How to Hire Next.js Developers: Rendering, SEO, Hosting

To hire a Next.js developer, test the decisions Next.js adds on top of React: what runs on the server, what gets cached and for how long, how pages earn their place in search, where login checks live and where the app is hosted. A developer who has made those calls on a live product shows it quickly in a practical exercise. One who has only built React screens usually does not.

This guide covers each decision, then a paid exercise, nine questions, costs by region and red flags, with examples from Next.js projects Empiric Infotech built for clients in the United Kingdom, Poland and the United States. Our terms are on the hire Next.js developers page.

The short answer

A Next.js hire comes down to seven decisions. Ask the developer to explain each from a product they shipped:

1. Server or client: what runs on the server, and where "use client" sits.
2. Rendering and caching: static, dynamic or incremental per route, and how stale data gets refreshed.
3. Search and speed: metadata, sitemaps and Core Web Vitals.
4. Data and writes: fetching in Server Components, and server actions that check permissions.
5. Authentication: cookie sessions, with checks next to the data.
6. Hosting: Vercel, AWS or your own servers.
7. Migration: off the Pages Router or a single-page React app without losing traffic.

Server Components and the App Router

The App Router, in the app directory, is the default for new Next.js projects. Its components are React Server Components unless the file starts with "use client". Server Components run only on the server, can read a database directly and send no JavaScript of their own to the browser. Client Components handle state, effects, clicks and browser APIs.

Look for discipline at that boundary. Everything imported into a Client Component joins the browser bundle, so marking a whole page "use client" for one button turns a server-rendered page into a client app. Strong developers push the directive down to the smallest interactive piece and keep pricing logic, permission checks and secrets on the server.

Static, dynamic or incremental rendering

Every route is static, dynamic or in between, which drives page speed, hosting cost and data freshness.

• Static: rendered at build time and served from a CDN. Right for marketing pages and documentation.
• Dynamic: rendered on each request. Reading cookies, headers or search parameters makes a route dynamic, which is how many apps lose their static pages unnoticed.
• Incremental: served static and regenerated in the background, after a time window or on demand through revalidatePath or revalidateTag.

Caching is where experience shows. Next.js can cache fetched data, rendered routes and visited pages in the browser, and its caching defaults have changed between major versions. A strong developer checks the defaults for your version, tags data so one webhook refreshes every page that uses it, and can explain why an editor still sees an old price.

B2B Medusa, for a client in Poland, shows the stakes. One developer built a Next.js and React storefront on a headless Medusa backend for catalogues of 100,000 to 200,000 SKUs, with delivery pricing that recalculates as buyers change quantities and dates. Product content can be cached for everyone. A price that depends on the buyer cannot.

Metadata, SEO and Core Web Vitals

Teams usually choose Next.js because pages must rank. In the App Router, each page exports a metadata object or a generateMetadata function for its title, description, canonical URL and Open Graph tags, and sitemap and robots files generate those routes from code. The Pages Router uses next/head. Ask how they handled canonicals on filtered pages and hreflang across regions.

Core Web Vitals measure loading (Largest Contentful Paint), responsiveness (Interaction to Next Paint) and visual stability (Cumulative Layout Shift), and Search Console reports them from real Chrome users. The Next.js tools are next/image for sized, responsive images, next/font to self-host fonts and next/script to delay third-party tags. A strong developer starts from field data, finds the actual LCP element on a mid-range phone and fixes that first.

Data fetching and server actions

A Server Component can be async and fetch its own data, from an API or straight from the database through Prisma or Drizzle. Experienced developers avoid waterfalls, where one await waits for another it never needed, by starting requests in parallel and streaming slow sections behind Suspense. They also know identical fetch calls in one render are deduplicated, while database calls need React's cache function.

Server actions handle writes: a function marked "use server" can be called from a form with no separate API route. Each one is also reachable as a plain POST request, so it must validate its input, check who is calling and refresh the affected cache.

A separate API still earns its place when several clients share the data. Superintelligence, an AI personal assistant for a client in the USA, was built by a two-person team: a Next.js web app and a Flutter mobile app on a Node.js backend with Strapi CMS. For that backend side, see hire Node.js developers.

Authentication and where checks belong

The usual choices are Auth.js, Clerk, Auth0 or Supabase Auth, with the session in an HTTP-only cookie. Where the checks sit matters more than the library. Middleware suits a quick redirect when there is no session, and should never be the only check. Layouts are a weak spot too, because a layout does not re-render on every navigation. The Next.js documentation recommends a small data access layer that verifies the session and the user's permission where data is read or written.

Secrets need the same care. Only variables prefixed NEXT_PUBLIC_ reach the client bundle, and the server-only package fails the build if server code is imported into a Client Component.

Vercel, AWS or your own servers

Vercel builds Next.js with no configuration, gives each pull request a preview URL and runs server code as functions. Its bill grows with traffic, so ask for an estimate from your real numbers.

On AWS, Amplify Hosting supports server-rendered Next.js, and a container on ECS gives full control. On your own servers, next start runs on any Node.js host and output: "standalone" produces a compact build for a Docker image. Self-hosting brings back work Vercel did: a CDN, image optimisation on your own CPUs, preview environments in CI and, with several instances, a shared cache so regenerated pages match.

Klavir, a bracelet store in the United Kingdom, was built by one developer on Next.js and React, with an interactive design studio, live pricing, shareable designs and Stripe checkout, deployed on Vercel.

