All posts

Test on phones you don't own: emulators, labs and swaps

· 8 min read

You have one phone. Maybe it's an iPhone. Your app works great on it. But a stranger will open it on a small Android, an old iPhone or a browser you never use, and you won't be there to see it. You don't need to buy a drawer full of devices. This post covers the free tools that stand in for them, what each one can't tell you, and how to fill the gap with real people.

Android Emulator: Android phones on your computer

I read Google's Android Emulator page on 10 November 2026. It says the emulator "simulates Android devices on your computer so that you can test your application on a variety of devices and Android API levels without needing to have each physical device". It comes with Android Studio, so there is nothing extra to install.

Each virtual phone is an Android virtual device, or AVD. Google's virtual devices page says an AVD contains "a hardware profile, system image, storage area, skin, and other properties". In plain terms:

  1. Open the Device Manager in Android Studio and click Create Device.
  2. Pick a hardware profile. Pixel phones are pre-loaded, and you can define your own screen size and resolution.
  3. Pick a system image for a particular API level. That is the Android version.
  4. Finish, then launch it and run your app.

That gives you a small screen and a large one, or an old Android version and a new one, with a few clicks. Google recommends creating "an AVD for each system image that your app could potentially support". One catch: your app won't run on an image with an API level lower than the minSdk you set in your manifest.

The emulator can also do some things a real phone does. Google lists simulated phone calls and texts, a set location, different network speeds and rotation. You can slow the network on purpose and see what your app does.

What does Google say it can't do? The page calls the emulator "high fidelity" and says it provides "almost all the capabilities of a real Android device". It doesn't list what is missing. It does warn about your computer. Google suggests at least 16 GB of RAM, and says that on weaker machines the emulator "might still run but not smoothly", in which case you should "consider testing on a physical device instead". If you have an older laptop, that matters.

Xcode Simulator: iPhones on your Mac

If you build native iOS apps, Simulator is the Apple equivalent. It needs a Mac. Apple's clearest written statement of its limits is the older Testing and Debugging in Simulator guide, which I read on 10 November 2026. Apple marks it as a retired document, last updated in 2018, so treat the details below as dated and check Simulator Help in Xcode for today's version.

The principle in that guide still holds. Apple calls Simulator "a great tool for rapid prototyping and development before testing your app on a device", then adds: "Simulator is a useful tool, but it should not be the only way you test an app."

On speed, the guide is direct. The simulator is an app on a Mac, so it uses the computer's resources, and Apple says:

"As a result, the simulator is not an accurate test of an app's performance, memory usage, and networking speed." Your interface "may appear to run both faster and smoother than on a device."

The guide also listed hardware that Simulator did not support as of iOS 8.2: motion sensors, camera and microphone, the proximity sensor, the barometer and the ambient light sensor. That list is from 2018, so it may have changed. The lesson is the same: anything that depends on real hardware needs a real device.

Browser device mode: for web apps

If your app runs in a browser, the quickest check is free and already on your laptop. Chrome DevTools has a device mode. I read the device mode docs on 10 November 2026. You can set the screen size from a preset or by dragging, rotate to landscape, change the pixel ratio, and switch between touch and click events. It can also slow things down. "Mid-tier mobile" approximates fast 3G with a CPU 4x slower than your own, and "Low-end mobile" approximates slow 3G with a CPU 6x slower.

Google is honest about the limits. The page calls it a "first-order approximation" of how a page looks and feels on a phone. It also says:

  • "With device mode you don't actually run your code on a mobile device."
  • "There are some aspects of mobile devices that DevTools will never be able to simulate."
  • For accuracy, it recommends running your page on a real phone and using remote debugging.

So use it to catch layout breaks at 320 pixels wide and slow-load problems. Then open the real thing on a real phone.

Cloud device labs: Firebase Test Lab

Google's Firebase Test Lab is "cloud-based app testing infrastructure" that lets you test on a range of devices and configurations. I read it on 10 November 2026. It offers two kinds of device:

  • Physical. Tests run on "real, production devices running in a Google data center".
  • Virtual. Android test devices "that run on faster Arm hosts".

