You need to migrate a legacy frontend codebase to a modern framework. What is your plan?

Avoid a big-bang rewrite. Assess and set goals, then migrate incrementally with the strangler-fig pattern — run old and new side by side, migrate route-by-route or feature-by-feature, keep shipping product, add tests as a safety net, and measure progress.

7 min read·~15 min to think through

The headline answer: incremental migration via the strangler-fig pattern — not a big-bang rewrite. Rewrites are high-risk, freeze product delivery, and usually fail or overrun badly.

1. Assess before touching anything

Why migrate? Performance, hiring, maintainability, security, ecosystem? The goal shapes the plan and justifies the cost.
Map the codebase — routes, features, shared components, dependencies, dead code, the riskiest/most-coupled areas.
Pick the target deliberately (React/Next, Vue, Svelte…) — team skills, ecosystem, requirements.
Get buy-in — leadership needs to know it's incremental and product keeps shipping.

2. Strangler-fig: new grows around the old

Run the legacy app and the new app side by side, and migrate piece by piece until the legacy is "strangled" out.

Routing seam — a proxy/router sends some routes to the legacy app, others to the new one. Migrate route by route (or feature by feature).
Or embed — mount new components inside the old app (or vice versa) at boundaries; web components or micro-frontends can bridge frameworks.
Shared shell — common header/nav/auth so the seams aren't visible to users.
Start with a low-risk, self-contained area to validate the approach and tooling, then move to higher-value areas.

3. Keep shipping product

Migration runs alongside feature work, not instead of it — pure-migration freezes don't survive contact with the business.
New features get built in the new stack; touched legacy areas get migrated opportunistically.
Avoid a long-lived branch — integrate continuously.

4. Safety nets

Add tests around legacy behavior before migrating it — characterization/E2E tests on critical flows so you can prove parity.
TypeScript, linting, CI gates on the new code.
Feature-flag the new implementation; roll out gradually; keep a rollback path.
Monitor errors and performance per migrated slice.

5. Watch the tradeoffs

Two stacks at once — bigger bundle, duplicated some logic, more complexity during the transition. Accept it as temporary; minimize the window.
Shared state/auth across the boundary needs a deliberate bridge.
Don't let "incremental" become "forever" — track progress (% migrated) and keep momentum.

6. Finish

Migrate the last pieces, delete the legacy code and its dependencies, remove the routing seam and any bridges.

The framing

"I'd resist a rewrite. Assess and set a clear goal, then strangler-fig: run new alongside old, migrate route-by-route behind feature flags with tests proving parity, keep shipping product throughout, and measure progress until the legacy is fully replaced and deletable."

Follow-up questions

•Why is a big-bang rewrite usually a bad idea?
•How does the strangler-fig pattern work in practice for a frontend?
•How do you keep shipping features during a migration?
•How do you handle shared auth/state across the old/new boundary?

Common mistakes

•Proposing a big-bang rewrite with a feature freeze.
•A long-lived migration branch that diverges from main.
•Migrating without tests, so you can't prove behavior parity.
•Starting with the hardest, most-coupled area.
•Letting the migration stall half-done, leaving two stacks forever.

Performance considerations

•Running two frameworks temporarily inflates bundle size and complexity — keep the transition window short and code-split the seam. Per-slice monitoring catches regressions as each piece moves.

Edge cases

•Shared global state/auth spanning old and new code.
•SEO/routing must stay stable through the transition.
•A legacy area nobody understands anymore.
•Third-party integrations tightly coupled to the old framework.

Real-world examples

•Route-by-route migration behind a proxy: legacy serves /old/*, new app serves migrated routes.
•Embedding new React components into a legacy app via a mount point, expanding outward.

Senior engineer discussion

Seniors reject the rewrite reflexively and lead with strangler-fig: assess and justify, run old/new side by side, migrate route/feature-by-feature behind flags with characterization tests proving parity, and keep product shipping throughout. They're honest about the temporary two-stack cost and stress measuring progress so 'incremental' doesn't become 'permanent.'