Contents
App monitoring and crash reporting ensure you learn about crashes, errors and performance problems before users describe them in a one-star review. Most users don't report bugs – they uninstall. This guide covers the tools available, the metrics that matter and how to set up monitoring in a GDPR-compliant way.
Why Monitoring Is Essential for Apps
Unlike a website, an app runs on thousands of different devices with different OS versions, storage levels and network quality. Bugs that never appear in testing surface only in the field. And you can't simply roll back a faulty version: an update must pass app review, and users install it with a delay. That makes early detection and targeted fixes all the more important.
Good monitoring answers three questions:
- What is broken? Crashes, frozen screens, failed API calls.
- Who is affected? Which app version, device, OS and how many users.
- Why does it happen? Stack trace, the user's last actions, network and memory state.
Key Metrics
| Metric | Meaning | What to watch |
|---|---|---|
| Crash-free users / sessions | Share of users or sessions without a crash | Compare the trend per version, not just the overall value |
| ANR rate (Android) | "Application Not Responding": the app freezes for several seconds | Google Play penalises high rates, including in visibility |
| Hangs (iOS) | Main thread blocked, UI frozen | Visible in Xcode Organizer and crash tools |
| Launch time | Time to the first usable screen | Measure cold and warm starts separately |
| API error rate and latency | Share of failed or slow backend calls | Correlate with server monitoring |
| Non-fatal errors | Caught errors that impair features | E.g. failed payments or uploads |
Concrete targets depend on app type and audience. More important than an absolute number is that each new version is at least as stable as the previous one. For load times and resources, see optimising app performance.
Tools for Crash Reporting and Monitoring
- Firebase Crashlytics: widely used, easy to integrate, good crash grouping. Part of Google's Firebase platform.
- Sentry: crash reporting, performance tracing and release health for apps and backend in one tool. Can also be self-hosted, which is attractive for keeping data in Germany or the EU.
- Platform tools: Xcode Organizer and App Store Connect for iOS, Android Vitals in the Google Play Console. Free and SDK-free, but delayed and less detailed.
- Observability platforms: Datadog, New Relic, Grafana-based stacks or OpenTelemetry setups when app and backend should be observed together.
For most projects, one crash reporting tool plus the platform consoles is enough. Larger products with their own backend benefit from end-to-end tracing from a tap in the app to the database query.
Setup Step by Step
- Integrate the SDK and initialise it as early as possible during app launch.
- Ensure symbolication: upload dSYM files (iOS) and R8/ProGuard mapping files (Android) automatically with every build. Without them, stack traces are unreadable.
- Tag releases and environments: attach version, build number and environment (staging, production) to every event.
- Define breadcrumbs: log navigation steps and key actions so the path to an error is traceable – without personal content.
- Report non-fatal errors: explicitly monitor critical flows such as login, payment or sync.
- Set up alerts: notify Slack, Teams or email for new error types, sudden spikes or a drop in the crash-free rate.
- Assign ownership: who checks the dashboard daily, and who decides on a hotfix?
Steps 2 and 3 fit neatly into the build pipeline – see our article on automating app releases with CI/CD.
Staged Rollouts: Contain Bugs Before Everyone Is Hit
Monitoring reaches its full potential when combined with gradual releases. Google Play supports percentage rollouts, and Apple offers phased releases over several days. If the crash rate rises in the new version, pause the rollout and fix the issue before all users are affected. Feature flags add another safety net by letting you switch off new features server-side without an app update.
Privacy in Monitoring
Crash reports contain device data, IP addresses and sometimes unintentionally personal content. Keep in mind:
- Never write names, email addresses, health or payment data into logs and breadcrumbs; enable the tools' scrubbing features.
- Sign a data processing agreement with the provider and check the server location; choose a self-hosted option if needed.
- Check whether certain analyses require consent and update your privacy policy accordingly.
- Limit retention periods for monitoring data.
Keep monitoring separate from product analytics: crash reporting serves debugging, while usage analytics serve other purposes. See app analytics: the KPIs that matter. This section is not legal advice.
Effort and Ongoing Costs
Integrating a crash reporting tool including symbolication and alerting usually takes a few days and should be part of every app project's standard scope. Many tools offer free tiers; as users and event volumes grow, usage-based fees apply. Also plan time for regular review – it is an essential part of ongoing app maintenance.
Agentic coding workshop for your team? From €799 net – request a quote by e-mailConclusion
Without monitoring, you are flying blind after launch. With a crash reporting tool, proper symbolication, clear metrics, alerts and staged rollouts, you detect problems early and fix them precisely. The key is to plan monitoring from the start and configure it in a privacy-compliant way. It is standard in the apps we build – request a fixed-price offer through our offer wizard.
Frequently Asked Questions
Which crash reporting tool is best for apps?
Firebase Crashlytics and Sentry are the most common options; Sentry is especially suitable if you also want to monitor the backend or self-host. Complement either with the data in App Store Connect and the Google Play Console.
Can crash reporting be GDPR-compliant?
Yes, if you keep personal content out of reports, sign a data processing agreement, check the server location and update your privacy policy.
What is a good crash-free rate?
There is no universal target; what matters is a high rate that stays at least stable with every version. A break in the trend after a release is the most important warning sign.
Why are my stack traces unreadable?
Usually the symbol files are missing: dSYM files for iOS or mapping files for Android must be uploaded for every build, ideally automatically in your CI/CD pipeline.