Four words, each planted in one place among an app's localizations, put to Apple's Search API in 27 storefronts on 2026-08-27: that is how we drew this map. The short answer is that the English (U.K.) keyword field reached 23 of the 26 storefronts outside the UK, the English (U.S.) field reached almost none, and the United States read our Spanish and French fields. Spend your keyword bytes accordingly, with the localization planner.
What we found
Seven findings, measured on 2026-08-27 across 27 storefronts and four planted words. The English (U.K.) field travels widest. The English (U.S.) field barely travels at all. Big storefronts read languages you may not expect. And a word that arrives through a foreign localization ranks low, not near the top.
- English (U.K.) reached 23 of the 26 storefronts outside the UK. The three that did not rank it: Canada, Japan and the United States. (The UK itself did not rank it either; see finding 7.)
- English (U.S.) barely travelled.
carbranked only in Mexico (182nd of 182 results) and Saudi Arabia (134th of 191). It did not rank in the United States itself. A word in the keyword field is not a promise of a match. - The United States read our Spanish fields —
cetosisranked 8th of 70 there — and our French fields, withmincir60th of 182. - Canada read the French fields:
mincir54th of 173. Canada did not read the English (U.K.) field. - Japan read neither English field.
- A word reached through a foreign localization ranked low.
slimmingcame 69th in Germany, 87th in the Netherlands, 144th in Denmark and 146th in Ukraine. Where the storefront's own language held the word, it ranked near the top:cetosis6th in Mexico, 15th in Spain. Presence and position are different things. - Two results we cannot explain. The United Kingdom did not rank the app for
slimming, though the word sat in its own English (U.K.) field. The United States did not rank it forcarb, from its own field. We write these down rather than smooth them over.
How we measured it
One app, four words, 27 storefronts, one day. Each word lived in exactly one place among the app's localizations, so a rank anywhere told us that storefront's search had read that localization. We queried one word at a time, kept negative controls, and tested cross-localization pairs to see whether Apple joins two localizations in a single query. It does not.
The app was one of our own, a fasting app. The four words:
| Word | Lived only in |
|---|---|
slimming |
the English (U.K.) keyword field |
carb |
the English (U.S.) keyword field |
cetosis |
the two Spanish localizations (Spain and Mexico) |
mincir |
the two French localizations (France and Canada) |
We put each word, alone, to Apple's Search API in Australia, Brazil, Canada, China mainland, Germany, Denmark, Spain, Finland, France, the United Kingdom, Indonesia, India, Italy, Japan, South Korea, Mexico, the Netherlands, Norway, Poland, Portugal, Saudi Arabia, Sweden, Thailand, Türkiye, Ukraine, the United States and Vietnam, and recorded where the app sat among the results Apple returned.
One word per query, never a phrase. A two-word query brings Apple's partial matching with it: fasting carb returned 177 apps in Germany where carb alone returned 156, so a pair proves nothing about either word.
Negative controls came back empty. banjo, slimming trombone, fasting trombone and cetosis trombone returned the app nowhere.
Cross-localization pairs were refused. Nine pairs joining a word from one localization to a word from another — abnehmen carb, ayuno carb, vasten slimming, intervallfasten slimming among them — returned nothing, while each of those words alone ranked. Apple matches a query against one localization at a time. That is the rule behind the practical advice further down: a phrase has to sit whole inside one localization.
How VisualASO's own readings work is in where the numbers come from. You can take the probe data (CSV) and check the cells yourself.
VisualASO, "Which localizations each App Store country searches", measured 2026-08-27, visualaso.com/research/app-store-cross-localization.
The full map
All 108 cells: 27 storefronts by four words, with the rank the app took and the number of results Apple returned for that query. Read a rank as "this storefront's search reached that localization". Read a blank as "it did not rank the app", which is weaker than "it cannot reach it" — see the limits below.
