# Localizing app keywords: why translation is not enough

> People in each country search in their own words. How to find the phrases locals type, test them in each store and fit them into a localized listing.

- URL: https://asomaster.app/blog/localizing-app-keywords/
- Published: 2026-10-27
- Updated: 2026-10-27
- Category: Localization
- Publisher: ASOMaster (Cressu OÜ), https://asomaster.app/

## Key takeaways

- People search with local habits, loanwords and compounds, so a correct translation is often not the phrase with demand.
- Collect two to four candidates per keyword and market from store suggestions, top local apps and native speakers.
- Test every candidate in its own market's store: rank, demand and difficulty.
- Each App Store localization has its own 30/30/100-byte fields; Japanese, Chinese and Korean keyword fields hold about 33 characters.
- Localize screenshots and visible copy too, then measure ranks and conversion per market.

Most apps that expand abroad translate their English listing and stop there. The result reads correctly and ranks for almost nothing, because people don't search with dictionary translations. A German looking for white noise types "weißes Rauschen", but a French user looking for a budget app is more likely to type "gestion budget" than a literal "planificateur de budget", and Japanese users often type the English loanword in katakana instead of a native term. Localizing keywords means finding the phrase each market uses, checking it in that market's store, and writing a listing around it.

This guide covers why translation fails, how to find real local phrases, how to test them, and how to fit them into a localized listing, with examples for a fictional sleep-sounds app called Tidepool.

## Why doesn't translating keywords work?

Because people search with habits, not dictionary entries. A literal translation can be correct and still rarely searched, many markets search the English loanword, compounds and inflections change how a phrase is written, and regional variants use different words. Only a search in that market's store shows which phrase has demand and relevant results.

There are four common ways a translated keyword goes wrong.

**1. The literal translation isn't what people type.** Translation tools aim for meaning; searchers use habit. A phrase can be correct and still almost unused.

**2. Loanwords win.** In many markets, people search the English term or a transliteration of it. In Japan, "ホワイトノイズ" (white noise written in katakana) is a natural search; a native descriptive phrase might not be. In Germany and the Nordics, plenty of tech and app vocabulary is searched in English.

**3. Word forms change.** Languages with cases, compounds or gender change how a phrase looks. German forms compounds: "Einschlafhilfe" (sleep aid) is one word, so splitting it into "Einschlaf" and "Hilfe" in a keyword field may not match. Slavic languages inflect nouns heavily.

**4. Regional variants differ.** Spanish in Spain and Spanish in Mexico use different words for many everyday things. Portuguese in Brazil differs from Portugal. Apple's App Store treats these as separate localizations for a reason.

Because of all four, the only reliable test of a local phrase is whether it brings up relevant apps and has real demand in that market's store.

## Step 1: decide which markets deserve the work

Localizing well takes effort per market. Prioritise:

1. **Markets where you already get downloads.** Check App Store Connect or Play Console analytics by country. Even unlocalized, some markets bring installs; a localized listing usually helps them most.
2. **Large markets with moderate competition.** Market size alone misleads: a huge market where your category is dominated by a few giants may be worth less than a mid-sized one with weak competitors.
3. **Markets that match your app.** A sleep-sounds app travels well. A budget app tied to US tax rules doesn't.

A sensible first wave is five to ten markets. ASOMaster ranks markets by an estimated potential score (audience size, buying power and access) and tracks 18 by default, but your own download data should have the final say.

## Step 2: find candidate local phrases

For each of your core English keywords, collect two to four candidate phrases per market. Sources, from most to least reliable:

- **The store's search suggestions in that country.** Set a device or test account to the storefront and type the start of the phrase. Completions show what locals type.
- **Top apps in that market.** Read the names and subtitles (App Store) or titles and short descriptions (Google Play) of the top ten local apps for your category. Repeated words are strong signals.
- **Native speakers.** A colleague, customer or translator who uses apps in that language. Ask "what would you type to find an app that does X?", not "translate X".
- **Machine translation, last.** Useful to generate candidates, never to choose them.

For Tidepool's core keyword "white noise":

| Market | Candidates |
|---|---|
| Germany | weißes Rauschen, white noise, Rauschen |
| France | bruit blanc, white noise |
| Japan | ホワイトノイズ, 白色雑音 |
| Brazil | ruído branco, som branco |
| Spain | ruido blanco, ruido blanco para dormir |
| Korea | 백색소음, 화이트노이즈 |

## Step 3: how do you test a local phrase?

Search it in that market's store and record three numbers: your rank, the demand behind the phrase and the difficulty of its top ten. Compare candidates only within the same market, and use Apple Ads or Google Ads popularity as the tie-breaker when you have it, because estimates alone can't separate close calls.

Test every candidate the same way you test English keywords, in that market's store, not in the US store.

| Market | Phrase | Rank | Volume (estimate) | Difficulty (estimate) |
|---|---|---|---|---|
| Germany | weißes Rauschen | 4 | 58 | 35 |
| Germany | white noise | 31 | 39 | 52 |
| Japan | ホワイトノイズ | 44 | 61 | 72 |
| Japan | 白色雑音 | none | 6 | 18 |
| Brazil | ruído branco | 7 | 55 | 29 |
| Brazil | som branco | none | 12 | 20 |

The pattern is typical. In Germany, the German phrase beats the English one on demand. In Japan the katakana loanword has real demand while the native term has almost none, despite being a correct translation. In Brazil the obvious translation wins. Without measuring, you'd have no way to know which case you're in.

