Learn from developers: Lyft and Google Maps share their experiences optimizing app performance
summary
Lyft and Google Maps report concrete wins: a single blocking screen removal yields 20% app launch reduction, and bundling baseline profiles yields ~40% speed improvement. They use Android Vitals, Perfetto traces, Jetpack macro/microbenchmark libraries and custom production tracing to find hotspots, then apply caching, asyncing network calls, and profile-driven AOT priming. Apps call reportFullyDrawn to mark readiness, run daily benchmarks to catch regressions, and prioritize low‑effort fixes to prove business impact.
AI-written summary · about
chapters (24)
- 0:00Intro
- 0:36Why does app performance matter
- 1:38Business metric improvements
- 2:01How to measure startup time
- 2:48Custom libraries
- 4:03How did you measure startup time
- 5:21Google Maps benchmarking libraries
- 6:09App startup
- 7:00Finding the problem
- 7:50Using a custom pipeline
- 8:42The most important thing
- 9:17Baseline profiles
- 10:10How to use baseline profiles
- 11:46Will baseline profiles be the future of app profiling
- 12:08How to improve app startup performance
- 14:30Caching
- 15:25Challenges
- 17:52What are the challenges
- 18:52How can Google help
- 22:10Time to first
- 23:13Other thoughts
- 23:52How to become an expert
- 26:54Conclusion
- 28:34Outro
related talks
39mAndroid App Performance in a Nutshell
41mModern App Performance
38mLow effort, High impact Android performance wins
45mDeep dive into Modern Android Performance
42mPerformance Paradox: Navigating the Complexities of Mobile App Optimisation
40mAndroid vitals: Debug app performance and reap rewards
32mChasing performance issues methodically
35mThe Impact of Micro-Optimizations on Large-Scale Apps