概要
公開日: 2026-06-16
翻訳生成に失敗したため、原文をそのまま保存しています。
原文
EAS Observe is a production performance monitoring tool for React Native. It tells you how your app actually performs in the hands of real users, across real devices, networks and conditions, so you can find what’s slow, see what a release changed, and fix it before it costs your users. Observe is free while it's in open beta, including the first 10,000 monthly active users. When the beta ends, we'll introduce paid plans that scale with your monthly active users, and we'll keep a free tier - up to 1,000 MAUs, with all five built-in metrics - for smaller apps. We'll give everyone at least 60 days' notice before any pricing takes effect. main-dashboard Why we built observability for React Native apps The React Native ecosystem is full of great tools for understanding your app while you're building it: React Native DevTools, Expo Atlas, Radon IDE. It’s also full of great tools for when your app breaks such as Sentry and Bugsnag. But there's been a gap in the middle. What about the app that doesn't crash and isn't broken, it's just… not fast enough? How long does it really take to become interactive after a cold launch? Did that "harmless" dependency bump add 400ms to startup? Which screen is the slowest to render, and for whom? Answering those questions used to mean guessing, or building your own telemetry pipeline. Observe is designed to give you answers out of the box. What you get with Observe Real startup metrics, from real sessions Observe automatically tracks the metrics that define how fast your app feels when it opens: - Cold launch - launching from scratch, the slowest and most important path. - Warm launch - returning to an app the OS kept in memory. - Time to first render - when something other than the splash screen finally appears. - Time to interactive (TTI) - when users can actually tap, scroll, and use the app. This is the number your users feel. - Bundle load - how long it takes to load and evaluate your JavaScript. These come from your real production sessions, so you see the full distribution, not just the happy path. Observe reports median, average, min, max, and P90/P99, because the experience of your slowest 10% of sessions is often where churn hides. Each data point carries the context you need to explain it: device model, OS version, app version and build, region, network type, even thermal state and low-power mode. So "startup is slow" becomes "startup is slow on older Android devices on cellular," which is a problem you can actually fix. single-metric See exactly what a release changed Most performance regressions ship in a release that "looked fine." Observe overlays your builds and updates directly on the timeline, so a regression isn't a mystery, it's a vertical line you can point at. You can compare version to version, build to build, and even embedded JavaScript against an OTA update, then drill from an app version down to a specific build number, update, or EAS build to find exactly where a number moved. version-comparison Per-screen navigation metrics Observe supports per-screen navigation metrics for both Expo Router and React Navigation. This integration is opt-in and relies on enabling the corresponding integration in the config object. Observe will then be able to track screen render metrics automatically. We separate these out into "initial" and "subsequent" renders: a subsequent render means the screen was already rendered in the stack when it was navigated to (for example, opening a modal and navigating back to the preceding screen by closing the modal). Additionally, if you add a markInteractive call to your remaining screens, you'll also capture the corresponding time to interactive metric for each screen. navigation-metrics Update download time If you use our OTA Updates, Observe automatically tracks how long your over-the-air updates take to download in the field, with no extra setup. You get download counts and median/P90 download times per published update, so you can see whether a la