| Storefront | slimming |
carb |
cetosis |
mincir |
|---|---|---|---|---|
| Australia | #96 of 151 | not ranked (176) | not ranked (7) | not ranked (132) |
| Brazil | #83 of 164 | not ranked (170) | not ranked (7) | not ranked (131) |
| Canada | not ranked (148) | not ranked (170) | not ranked (7) | #54 of 173 |
| China mainland | #60 of 138 | not ranked (165) | not ranked (7) | not ranked (128) |
| Denmark | #144 of 153 | not ranked (180) | not ranked (5) | not ranked (129) |
| Finland | #119 of 155 | not ranked (185) | not ranked (5) | not ranked (129) |
| France | #101 of 145 | not ranked (188) | not ranked (5) | #117 of 170 |
| Germany | #69 of 147 | not ranked (156) | not ranked (5) | not ranked (131) |
| India | #75 of 162 | not ranked (187) | not ranked (7) | not ranked (130) |
| Indonesia | #65 of 157 | not ranked (192) | not ranked (7) | not ranked (130) |
| Italy | #107 of 148 | not ranked (178) | not ranked (5) | not ranked (131) |
| Japan | not ranked (143) | not ranked (183) | not ranked (7) | not ranked (130) |
| Mexico | #74 of 159 | #182 of 182 | #6 of 70 | not ranked (131) |
| Netherlands | #87 of 148 | not ranked (176) | not ranked (5) | not ranked (130) |
| Norway | #68 of 161 | not ranked (180) | not ranked (7) | not ranked (132) |
| Poland | #67 of 158 | not ranked (177) | not ranked (5) | not ranked (130) |
| Portugal | #78 of 159 | not ranked (187) | not ranked (5) | not ranked (130) |
| Republic of Korea | #48 of 163 | not ranked (186) | not ranked (7) | not ranked (130) |
| Saudi Arabia | #81 of 164 | #134 of 191 | not ranked (7) | not ranked (130) |
| Spain | #86 of 152 | not ranked (177) | #15 of 68 | not ranked (131) |
| Sweden | #98 of 150 | not ranked (167) | not ranked (5) | not ranked (130) |
| Thailand | #83 of 161 | not ranked (184) | not ranked (7) | not ranked (130) |
| Türkiye | #106 of 161 | not ranked (181) | not ranked (7) | not ranked (132) |
| Ukraine | #146 of 164 | not ranked (186) | not ranked (7) | not ranked (131) |
| United Kingdom | not ranked (144) | not ranked (182) | not ranked (7) | not ranked (135) |
| United States | not ranked (146) | not ranked (166) | #8 of 70 | #60 of 182 |
| Vietnam | #35 of 145 | not ranked (181) | not ranked (7) | not ranked (130) |
Apple's own table, and how well it predicts search
Apple publishes a table of which languages customers can be shown in each country. It says nothing about search. We checked our 108 measured cells against it anyway: the table's languages matched what search read in 103 of 108. So the table is a fair first guess for search, our measurement says, with five exceptions we can name.
Apple's App Store Connect help page App Store localizations (fetched 2026-10-29) lists a default language and "additional supported language(s)" for 175 countries or regions. English (U.K.) is the default language in 135 of those 175 rows and an additional language in 37 more, which fits how far the U.K. field reached in our probes.
Three rows worth reading closely:
| Country | Default language | Additional languages |
|---|---|---|
| United States | English (US) | Arabic, Chinese (Simplified), Chinese (Traditional), French, Korean, Portuguese (Brazil), Russian, Spanish (Mexico), Vietnamese |
| Japan | Japanese | English (US) |
| Canada | English (Canada) | French (Canada) |
The US row matches findings 3 and 4: the US lists French and Spanish (Mexico), and search read both our French and Spanish fields. The Canada row matches what Canada did: it read French and not English (U.K.), which Apple does not list there.
| Probe word | Lived in | Apple lists it, search read it | Apple lists it, search did not | Apple does not list it, search read it | Neither |
|---|---|---|---|---|---|
slimming |
English (U.K.) keyword field only | 23 | 1 | 0 | 3 |
carb |
English (U.S.) keyword field only | 0 | 2 | 2 | 23 |
cetosis |
both Spanish localizations | 3 | 0 | 0 | 24 |
mincir |
both French localizations | 3 | 0 | 0 | 24 |
The five cells where the table and search disagreed:
- The United Kingdom, which did not rank
slimmingfrom its own field. - The United States, which did not rank
carbfrom its own field. - Japan, where Apple lists English (US) but
carbdid not rank. - Mexico and Saudi Arabia, where Apple does not list English (US) but
carbranked last or near last.
Treat the table as a guess until a probe on your own app confirms it. Third-party maps of this behavior exist — MobileAction's article on cross-localization, dated 15 April 2026, says of itself that "Apple does not describe keyword indexing behavior in the documentation cited in this article". That is exactly why a measured map is worth having: nobody can read this off Apple's docs, because Apple's docs do not cover search.
