in traffic. That's why the keyword field must be researched by someone (or some tool) that knows what locals actually type. MT is acceptable for body copy, never for the keyword field. Rank #1 for a phrase no one searches and you've built a monument, not a channel.
Keyword translation vs keyword localization
| Keyword translation | Keyword localization |
|---|---|
| Keeps the meaning of your English keywords | Keeps the search intent of local users |
| Output: dictionary-correct terms | Output: terms people actually type, slang and misspellings included |
| Tool: a translator or MT | Tool: local autocomplete, Apple Search Ads popularity, a native speaker |
| Fine for description and screenshots | Required for title, subtitle and the keyword field |
How to translate keywords when you have to
No native speaker, no budget, 40 locales staring at you. Translate anyway, but as a first draft, not the final field:
- 1. Machine-translate each English keyword into three to five candidate terms, not one. Ask for synonyms and the colloquial form, not just the dictionary form.
- 2. Open the App Store on a device set to that storefront and type each candidate. Autocomplete suggestions are the real search vocabulary; a candidate with no suggestions is a dead keyword.
- 3. Check the Apple Search Ads popularity score for the survivors (free with an Apple Search Ads account, no campaign needed). Keep the ones with a score, drop the rest.
- 4. Pack the keyword field as usual: 100 characters, commas, no spaces, no words already in the title or subtitle.
- 5. Ship, wait 2 to 4 weeks, and watch which locale moved before translating the next batch.
That is still re-research. The translator only supplies candidates; the storefront decides.
The free wins: English locales you already speak
Before touching a translator: English (U.K.) and English (Australia) are indexed separately from U.S. English. Each is a fresh 30-character title, subtitle and 100-character keyword field, in a language you already write. Same logic for Spanish (Mexico) vs Spain and Portuguese (Brazil) vs Portugal: distinct locales, distinct indexes. Filling en-GB is the cheapest ASO win available: one afternoon, no budget, more indexed keywords (the character-budget mechanics are in the keyword field guide).
Which markets are worth real research
- Japan: top-three revenue market, the highest per-user spend, and low localization competition.
- Germany & France: top-ten revenue, established paying users.
- Brazil & Mexico: huge download volume; pt-BR and es-MX specifically, not the European variants.
- South Korea: high revenue, and the hangul barrier keeps most competitors out entirely.
Metadata for every locale ships in one metadata-only submission, no new binary needed when only the version fields change.
Localize the pitch, not just the keywords
Keywords get you found; the listing still has to convert. A user in Brazil should feel the app was built in Brazil. That means the subtitle and screenshot captions get rewritten per country, not run through a translator. Discovery is only half the funnel; a localized keyword pulling people to an English-only page leaks at the conversion step instead.
Judge each locale like a separate experiment
After shipping, the totals will lie to you: a win in Mexico hides inside a flat global number. Baseline before submitting, give indexing its 2 to 4 weeks, then check which countries actually moved and double down there instead of translating 40 more languages on faith. The full discipline is in how to tell if your ASO change worked; FunnelHound's funnel view from App Store Connect makes the before/after readable from your phone.
See which markets your localization actually moved
FunnelHound tracks impressions → installs → purchases from App Store Connect, so a win in one storefront doesn't hide in the totals.
Get FunnelHoundData notes: per-locale indexing, field limits and metadata-only submissions per Apple's App Store Connect documentation as of 2026. Market rankings are directional, drawn from public store reports. Locale cross-reading behavior (which storefront reads which locales) shifts over time. Verify current behavior before betting a launch on it.