Disclosure up front: I work with DuoPlus, so read the tool picks here knowing that. The method itself works with any real Android device you can park in a region and leave alone. What I want to walk through is narrower than a full localization audit: the first five minutes a new user spends with your own app in three markets, and the store listing they saw right before that.
I picked Brazil, Japan and Germany because they break things in different ways. Brazil flips your decimal separator and writes dates day-first. Japan tests whether your layout survives a script with no spaces and whether your date strings know which way round year, month and day go. Germany makes every button label longer than you budgeted for and puts the currency symbol after the number. If your onboarding survives those three, it usually survives most places.
Why one phone per region, and why keep it
The tempting shortcut is one test device where you flip the system language between runs. It catches translation gaps. It misses almost everything else, because a lot of first-run behaviour hangs off more than the language setting: the region, the time zone, the network the device sits on, and what the store decides to show that device.
So the setup is three cloud phones, one per country, each with its own region and proxy settings, and each kept as a long-lived environment rather than wiped every session. A DuoPlus cloud phone is a separate Android device that keeps its own storage and sessions, and you can set it up in a local environment in the country you care about (they cover 150+ countries and regions). Keeping the phone around matters more than people expect. When something looks wrong on Tuesday, you want to reopen the exact same device on Thursday and see whether your fix landed, not rebuild the conditions from memory.
If you can afford a second pass, run the same check on Android 10, 11, 12 and 15 for at least one region, since permission prompts shift between them.
Before you start: write down what "correct" looks like
This is the step everyone skips and then regrets. For each country, write a short expected-values sheet before touching a device:
- Language and locale: pt-BR, ja-JP, de-DE. Note any strings you know are still in English on purpose.
- Currency: how a price should render. For example, Brazil and Germany both use a comma as the decimal separator, but the symbol position and spacing differ. Japan usually shows yen without decimals.
- Dates and times: day/month/year for Brazil and Germany, year/month/day for Japan, 24-hour clock in all three for most contexts.
- Store listing: which title, short description and screenshots you expect each country's store page to show.
- Legal and consent screens: anything your team has decided must appear in that market, such as a cookie or data consent prompt in Germany.
Without this sheet, testers "check" a screen and say it looks fine, because it looks fine to someone who doesn't read Japanese.
Step 1: Set up the three phones
Create one cloud phone per country and name them plainly, something like app-firstrun-BR, app-firstrun-JP, app-firstrun-DE. Point each phone's region and proxy at its country, and set the system language and time zone to match too. Then leave them that way. Don't reuse the Japan phone for a quick German check later; that's how you end up with a device whose cache and store history belong to two countries at once.
If more than one person on your team will run these checks, share the phones through team sharing instead of passing credentials around. Whoever opens app-firstrun-JP next finds it in the same state the last person left it.
Step 2: Look at the store listing from inside each country
Open the store on each phone and search for your app the way a local user would, in the local language if your title is translated. Screenshot the listing. What I've seen go wrong:
- The listing falls back to English in one country because a translation was saved as draft and never published.
- Screenshots show prices in dollars on a market where the app displays reais or euros.
- The short description is cut off mid-word because the German version runs long.
Compare against your expected-values sheet, not against your memory of the English page.
Step 3: Install fresh and watch the very first screen
Install from the store on each phone, not by sideloading, because you want the real first-run path. If you're testing a build that isn't public yet, use bulk install to put the same APK on all three phones at once, and note in your sheet that this run skipped the store step.
Then open the app and record the first screen. Splash, language prompt, any welcome carousel. Questions to answer:
- Did the app pick up the device language, or did it default to English and ask?
- Do the welcome slides fit? German text overflowing a fixed-width button is the classic one.
- Are there any hard-coded strings? They tend to hide in error toasts and permission explanations.
Step 4: Go through signup and the first paid-looking screen
Sign up as a new user on each phone. If your signup sends a verification SMS, you can use a cloud number to receive it on your own test account; that's the use case they're built for. Watch for:
- Phone number fields that reject the local format or don't default to the right country code.
- Address forms that demand a US-style state field in Japan or a ZIP code pattern in Germany.
- The first price the user sees. This is where currency formatting breaks most often, because the number usually comes from a server and the formatting from the client. A price rendered as "R$ 19.90" instead of "R$ 19,90" looks broken to a Brazilian user.
- Dates on trial-end notices or "member since" labels. A date like 03/04 means different days in Brazil and in a US-formatted string.
Step 5: Leave it, come back, check it stuck
Close the app, wait until the next day, and reopen it on each phone. This is where a long-lived device earns its keep. You're checking that the login persisted, that the language choice stuck, and that any scheduled notification arrived at a sensible local time rather than at 3 a.m. because the server assumed your office time zone.
Step 6: Log findings per country, not per screen
I keep one short table per country with three columns: what we expected, what we saw, screenshot link. Mixing all three countries into one list makes it hard for whoever fixes the bugs to tell which market they're working on. After a fix ships, rerun Steps 3 to 5 on the same phone. Because nothing was wiped, you're comparing like with like.
A few things this won't tell you
It won't replace a native speaker reading your copy for tone. It won't tell you whether your pricing is right for the market, only whether it displays correctly. And three countries is a sample, not coverage. But a two-hour pass across three regional phones catches a surprising share of the bugs that otherwise show up first in one-star reviews.
For anyone wanting to copy this, the cloud phones and the regional options are on duoplus.net. Billing is per minute, which suits this kind of check: you open the phones for a focused session, then close them until the next release.

