The only Polish ad copy in this whole category is running in Portugal.
Across 11,022 App Store ad records and 118 distinct headlines, swept from 29 entities in all 25 EU storefronts, the language tag pl never appears once. Poland gets English from every advertiser that runs there. The single Polish sentence in the file belongs to tonestro, it is tagged pt-PT, and it runs in the Portugal storefront.
All of this comes out of Apple's Ad Repository, the public dataset the EU Digital Services Act requires Apple to publish. It records the headline, the language tag and the country of every App Store ad, so the translation setup of an entire category is readable from outside. Twenty two advertiser apps, all 25 EU storefronts, zero failed country queries across 725 country slices, data window ending 26 July 2026.
The Polish headline in Portugal
tonestro, published by fun.music IT GmbH, is the third heaviest advertiser here at 3,299 repository records, and one of only two lessons apps with volume in all 25 storefronts. It ran a single English headline, "Learn to play an instrument", everywhere, then localized in two dated waves in May 2026: France and Portugal on 19 May, then Germany, Italy, Portugal and Spain on 29 May.
Four of the five translations landed where you would expect.
| Storefront | Tag | Headline |
|---|---|---|
| FR | fr-FR | "Apprends un instrument" |
| AT, DE | de-DE | "Lerne ein Instrument" |
| IT | it | "Impara a suonare" |
| ES | es-ES | "Aprende a tocar" |
| PT | pt-PT | "Naucz się grać" |
That last line is Polish. It means learn to play.
Everything else on the creative is correct Portuguese. The app name is the localized one, tonestro with the Portuguese subtitle Aulas de Música, the promotional text is Portuguese, the tag is pt-PT, and the only storefront it runs in is PT. Two creative groups, 15 records, first recorded 19 May 2026, still recorded at the window end.
We verified it twice, because a finding like this is usually a grouping artifact. Once in the grouped sweep, and once by pulling the raw Portugal slice and reading the fields off individual records, where the row prints as PT | pt-PT | tonestro - Aulas de Música | "Naucz się grać" | 2026-07-19 to 2026-07-26. Both passes agree.
Poland, meanwhile, takes 215 tonestro records, and all 50 the repository returned are English: PL | en-GB | "Learn to play an instrument".
What the data supports is that the copy is Polish, the tag is pt-PT and the storefront is Portugal. Whether that is a paste error, a leftover from a market that never launched, or something intentional, the repository does not say, and we are not going to guess.
Eight languages, and the list stops there
Eight language tags exist in the entire 11,022 record dataset: en-GB, de-DE, fr-FR, it, es-ES, pt-PT, nl, sv. That is the whole set.
Which means eight storefronts get copy in the local language: AT, DE, ES, FR, IT, NL, PT, SE. German is the only tag that covers two storefronts, Austria and Germany.
And seventeen only ever see English: BE, BG, CY, CZ, DK, EE, FI, GR, HR, HU, IE, LU, LV, PL, RO, SI, SK.
None of the 22 apps in this sweep has written a Polish, Czech, Greek, Danish, Finnish, Hungarian, Romanian, Croatian, Bulgarian, Slovak, Slovenian, Estonian, Latvian, Lithuanian or Irish line in the past year. Belgium never gets Dutch or French, not even from publishers who already write both. Denmark, Finland and Poland sit in the English tier next to Cyprus and Luxembourg.
tonestro and Ultimate Guitar, the two widest advertisers, serve the identical set of 19 storefronts in English only: BE, BG, HR, CY, CZ, DK, EE, FI, GR, HU, IE, LV, LU, NL, PL, RO, SK, SI, SE. Poland alone takes 215 tonestro records and 221 Simply Piano records, every one of them in English.
The order publishers translate in
Counting the per advertiser map, the languages stack up in a strict order. Each number is how many of the 22 apps run at least one line in that language.
| Language | Apps writing it | Storefront |
|---|---|---|
| German | 19 of 22 | AT, DE |
| French | 16 | FR |
| Spanish | 15 | ES |
| Italian | 13 | IT |
| Portuguese | 11 | PT |
| Dutch | 8 | NL |
| Swedish | 3 | SE |
| everything else | 0 | none |
Only three advertisers write Swedish at all: Moises, Tuner Pro and Sheet Music Scanner. After Swedish the ladder does not continue, it stops.
The map has room at both ends. Piano Simple Piano Recording runs in 24 storefronts and translates for none of them. Tuner & Metronome Attune runs in 25 and translates for exactly one, Portugal. At the other end, GuitarTuna runs in three storefronts only, France, Germany and Spain, and every line is in the local language, and Skoove runs one storefront, Germany, in German. The two advertisers with no English-only storefront are also two of the narrowest footprints in the table.
One caveat on reading any of this: Ultimate Guitar's top creative carries six language tags at once, de-DE, en-GB, es-ES, fr-FR, it and pt-PT, under one headline group with 683 records across all 25 storefronts. So a storefront in the English tier is a storefront where the copy that reaches it is English, not one where the advertiser filed no other tag.
Two piano apps, one German sentence
Simply Piano and flowkey are direct competitors selling the same thing to the same person. In Germany and Austria they run the same headline, character for character: "Songs und Kurse am Piano". In France they run the same one again: "Cours et chansons au piano".
Simply Piano's German line has 38 records from 29 May 2026 and was still recorded at the window end. flowkey's German line has 52 records from 28 September 2025, also still recorded. On French it is 33 records from 24 May 2026 against 26 records from 8 December 2025. flowkey was there first by eight months on German and five on French.