What the map does not prove
One app, one category, one day. A rank shows a storefront can reach a localization; it does not show how much weight Apple gives that localization, how the behavior changes by category, or whether it held a week later. Four more limits sit on top of that, and they decide how far you should push a plan built on this page.
It cannot separate the two Spanish or the two French localizations. cetosis sat in both Spanish localizations and mincir in both French ones, so inside any one storefront the study cannot tell es-ES from es-MX, or fr-FR from fr-CA. We infer that the US reads Spanish (Mexico) because Apple's table lists Spanish (Mexico) for the US and lists no Spain row for it — an inference, not a measurement.
The ranks come from Apple's Search API, whose order can differ from the order the App Store app shows a customer.
Blank cells are soft. The UK and US own-field misses prove that a word present in a field read by that storefront can still fail to rank. So a blank tells you less than a rank does.
Re-run before spending many characters on a single claim. VisualASO's probe tool plants a word in your next version and measures which storefronts rank it, for your own app, in your own category. That turns a borrowed map into your map.
How to plan your localizations with it
Four rules follow from the findings. Write each country's default language first, because that is where a word ranks well. Add other localizations as extra keyword space, but only the ones that storefront actually reads. Never repeat a word across two localizations the same storefront reads. And keep every phrase whole inside one localization.
Your own language listing is where a word ranks well. The gap between cetosis at 6th in Mexico and slimming at 144th in Denmark is the whole argument. A foreign localization gets you into the result set; the local one gets you somewhere a customer will see.
Extra localizations are extra bytes, if the storefront reads them. Each localization carries its own 100-byte keyword field. For a US-first app, Apple's table and our probes agree that Spanish and French reach US search, so those fields are real keyword space rather than decoration. The localization planner runs this map against your own localization set and tells you which fields a given storefront reads.
Do not repeat a word in two localizations the same storefront reads. The second copy adds nothing and costs bytes you could have spent on a word you do not rank for yet. Pack each field once, deliberately: the keyword field packer counts the bytes per localization so you can see what a duplicate costs.
A phrase has to sit whole inside one localization — in its name, subtitle or keyword field — to match as a phrase. The nine refused pairs settle that. You cannot split "intermittent" into English and "fasting" into another field and expect the pair to match.
The U.K. English field reaches far but ranks low outside the UK. Use it as the wide net it is, and do not expect it to carry a competitive keyword in Germany or Japan. The rest of the keyword rulebook, including how to choose words worth the bytes, is in our App Store keywords best practices guide.
Questions
Which languages does the US App Store index?
In our 2026-08-27 probes, US search read our Spanish fields (cetosis 8th of 70) and our French fields (mincir 60th of 182). Apple's table also lists Arabic, Chinese (Simplified), Chinese (Traditional), Korean, Portuguese (Brazil), Russian and Vietnamese as additional languages for the US, which our four-word study did not test.
Should I fill in the English (U.K.) keyword field?
Yes, if you want reach: it ranked our app in 23 of the 26 storefronts outside the UK. But those ranks were low — 69th in Germany, 146th in Ukraine — so treat it as a net for storefronts you have no local listing for, not as a substitute for one.
Does the US App Store read my English (U.K.) keywords?
Not in our measurement. slimming, which sat only in the English (U.K.) field, did not rank in the United States, Canada or Japan. Apple's table agrees for those three rows: none of them list English (U.K.).
Can a search phrase span two of my localizations?
No. We tested nine pairs that joined a word from one localization to a word from another and none returned the app, while each word alone ranked. Apple matches a query against one localization at a time, so a phrase must sit whole in one localization's name, subtitle or keyword field.
How often should I re-check this map?
Before you commit many characters to one claim, and on your own app. This study is one app, one category, one day; the five cells where Apple's table and search disagreed show the behavior has edges we cannot predict from documentation.
Sources, checked 29 October 2026
- Apple, App Store localizations: https://developer.apple.com/help/app-store-connect/reference/app-information/app-store-localizations
- MobileAction, App Store cross-localization (15 April 2026): https://www.mobileaction.co/blog/app-store-cross-localization/
- Our probe data, measured 27 August 2026 (CSV): https://visualaso.com/research/app-store-cross-localization.csv