Crash reports before launch: Crashlytics and Sentry basics
· 7 min read
A tester opens your beta, taps around, and the app closes. They shrug, reopen it and move on. Maybe they tell you "it crashed once". That is not enough to fix anything. A crash reporter is a small piece of code that records what happened at the moment of the crash and sends it to you. You can add one in under an hour, and it is much easier to do before launch than after. This post covers the two free-to-start options most indie makers pick, Firebase Crashlytics and Sentry, plus the tools your app stores already give you.
Why tester notes are not enough
Testers are not engineers, and they should not have to be. They rarely know the exact screen, the steps, or what the app was doing. Many crashes happen once and never again, so there is nothing to reproduce. And some testers never write anything at all. A crash reporter does not replace their notes. It adds the part they can't give you: the technical record. Use both.
What a crash reporter captures
I read the docs of both tools on 10 November 2026. The details differ, but the idea is the same.
- A stack trace. The list of code calls that led to the crash. Sentry's source maps page says its goal is to "see the exact line of code that caused an error".
- Device and OS. Google's Crashlytics customize page says an issue summary has breakdowns "by Android version and hardware device". That tells you if a crash only happens on one phone.
- Breadcrumbs and logs. The steps before the crash. Sentry uses breadcrumbs to "create a trail of events that happened prior to an issue" (see its breadcrumbs page). Crashlytics shows breadcrumb logs and your own custom logs in the Logs tab of an issue. Its breadcrumb logs need Google Analytics enabled for your Firebase project, and custom logs are limited to 64kB.
- Non-fatal errors. Crashlytics reports "crashes, non-fatal errors, and ANRs". An ANR means "Application Not Responding", an app that froze. Sentry says it "automatically reports errors, uncaught exceptions, and crashes".
Firebase Crashlytics
The Crashlytics get started page lists guides for Apple platforms, Android, Web, Flutter, Unity and Android NDK. The pricing table on the Firebase pricing page lists Crashlytics for mobile apps as "No-cost" on both the Spark and Blaze plans. For web apps the Spark column reads "Not applicable", so for a web app use Sentry instead.
The outline is the same on every platform. The guides give you "step-by-step instructions to set up Crashlytics in your app and force a test crash to check your setup."
- Add the SDK. On Android, the Android guide has you add the Crashlytics library to your module-level Gradle file, then add the Crashlytics Gradle plugin to the root-level and module-level Gradle files.
- Force a test crash. The guide is titled "Force a test crash to finish setup". You add a button that throws an exception and press it.
- Check the console. Open the Crashlytics dashboard in the Firebase console and look for the test crash report. If it has not shown up after five minutes, the Android guide suggests turning on debug logging.
iPhone: upload your dSYM files
The Apple platforms guide has three main steps: add the SDK, "Set up Xcode to automatically upload dSYM files", and force a test crash. The dSYM step is a Run Script build phase. For the test crash, run the app with the Xcode debugger disconnected, then relaunch it so the report is sent. The guide also says to enable Google Analytics to get breadcrumb logs, which are available on every Apple platform Crashlytics supports except watchOS.
The dSYM page says to check that Debug Information Format is set to "DWARF with dSYM File" for all build types. If an upload fails, the console shows a "Missing dSYM" alert. You can also upload by hand with the drag-and-drop zip option in the console or the upload-symbols script.
Android: the mapping file
If your release build is obfuscated, the stack trace is a pile of short meaningless names. The deobfuscated reports page says the plugin uploads the mapping file so the servers can show stack traces as "unobfuscated and human-readable code". When an obfuscated build makes a mapping file, the plugin uploads it with no extra setup. The same page shows a setting, firebaseCrashlytics.mappingFileUploadEnabled, that turns uploads off for a variant. If you set it to false, those reports stay obfuscated. So leave it on for your beta build.
Sentry
Sentry covers mobile and web with the same dashboard. It is the natural pick if your app is a web app. Its pricing page, which I read today, says "Start free on the Developer plan forever". The comparison table lists "5k errors" for error monitoring and "One user" on that plan. That is plenty for a small beta. Check the page again before you rely on it, since plans change.
Web app (JavaScript)
The JavaScript guide recommends installing the SDK with a package manager, for example npm install @sentry/browser --save. Then you initialize it with Sentry.init and your dsn, which is the address that tells the SDK which Sentry project to send to. The guide says to start Sentry "as early as possible". To verify, add a test button that throws an error and click it. The page notes that errors triggered from the browser developer tools won't be captured, so use a real button in your code.
Production web code is usually minified, which makes stack traces unreadable. The source maps page says "To enable readable stack traces in your Sentry errors, you need to upload your source maps to Sentry". The easiest route is the wizard, npx @sentry/wizard@latest -i sourcemaps. For webpack, Rollup, Vite and esbuild you can use the matching Sentry plugin. One catch from the same page: "Source maps are only generated and uploaded during production builds", so test with a production build.
Android
The Android guide recommends installing the SDK through the Sentry Wizard. The DSN goes in AndroidManifest.xml as a meta-data entry. To verify, add the guide's snippet that throws and catches a deliberate exception, then look for it in Sentry.
The tools your stores already give you
Both stores collect crashes without any SDK. On Android, Play Console has crash reporting under Android vitals. I covered it in the pre-launch report and Android vitals, so I won't repeat it here. On iPhone, look in Xcode's Organizer and in the TestFlight area of App Store Connect. These are free and need no code. The catch is that they only show what the platform collected, and not your own breadcrumbs or custom logs. Use them as a safety net, and add Crashlytics or Sentry for the detail.
Privacy: crash data is data
A crash reporter sends information from your testers' phones to a third party. Treat it as collected data when you fill in Apple's App Privacy details and Google Play's Data safety form. Two posts walk through those: the App Store product page and the Google Play data safety form.
Keep personal data out of your logs. Sentry's sensitive data page says sensitive data may appear in several places, and that beforeSend lets you "scrub any data before it is sent". It also says to "use beforeBreadcrumb to filter it out from breadcrumbs before it is attached", and that automated behavior "is controlled via sendDefaultPii". Crashlytics says it "includes a way to anonymously identify users in your crash reports". Use that, and never put an email address or name in a log line.
Your first week of beta
- Day 1: force a test crash on a real build and confirm it appears in the dashboard.
- Every day: open the dashboard for five minutes. Look for new crash groups and any that repeat.
- Match to notes: when a tester writes "it closed on me", find the crash from that phone and time. Then sort everything as in turning beta feedback into a fix list.
- Fix the top crashes first: the ones that hit the most testers, or that happen on the first screens.
- Ship a new build and watch the same group. If it stops, you are done with it.
The checklist
- Pick one tool: Crashlytics for a mobile app, Sentry for a web app or if you want one dashboard for both.
- Add the SDK to the build your testers will install.
- Force a test crash or a test error, and see it in the dashboard.
- iPhone: confirm dSYM files upload. Android: keep mapping file upload on. Web: upload source maps from a production build.
- Strip personal data from logs and breadcrumbs.
- Update the App Privacy details and the Data safety form to match.
- Check the dashboard daily during the beta, and fix the top crashes first.
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 link to wherever testers can open or install it, such as your site, a TestFlight link or a Google Play testing link. Real people on their own phones will hit crashes that you never see on your own phone.
TestSwap does not collect crash reports for you, so set up one of the tools above yourself. A review has an optional private bug notes box that only the maker sees. Ask testers to put what they were doing and which phone they used there. Then match those notes to your dashboard.
Try it
Add a crash reporter to your beta build today and force one test crash. Then post your app on TestSwap with a test task and ask testers to say what phone they used. 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).