Migrating from Pages Router or a React SPA

From the Pages Router. The pages and app directories can run side by side, so a migration moves one route at a time. getServerSideProps and getStaticProps become async Server Components, getStaticPaths becomes generateStaticParams, _app and _document become the root layout, and next/router gives way to next/navigation. Start with low-traffic routes and leave the pages that earn search traffic until last.

From a single-page React app. The official Next.js guides for Create React App and Vite first run the whole app as one client-rendered catch-all route, then move routes into real pages one at a time.

Either way, keep every URL that ranks, add permanent redirects in next.config.js for any that change, and compare Search Console data before and after each batch.

A practical exercise worth paying for

An interview shows what a developer knows. A short paid exercise shows how they decide. Keep it to about four hours: a product catalogue with three routes.

• A static listing page that refreshes when a product changes.
• A product page with its own title, canonical URL and Open Graph image.
• A review form that requires sign-in and saves through a server action.

Ask for a one-page README saying what is static, what is dynamic and what clears each cache, then review it together for 30 minutes. Look for "use client" only where needed, a session check inside the action, and images with set dimensions.

Then show them a file with four planted problems: a page marked "use client" for one button, an API key in a NEXT_PUBLIC_ variable, a server action with no permission check, and three await calls that could run in parallel. A strong developer finds all four within minutes. Empiric Infotech introduces one developer for you to meet first, and the 7-day risk-free trial then puts them on your real backlog.

Nine questions that test real experience

Each question is followed by what an experienced developer should mention.

1. A product page has a filter and an add-to-cart button. Where does "use client" go? On those two parts only. The page and its data stay in Server Components.
2. A pricing page changes twice a month. How should it render? Statically, refreshed on demand when a CMS webhook calls a route handler that runs revalidateTag.
3. An editor changes a price and the site shows the old one. Where do you look? The data cache, the rendered route, the browser's router cache, the CDN, and whether the save triggered a revalidation.
4. How do you handle metadata and sitemaps for 150,000 product pages? generateMetadata per page, canonicals that ignore filter parameters, and several sitemap files, since one holds at most 50,000 URLs.
5. Mobile LCP is poor on the home page. What do you check first? Field data in Search Console, then the LCP element: often a badly sized hero image, a web font, a needlessly dynamic page or early third-party scripts.
6. What does every server action do before writing to the database? Validates the input, for example with Zod, checks the session and permission, and refreshes the affected cache after the write.
7. Where do you check that a user may see a page? In a data access layer on the server, next to the query, with middleware only for quick redirects.
8. We want to move from Vercel to AWS. What changes? A standalone container or Amplify Hosting, a CDN, image optimisation, a shared cache, previews in CI and a cost estimate.
9. How would you move a 150-route Pages Router app to the App Router? Route by route with both routers running, low-traffic routes first, URLs unchanged, and rankings checked after each batch.

What a Next.js developer costs

BLS figures cover all software developers, with no separate pay band for Next.js or React.

• US employee: a median salary of $135,980 a year, which is $11,332 a month (BLS, May 2025), or about $15,298 a month with employer costs. Filling the role takes about six weeks and about $5,475 per hire (SHRM 2025).
• UK, EU or Australian employee: about EUR 7,875 a month in the UK and EU, and about AUD $13,455 in Australia, with employer costs included (our cost benchmark).
• Agency, per month at 160 hours, junior to senior: $3,840 to $6,560 in Asia, $4,960 to $12,160 in Eastern Europe and $5,280 to $12,000 in Latin America (Accelerance 2026).
• Empiric Infotech dedicated developer, 2026 rate: $2,000 (EUR 2,000, AUD $3,000) for 160 to 172 hours. A senior team lead reviews and tests every release, so there is no separate QA seat, and a project manager joins your meetings.

Dedicated developer cost per month compares these options region by region.

Red flags in a Next.js hire

• Every component marked "use client", with no clear reason.
• No idea which of their routes were static or dynamic.
• Private keys in NEXT_PUBLIC_ variables.
• Server actions that trust whatever the form sends.
• "It worked on Vercel" as the full answer to every hosting question.
• A migration plan that moves every route in one release.

Frequently asked questions

Do we need a Next.js developer or a React developer?

A Next.js developer, when pages must rank in search, load fast on phones or run server code next to the interface. When another team owns the backend and search does not matter, a React developer fits: see hire React developers. Next.js developers write React every day, so they cover both.

Can a Next.js app run on AWS or our own servers?

Yes. next start runs on any Node.js server and the standalone output fits in a Docker image. Plan for a CDN, image optimisation and a shared cache across instances.

Who holds the repository, domain and hosting accounts?

Your company does, from day 0. The repository, the Vercel or AWS account, the domain, the database and any CMS sit in your name, and the NDA and IP assignment are signed before work starts.

How many hours does a Next.js developer in India overlap with a UK team?

5 to 7 hours a day, 9:00 AM to 3:00 PM in the UK and 9:00 AM to 4:00 PM in Central Europe. That is enough for a daily standup and live code review.

How quickly can a Next.js developer start?

Within 48 hours of your go-ahead. You meet the developer we introduce first, and every mid to senior developer has cleared technical interviews, real-world coding tests and a portfolio review.

Meet a Next.js developer

Write down your routes, the ones that must rank, your data sources and your hosting choice, then put the exercise and questions above to the developer we introduce. They work only on your product for 160 to 172 hours a month, inside your repository, standups, code review and release process, with seven days' notice to stop. US clients get a W-8BEN-E, and UK and EU clients sign a DPA with Standard Contractual Clauses. The process and each region's terms are on the hire Next.js developers page.