Migrating a Shipping App to React Native's New Architecture
When I took over the Topship mobile app, the crash rate was the single biggest threat to the product. People trust a logistics app with real cargo and real money, so every crash is a booking someone loses faith in. The old bridge-based architecture was the root of a whole family of these failures, so the New Architecture was not a nice-to-have, it was the fix.
The mistake most teams make is treating a New Architecture migration as one big cutover. I did the opposite. I turned on the new renderer behind a flag, moved the app screen by screen, and leaned on Sentry to tell me which native modules and third-party libraries were not yet interop-ready. Anything that was not compatible got wrapped, replaced, or upgraded before it reached production.
The order mattered. I started with the highest-traffic, lowest-risk screens to build confidence and catch interop issues early, then moved to the complex tracking and animation screens where the old bridge caused the worst jank. Because the work shipped incrementally, feature development never had to stop, and each release proved out a slice of the migration in front of real users.
By the end, crashes were down about 90 percent and the app held a steady frame rate through its heaviest screens. The lesson I carry forward: a migration you can ship in slices is a migration you can actually finish, because it earns trust with every release instead of asking for it all at once.