Hanson Ho

Android Architect at Embrace

Talk Title

Combating sampling bias in production: How to collect and interpret performance data to drive growth


Robertson 2




13:30 > 40 min


on Twitter

App performance data collected in production can be affected by several forms of sampling biases, including the oversampling pf heavy users, undersampling of churned users, and other forms of survivorship bias. This results in an incorrect picture of how your app is actually being experienced in the wild. In this session, we will discuss these biases, including techniques you can use today to mitigate them. You'll learn what you need to get a clear understanding of how all folks in the *very* heterogeneous Android ecosystem are experiencing your app. That way, you can tailor your app to enhance everyone's user experience and drive growth for all segments and device cohorts.

Suggested tags: Observability, App Performance, Performance Metrics, User Growth

Speaker Bio

Hanson was the former Tech Lead of Android Performance and Stability at Twitter, where he spent a lot of time on collecting and interpreting performance data in order to improve the app experience for all Twitter users on all Android devices all around the world. He is now at Embrace trying to help all app developers do the same as what he did at Twitter, but a lot more quickly and easily.