How We Improve Android App Performance (Part 1)
Article Summary
Tokopedia's Android team achieved a 40% improvement in app start time and page load speed. Here's the systematic approach that got them there.
Vishal Gupta from Tokopedia Engineering shares their four-stage performance improvement cycle, focusing on how they monitor production apps. This is part one of a series on scaling Android performance for millions of users.
Key Takeaways
- Three core metrics: App Start Time (target: under 2.5s), Frame Performance Index (target: 90%+), Page Load Time (under 2s)
- Built custom Data Studio dashboard on Firebase + BigQuery to monitor all pages in one view
- Optimized BigQuery costs by running scheduled queries once daily instead of continuous scans
- Custom traces capture Page Load Time from onCreate to API response and UI render
- Color-coded scorecards show red/green health status against performance targets per version
A structured monitoring system using Firebase Performance, BigQuery, and Data Studio helped Tokopedia improve app performance by 40% while preventing regressions.
About This Article
Tokopedia couldn't see how their production app was performing across all pages at once. Firebase's built-in tools couldn't show them the full picture of metrics or how things were doing compared to their targets.
Vishal Gupta's team pulled Firebase performance data into BigQuery and set up daily scheduled queries. They connected everything to Data Studio to build a single dashboard with color-coded scorecards that showed what they needed to know.
The dashboard let the team track performance across all their important pages in one place. They could spot problems before they hit production and keep the 40% improvement they'd gained in app start time and page load speeds.
