Shipping a new feature is easy. Knowing if it actually improved anything? That's the hard part.
In production React applications, small UI changes can have a big impact on user behaviour. A new form, validation change, onboarding step, checkout flow, or performance improvement might reduce friction - or quietly make the experience worse.
Many teams still ship these changes based on intuition rather than evidence. Without measurement, it is impossible to know whether users succeeded more often, where they dropped off, or whether the change created unintended consequences.
This talk makes the case that experimentation is part of frontend engineering. Engineers are not only responsible for building the interface; we also help define what success looks like, add the instrumentation that makes behaviour visible, and release changes in a way that lets teams learn safely.
In this talk, we’ll walk through the experimentation lifecycle from a React engineering perspective: forming a hypothesis, defining success and guardrail metrics, instrumenting meaningful user behaviour, gathering data through feature flags or staged rollouts, and learning from the results.
Attendees will leave with a practical framework for embedding experimentation into React feature delivery, so teams can build confidence in the true impact of what they ship.
This talk has been presented at React Day Berlin 2026, check out the latest edition of this React Conference.





