![One keyword with its local phrase, rank and metrics in each market](https://asomaster.app/shots/07-keywords-by-market-light-1600.webp)
Keep estimates and real numbers apart here too. If you have Apple Ads or Google Ads accounts, they report real popularity per country, which is the best tie-breaker between two candidates. See [real popularity](https://asomaster.app/docs/popularity/) for connecting them, and [metrics explained](https://asomaster.app/docs/metrics/) for what the estimates mean.

In ASOMaster, adding a keyword writes a local phrase for every tracked market and checks it in that store; you can replace any phrase with one of your own candidates and check again. The point is the same whatever the tool: a local phrase is a hypothesis until a search in that store confirms it.

## Step 4: pick a winner per market

Choose the phrase with the best combination of demand and difficulty that genuinely describes your app. A simple rule of thumb:

1. Prefer the phrase with higher real popularity when you have it.
2. Between similar demand, prefer lower difficulty.
3. Prefer a phrase where you already rank in the top 30: you have relevance to build on.
4. Penalise reusing the English phrase in a non-English market unless the measurements clearly favour it.

For Tidepool: "weißes Rauschen" for Germany, "ホワイトノイズ" for Japan, "ruído branco" for Brazil.

## Step 5: write the localized listing

Each localization has its own fields. On the App Store, that's a name (30 characters), subtitle (30) and keyword field (100 bytes); on Google Play, a title (30), short description (80) and full description (4,000). Write each in the target language, using the winning phrases.

A localized App Store listing for Tidepool in German:

- Name: `Tidepool: Einschlafgeschichten` (30/30)
- Subtitle: `Weißes Rauschen & Naturklänge` (29/30)
- Keywords: `regen,meer,wellen,baby,schlaf,einschlafen,braunes,einschlafhilfe,ventilator,schlaflos` (85/100 bytes)

Notes on the German version:

- The brand stays; the descriptor is a researched German phrase ("Einschlafgeschichten", bedtime stories), not a translation of "Sleep Sounds".
- Compound words appear whole where they're searched whole ("einschlafhilfe", sleep aid).
- No word repeats across the three fields: "Rauschen" is already in the subtitle, so the keyword field doesn't spend bytes on it.
- Lower case in the keyword field is fine; search isn't case sensitive.

### Byte limits in non-Latin scripts

Apple limits the keyword field to 100 bytes. Most Latin characters take one byte, but accented letters such as "ü" or "é" take two, and Japanese, Chinese and Korean characters take three each. A Japanese keyword field holds about 33 characters. Spend it on the highest-value phrases and rely on the name and subtitle for the rest.

### More than one localization can count

On the App Store, a storefront can support more than its default language. Apple's [App Store localizations reference](https://developer.apple.com/help/app-store-connect/reference/app-information/app-store-localizations) lists, for each country or region, a default language and additional supported languages. Canada, Switzerland and Belgium list several; the United States lists Spanish (Mexico), Chinese, French, Korean, Portuguese (Brazil), Russian, Vietnamese and Arabic as additional languages.

This matters in two ways. First, people in multilingual markets may search in more than one language, so a French listing can matter in Canada. Second, developers often use an additional localization's fields to cover more search terms in a storefront. Treat that second use with care: the localization still needs a real name and subtitle in its language, because some users will see it, and whether its words count for search in a given storefront is something to confirm with rank checks rather than assume. Apple's [localize app information](https://developer.apple.com/help/app-store-connect/manage-app-information/localize-app-information) page covers adding localizations in App Store Connect.

## Step 6: localize the visible copy too

Ranking brings people to your listing; the rest of the page decides whether they install. For each priority market:

- Translate and adapt screenshots and their captions, not just the text fields. Captions in the wrong language undo much of the work.
- On Google Play, localize at least the opening of the full description, since it's indexed and visible.
- On the App Store, localize the promotional text and description for conversion, even though they don't affect search.
- Check that the app itself is localized, or say clearly in the description that it isn't. A localized listing for an English-only app leads to disappointed users and poor ratings.

## Step 7: measure per market

After release, check ranks per market after one and two weeks, the same way you do for your main market. Expect differences:

- Markets where you had no localization before often show the largest jumps.
- Some phrases won't move. Either the competition is too strong or the phrase isn't what people search. Go back to step 2 for those.
- Conversion can change as much as ranks. Watch downloads per product page view by country in your store analytics.

![A table of languages, each with its localized name and subtitle and a status](https://asomaster.app/shots/13-listing-studio-languages-light-1600.webp)
## Can you localize keywords at scale?

Yes, if tools do the measuring and people do the judging. Generating candidates, searching each one in its own store and picking winners by the same rule all scale well. Deciding whether a phrase fits your app and whether the visible copy reads naturally still needs someone who knows the product and the language.

Done by hand, this process takes a few hours per market. That's fine for five markets and impractical for forty.

ASOMaster's Listing Studio localizes a chosen listing to up to 49 languages by proposing local candidates per language, searching each in that store, and picking the winner by demand and difficulty, then writing the localized fields; you review each language before anything is published. It runs on your own AI subscription. More on the [App Store localization tool](https://asomaster.app/app-store-localization-tool/) page.

## Checklist

1. Choose five to ten markets from your download data and market potential.
2. For each core keyword, collect two to four local candidates per market from search suggestions, top apps and native speakers.
3. Measure every candidate in its own store: rank, demand, difficulty, and real popularity where available.
4. Pick a winner per market by demand, difficulty and fit.
5. Write each localization's fields in its own language, within its own limits.
6. Localize screenshots and visible copy for conversion.
7. Measure ranks and conversion per market after one and two weeks.

For the English side of the same process, start with the [keyword research guide](https://asomaster.app/blog/app-store-keyword-research/) and the guide to [writing the name, subtitle and keyword field](https://asomaster.app/blog/app-store-name-subtitle-keyword-field/).