The Robo test is the one to know if you have no test code. Google says it "does not require a pre-written test", so you upload your Android app and it explores. The other test types listed are instrumentation tests, game loop tests and, for iOS, XCTest.

On cost, the usage, quotas and pricing page says the no-cost Spark plan gives you "10 test runs per day on virtual devices" and "5 test runs per day on physical devices", up to 15 test runs per day in total. For a solo maker with a small app, that is plenty to start. Check that page before you rely on it, since quotas can change.

Commercial device clouds also exist. They rent you access to many real phones, so compare their current prices if the free quota isn't enough.

If you plan to ship builds to testers through Firebase too, I wrote about Firebase App Distribution for test builds.

What emulators and labs still miss

All of these tools answer one question: does the app run? None of them answer the other question: do people get it? Here is what is left over.

  • How people actually use it. A Robo test taps everything. A person taps the one button they think is the right one, and sometimes it's the wrong one.
  • A first-time user's confusion. You can't simulate someone who has never seen your onboarding. Only a new person can be new. See testing the first 60 seconds of your app.
  • Real networks. Throttling is a model. A phone on a train with one bar is the real thing. (Throttling still catches a lot; see testing on a slow or flaky network.)
  • Notifications and interruptions. Calls, battery saver modes, a full phone, and the app being sent to the background mid-task.
  • Hands and thumbs. A mouse pointer is not a thumb. Buttons that are fine on a desktop can be hard to hit on a phone.

Real people on their own devices

This is the cheapest "device lab" there is: people who already own different phones. Friends and family, a community you belong to, or other makers who will trade a test with you. I covered where to look in finding beta testers for your app.

Build a small device matrix

You can't cover every phone, and you don't need to. Pick 3 to 5 combinations that matter for your app:

  1. The newest iPhone you can get someone to use.
  2. An older iPhone, ideally with a smaller screen.
  3. A mid-range Android phone, not a flagship.
  4. A small screen, on either platform.
  5. A tablet, only if people will really use your app on one.

Then ask every tester to say what they used. A line like "iPhone, Safari" or "Android phone, Chrome" is enough, and the model and OS version are better. Without it, a bug report is hard to repeat. With it, you can spot patterns, such as every broken layout coming from the smallest screens.

Where TestSwap fits

TestSwap is a free site where app makers review each other's apps: you test someone's app, they test yours. You can post web, iOS and Android apps, and you link to wherever testers can open or install it, such as your site, a TestFlight link or a Google Play testing link. The reviews come from real people using real apps.

TestSwap does not let you choose which device a reviewer uses, so I won't pretend it does. What you can do is ask for it. A review has an optional private bug notes box that only the maker sees, and bugs go there with the device and the steps to reproduce. Put "tell me your phone and browser" in your test task and you get a matrix for free. Here's how to write a test task. If you're an iPhone-only maker with an iOS app, the TestFlight external testers post shows how to get the build onto other people's phones.

Checklist

  1. Decide which platforms and screen sizes your users probably have.
  2. For web apps, run Chrome device mode at 320 pixels wide and with slow mobile throttling.
  3. For Android, create two or three AVDs: a small screen, a large one, and an older API level at or above your minSdk.
  4. For iOS, run Simulator on a Mac, but don't trust it for speed, memory or sensors.
  5. Try a Firebase Test Lab Robo run on a virtual device, within the free daily runs.
  6. Write your 3 to 5 device matrix.
  7. Ask every tester for their phone, OS version and browser or app version.
  8. Watch for first-use confusion, not only crashes.
  9. Use real hardware for anything involving the camera, sensors, notifications or a poor network.

Try it

If you want real people on real phones, post your app on TestSwap with a test task that says how to open it and what to do, and ask reviewers to note their device in the private bug box. Reviewing someone else's app first is the quickest way to see how it works. Tell me what's missing or broken (@dizid on X).