Every team says they want to be data-driven, but most product decisions still happen somewhere between a dashboard and a gut feeling. We ship features because a metric looked promising, or we ignore the metric entirely because we believed in the idea anyway. Neither approach is wrong all the time, but almost nobody talks honestly about how these decisions actually get made in real React Native teams.

This roundtable brings together developers and product-minded engineers to compare real experiences with A/B testing, feature flags, and rollout decisions in React Native apps. We will talk about the experiments that clearly worked, the ones that quietly failed, and the moments when data pointed one way but the team shipped the other anyway. We will also get honest about the hidden cost of experimentation itself: the extra engineering overhead, the slower release cycles, and the times testing infrastructure created more confusion than clarity.

Your app's slowest 500 milliseconds happen before you write a single line of render logic. In 2026, sub-second cold starts on mid-range devices are achievable, and the New Architecture has eliminated most obvious bottlenecks. So where do the remaining milliseconds hide?
This talk walks through the complete modern React Native cold start: native process launch, JS engine initialisation, JSI setup, module registration, bundle loading, and the handoff to first render.
You'll learn what Hermes V1 actually changed, and what's still ahead. Why native module initialisation remains a hidden cost post-migration. How bundle strategy and load order impact Time to First Render. Real production profiling data on where teams still lose time. And practical, high-leverage defaults that deliver consistent wins.
Walk away with a clear mental model of the startup path and a checklist to hit sub-1.2s iOS, sub-2s TTI Android in 2026.