On this page
01. Your React application suddenly becomes slow after adding new features. How would you identify and fix the issue?
- Your React app suddenly becomes slow after adding new features. How do you find and fix the issue? (Debugging Performance)
Short answer
Because it got slow "after new features", I treat it as a regression: find which change caused it (compare releases, git bisect, bundle diff), then measure with the React Profiler and Chrome Performance panel to see whether the cost is extra re-renders, an expensive computation, a bigger bundle or more network requests. I fix that specific cause and add a performance budget so it cannot creep back unnoticed.
Why it happens
New features commonly bring:
- State lifted too high or a new value in a widely used Context, so far more components re-render.
- Unstable props: inline objects/functions that defeat
memo. - Heavy dependencies (charting, date, editor libraries) added to the main bundle.
- Expensive work in render: filtering/sorting large arrays on every keystroke.
- Request waterfalls: new components fetching in sequence inside
useEffect. - Effects that loop:
setStatein an effect with a dependency that changes every render.
How to fix it
1. Find the culprit change
git bisect start
git bisect bad HEAD
git bisect good v2.3.0 # last fast release
# test each step, mark good/bad until the commit is found
Also compare bundle reports (rollup-plugin-visualizer, @next/bundle-analyzer) and Lighthouse scores between the two versions.
2. Measure where time goes
React Profiler -> which components render, why, how long
Performance tab -> long tasks, scripting vs layout vs paint
Network tab -> new requests, sizes, waterfalls
3. Fix by cause
// Before: new "theme" state in a top-level context re-renders the whole app
<AppContext.Provider value={{ user, theme, setTheme }}>
// After: split contexts and memoize the value
const themeValue = useMemo(() => ({ theme, setTheme }), [theme]);
<UserContext.Provider value={user}>
<ThemeContext.Provider value={themeValue}>
{children}
</ThemeContext.Provider>
</UserContext.Provider>
| Cause | Fix |
|---|---|
| Too many re-renders | Move state down, split context, memo + stable props |
| Expensive render | useMemo, Web Worker, useDeferredValue |
| Big bundle | React.lazy / dynamic import(), lighter library |
| Waterfalls | Parallel fetch, route loaders, React Query prefetch |
| Long lists | Virtualization |
💡 Interview tip: Say "measure first". Wrapping everything in
memowithout data is a red flag to interviewers.
4. Guard against regressions
Add bundle-size budgets (size-limit) and Lighthouse CI to pull requests, and track Web Vitals per release.
How to verify
- Profiler recording of the same interaction shows fewer/shorter renders.
- Lighthouse and bundle size are back to the old baseline.
- Field INP/LCP per release returns to normal.
Follow-up questions
- What is
git bisect? A binary search over commits to find the first bad one. - Will the React Compiler fix this? It removes most manual memoization needs, but not big bundles, waterfalls or heavy algorithms.
- How do you profile production? Use a profiling build or the
<Profiler>API and RUM data.
Quick recap
- Treat it as a regression: find the change.
- Measure renders, long tasks, bundle and network.
- Fix the actual cause with the matching technique.
- Add budgets and CI checks to stop it recurring.