Next.js for the startups still on Vercel hobby.
From hobby plan to production, App Router, ISR, Server Components, and Edge. We ship Next.js the way Vercel ships their own product.
When Vercel hobby stops being enough.
Next.js is a great hammer. The trouble is most teams ship to it like it is still a static site generator. App Router, Server Components, and ISR demand a different mental model for caching, data fetching, and deployment topology. By the time a startup has 50,000 monthly users on the Vercel hobby plan, the next six months are going to be a decision about where state lives, what runs on the Edge, what runs in a Node lambda, and what runs on a separate backend. We have made these decisions for plenty of production Next.js stacks.
What changes when a Metafic pod is in your repo.
App Router from day one
If you are starting fresh or on a small pages-directory codebase, App Router is the right default. We move you there in week one and never look back.
Cache layers stated explicitly
Next.js has six different caches. We document which one holds each piece of state, and we put cache invalidation in the code-review checklist.
Edge runtime where it pays
Auth, redirects, A/B variants on the Edge. Heavy data work in Node lambdas or your own backend. We do not push everything to Edge as a fashion.
Observability from staging onward
Vercel logs, Sentry, OpenTelemetry traces hitting Honeycomb or your warehouse. So you can answer "is this slow for everyone or just this one user".
ISR and on-demand revalidation done right
Most teams either over-rebuild or under-rebuild. We tune revalidation per route based on how often the underlying data changes.
Who is on the pod for this work.
Pods scale up from here for Enterprise engagements.
Has shipped multiple Next.js apps to production, including App Router migrations.
5+ years JS, deep Next.js, comfortable with Server Actions and form mutations.
Builds the Playwright suite that runs on every PR against a preview deployment.
Tuned to read your repo and propose Server-Component refactors, plus generate route handlers and Zod schemas.
The bugs that bite this stack.
Form actions vs API routes vs route handlers
Three ways to handle a form submission, each with different cache implications. Picking the wrong one means re-fetching the world on every keystroke.
Cookie-based auth and Server Components
Auth state in Server Components is a class of bugs all on its own. We use a server-action-first auth pattern and avoid the worst footguns.
Bundle size sneaking up
Server Components do not ship to the browser, but the package you accidentally imported into a client component does. We audit in week three.
Vercel costs exploding at scale
Vercel is great until you hit 1M+ MAU. Function-invocation cost can dominate. We help model whether your traffic profile makes sense on Vercel or on your own infra.
Honest about scope.
We will not push you off Vercel just because. Most startups should stay on Vercel until at least Series B. We will model the cost so you can defend the decision either way.
Common questions.
Vercel, Netlify, or our own infra?
Vercel for almost everyone in the first two years. Cloudflare Pages or your own infra if you have very high traffic and engineering capacity to operate it. We model your specific case in week one.
Are Server Components production-ready?
Yes. The ecosystem has caught up. Use them by default; reach for "use client" deliberately.
Will you train our team on Next.js patterns?
We pair-program with your engineers in the first few weeks specifically so the patterns we introduce do not leave the building when the pod pauses.
What if we are on Pages Router today?
We stay there if it is working and migrate route-by-route to App Router only where it gives real lift. Big-bang App Router migrations are how teams get stuck.
Ready to scope it?
A 25-minute call. We will tell you what we would do, what we would not, and whether a pod is the right shape.
Or stay in the loop. One engineering teardown a week.