Both are translating the same English promise, and the translation is so literal that two companies arrived at identical strings. Simply Piano's English "Piano Songs & Lessons" becomes "Songs und Kurse am Piano" in German, "Cours et chansons au piano" in French, "Canzoni e lezioni pianoforte" in Italian, "Canciones y Clases de Piano" in Spanish, "Musicas e Aulas de Piano" in Portuguese, "Piano liedjes en lessen" in Dutch. Same promise, six translations, no change of angle in any market. flowkey does the same with "Piano songs and lessons".
Moises is the one advertiser in the table whose top creative sells a feature instead of a category. "Remove vocals or instruments", carried into German, Swedish, Dutch, Portuguese, Spanish, French and Italian. Everyone else translates a category claim.
The other pattern worth knowing before you budget for this: translation lands in batches, on single days. Ultimate Guitar launched six language variants on 17 September 2025, then repeated the move on 3 February 2026 and again on 7 April 2026. Soundcorset launched six on 8 August 2025. Metronome Tuner X launched seven on 20 November 2025, its first day in the data. Sheet Music Scanner shipped six on 5 June 2026 and three more the next day. Nobody in this sweep drips languages out one at a time.
How long the leaders waited before translating
Simply Piano ran English only in every market for 168 days before it localized anything. Its English "Piano Songs & Lessons" starts 26 October 2025 and its first non English creative, Portuguese, starts 12 April 2026. MuseScore has almost exactly the same gap: English from 28 October 2025, first non English 13 April 2026, 167 days. Chordify's gap is 122 days.
Those three are the ones Apple gives a real start date for. Any advertiser whose English creative is flagged as older than the one year window is left out of that figure, because the true start is unknown.
Four months of English in Germany and France, from two of the largest advertisers in the category, is a useful number if you are deciding whether translation blocks your launch. It did not block theirs.
The video gap
Translation is where the production budget in this category goes. Video is close to absent.
Twenty five of 253 creatives carry a video asset, 9.9%, and those account for 166 of 11,022 records, 1.5%. Three advertisers run any video at all: Sheet Music Scanner (131 of 131 records, three distinct video URLs), Sheet Music Reader Halbestunde (23 of 125) and Skoove (12 of 12, one video URL). Skoove is the smallest advertiser in the table at 12 records.
The four heaviest advertisers, 13,393 repository records between them, run none. We checked that at record level rather than trusting the rollup: Simply Piano 0 video records out of 900 retrieved, Ultimate Guitar 0 of 1,250, MuseScore 0 of 1,250.
That is a gap, not a recommendation. The repository shows what runs and how long it runs, never what converts. But if you are the one person in this category with a video, you are shooting at an open goal in a format the four biggest spenders have left alone, and localizing a video costs a great deal more than localizing a one line headline, which may be the whole reason the gap exists.
What to do with this if you are one person
- Translate German first, then French, Spanish, Italian. Nineteen of the 22 apps here write German, 16 write French, 15 Spanish, 13 Italian. Those four languages cover five storefronts, AT, DE, ES, FR and IT, and that is where competitive pressure on copy actually exists.
- Portuguese and Dutch are optional, Swedish is nearly empty. Eleven apps write Portuguese, eight write Dutch, three write Swedish. Sweden gets English from almost everyone here, including advertisers running in all 25 storefronts.
- Buy the other seventeen storefronts in English and stop thinking about it. BE, BG, CY, CZ, DK, EE, FI, GR, HR, HU, IE, LU, LV, PL, RO, SI and SK get English from every one of the 22 apps in this sweep that runs there. Poland is not an exception, it is the clearest case: 215 records from one advertiser, 221 from another, all English.
- Do not hold your launch for the translations. Simply Piano ran 168 days of English before its first localized line, MuseScore 167, Chordify 122. Ship English, read the numbers, translate after.
- Budget the translation as a line item. In this category the local version is the same sentence with the same promise. Two competing piano apps ended up with identical German and French headlines, which tells you how little room anyone is using.
- Check the language tag against the storefront before you launch. One advertiser in this sweep is running Polish copy in Portugal, first recorded in May 2026 and still recorded in late July, on an otherwise fully localized creative. Another shipped an English headline tagged
fr-FRinto France and a Portuguese headline taggedes-ESinto Spain, and ran both for months before replacing them in 2026. Nothing in the App Store flags this for you.
The thing the data cannot tell you is whether any of it converts. Longevity is the only signal the repository carries, and a line somebody keeps paying for month after month is the closest thing to a verdict you get from outside.
You can pull any advertiser in this table yourself at recao.app, free and without an account, in about a minute per app.
Method note: figures come from Apple's public Ad Repository (DSA Article 39), swept on 31 July 2026 across all 25 EU storefronts with zero failed country queries over 725 country slices. Twenty nine entities swept, 22 advertiser apps in the table, 11,022 records retrieved, 253 distinct creatives, 118 distinct headlines. Apple's returned data window runs 26 July 2025 to 26 July 2026, and the repository lags real time, so "still recorded" means recorded at 26 July 2026 and not today. Language is read from Apple's defaultLanguageTag on the creative plus the storefronts the creative ran in. English glosses of local copy are ours. A creative is grouped by placement, format, headline and asset count, and its span runs from its earliest first impression to its latest last impression. Records are repository records, a proxy for recorded activity and not for spend, installs or reach. Apple caps each country query at roughly the 50 most recent records, so for 13 of these advertisers we did not read every record: for those, first impression dates can only move earlier and campaign spans can only get longer, which makes every longevity figure here a floor rather than an exact value. Apple reports creatives older than its one year window with a marker rather than a date, so those cannot be dated more precisely. Apps with no entity in Apple's index are absent from this sweep, and absence from the index supports no conclusion about anyone's marketing. Refreshed monthly.