Crash-reporting taken to next-level
summary
Built a BigQuery + Data Studio pipeline that ingests Firebase Crashlytics (real-time export of ~40–50 tables), cleans with Airflow, joins Firebase Analytics and backend datasets, and produces a single crash-free user calculation across 20+ apps and Android/iOS in 50+ countries handling 14 billion events. Adds targeted mitigations like forbidden versions and backend-controlled updates, surfaces region/screens/stack traces, links crashes to business metrics (crashers show ~40% higher cancellations) and saves engineers >40% of troubleshooting time.
AI-written summary · about
chapters (14)
- 0:00Introduction and speakers
- 1:11About Delivery Hero
- 3:38Why we started crash reporting
- 7:51Limitations of Firebase Crashlytics
- 9:58Custom keys and data export
- 12:05Data architecture setup
- 16:39Building dashboards with Data Studio
- 17:21Dashboard features and filters
- 18:45Software development and release process
- 21:35Feature flag rollout monitoring
- 23:02Last active screen analysis
- 24:05Business impact of crashes
- 25:31Country-level crash analytics
- 27:17Summary and next steps
related talks
32mChasing performance issues methodically
28mFirebase analytics for BigQuery
46mFirebase & BigQuery - Do Mobile App Analytics Easily & at Scale - Queries Included
44mA Crash Course in Google Analytics for Firebase and BigQuery
44mEfficient Android app monitoring
32mTurn Your App Data into Answers with Firebase and BigQuery
22mSupercharge your app with Firebase and BigQuery
11mCrashlytics your way: Your most important crashes with Custom Keys