Mobile web page crashes from excessive memory usage are notoriously hard to detect—especially on iOS, where browsers offer no native crash APIs. So how do you know if your app crashed, or if a user simply left? Yahoo's Charles Truelson (Director of Engineering) and Ricky Blaha (Principal Engineer) share how their product and monetization engineering team built an inferred crash detection system using session markers and browser lifecycle events. They discuss how they used that crash detection system along with data-driven A/B testing to isolate the root cause—ad memory overload—and the iterative process they are using to address it. Charles and Ricky share an end-to-end strategy to combat iOS mobile web crashes caused by ad memory leaks, detailing the development of an Inferred Crash Detection System and a move toward Intelligent Ad Loading to balance crash reduction with monetization. You’ll learn about the detection methodology, key experiments (including one that showed a 70% crash reduction with ads disabled), optimizations they shipped, and their roadmap toward better ad loading. And you will gain a practical, data-driven framework to resolve similar complex, revenue-sensitive technical crises.

 17      0