Every Android developer knows the frustration: You shipped a new version of your app and unexpectedly, your: - Application launch timings trend up - ANR rate spikes - Play Store rating is at risk of slipping You jump into your observability tool to investigate. It tells you what is broken, but not why. Born out of a direct collaboration with Google's Android team, it is now possible to look under the hood of your applications at scale through Datadog’s Mobile Real User Monitoring (RUM) integration with the official ProfilingManager API. In this talk, we will share what we built, what we learned, and walk through an example of end-to-end troubleshooting workflow: - Creating monitors or SLOs on top of time to initial display (TTID), time to full display (TTFD), ANR rate, or error volume metrics. - Pivoting from a monitor alert to problematic RUM sessions with profiling data. - Drilling into flame graphs, thread timelines, and call graphs to identify the specific methods responsible for application slowness. Additionally, we will share upcoming enhancements to Android profiling data collection, including: - Dynamically changing how frequently your application is profiled. - Manually starting and stopping profiling during key moment. - Reviewing your profiling data in aggregate to spot system-level bottlenecks.