Next.js / Astro migration in Copenhagen

Copenhagen teams with a heavy Next.js marketing build can move read-mostly routes to Astro islands, keep auth on Next.js, and load scripts only after Danish cookie consent.

Next.js / Astro migration

Website & application migration in Copenhagen

I migrate sites from WordPress, Joomla, Drupal, Angular, Vue and other technologies to Astro and Next.js. I plan each migration to keep SEO, content and features intact and switch over in an agreed window. I work alone and have years of experience with both legacy and modern stacks.

  • I have had a wordpress.org account since 2006.
  • I contributed to WordPress core for versions 7.0 and 7.1.
  • I maintain 30 plugins on wordpress.org.
  • I organise and speak at WordCamps.
  • I have attended every WordCamp Europe from 2017 to 2026.
  • Local WordPress meetup: WordPress Copenhagen

My WordPress.org profile

This page is for a Copenhagen team whose Next.js deployment has grown far beyond what the traffic pattern needs: investor pages, handbooks, partner directories and Danish or English press rooms that change a few times per week, yet still pull a full client React bundle on every request. The public site behaves like a brochure, but the runtime behaves like a single-page application. I split that stack: Astro for read-mostly routes with islands for the few interactive widgets, and Next.js only where session state, MitID-style login flows or checkout logic genuinely requires it. I work as one developer from Gdynia, remotely and in writing, and I document every redirect and consent decision before production traffic moves.

Denmark’s Agency for Digital Government (Digitaliseringsstyrelsen) publishes digitalisation standards for the public sector that many Greater Copenhagen procurement officers paste into vendor questionnaires. Private life-science and logistics sites inherit the same WCAG vocabulary even when they are not bound by public-sector law verbatim.

Copenhagen readers skew mobile on commuter networks, so marketing teams benchmark article routes against the official Core Web Vitals thresholds on web.dev/articles/vitals. Astro islands let a Medicon Valley factsheet ship as HTML while a trial-locator widget hydrates only after a deliberate click, which keeps interaction budgets closer to the 200 millisecond INP engineering band without dropping Next.js on authenticated study portals. Danish cookie consent texts still reference Datatilsynet examples, so staging captures a screen recording before any proxy flip.

#Where a heavy Next.js build hurts Copenhagen teams first

In Medicon Valley life-science marketing, Nordic SaaS documentation hubs and harbour-adjacent industrial sites, the pain rarely sits in the CMS. It sits in the default hydration model. Next.js pre-renders HTML, but the official Pages Router documentation still describes a hydration step that makes each page fully interactive in the browser. When almost nobody clicks a client-side filter on a compliance PDF index, that machinery is mostly idle yet still parsed on the main thread.

The symptoms are familiar: marketing asks for faster mobile article loads, finance questions rising serverless invocations on static guides, and compliance asks why analytics pixels appear in network traces before anyone accepts cookies. Those three threads point to architecture, not a missing WordPress plugin.

Astro’s islands documentation describes the alternative clearly: most of the page is static HTML, and only components marked with a client:* directive ship JavaScript. That matches how Copenhagen sites are actually read.

#What I move to Astro, and what stays on Next.js

I classify every route before writing templates. The split is deliberate and written down so your team can challenge it.

Astro (default for public marketing and knowledge content):

  • Corporate storytelling, ESG statements, job landing pages without logged-in state
  • Product catalogues and spec sheets that editors update from Storyblok, Sanity, Contentful or headless WordPress
  • Danish and English documentation trees where search can be server-rendered HTML plus a small island for the search box
  • Campaign pages and event microsites that must stay fast on mobile networks

Next.js (keep or extract, same registrable domain):

  • Customer portals with session cookies and server-side authorisation
  • Configurators that need per-user state or live pricing from an ERP API
  • MitID or other OIDC callback routes that must run server-side
  • Webhook receivers and other paths that cannot be fully pre-rendered

Both frameworks can coexist. Astro pages link to Next.js paths with normal anchors; cookies and consent banners stay consistent because I wire one CMP configuration and one script loader policy.

#Example island on a Danish product page

Below is a simplified pattern I use on Copenhagen builds: static compliance copy, one React island for a tariff or dosage calculator that hydrates only when visible.

---
import BaseLayout from '../../layouts/BaseLayout.astro';
import DkComplianceList from '../../components/DkComplianceList.astro';
import DosageEstimator from '../../components/react/DosageEstimator.tsx';
import { getProductBySlug } from '../../lib/cms';

const { slug } = Astro.params;
const product = await getProductBySlug(slug, 'da-DK');
---

<BaseLayout title={product.title} lang="da-DK">
  <main>
    <h1>{product.name}</h1>
    <DkComplianceList items={product.regulatoryNotes} />
    <DosageEstimator client:visible productId={product.id} locale="da-DK" />
  </main>
</BaseLayout>

The layout, regulatory list and body copy ship without a default Astro JavaScript runtime. Only DosageEstimator loads client code, and only after it enters the viewport, which keeps interaction work off the critical path for Largest Contentful Paint.

Danish sites must respect EU GDPR and national supervision by Datatilsynet (datatilsynet.dk). For terminal storage, Denmark implements the ePrivacy logic through executive orders on retsinformation.dk; the cookie rules are published as BEK nr. 1147 af 08/12/2011 under the Ministry of Business (retsinformation.dk/eli/lta/2011/1147).

In migration terms, the order of operations matters more than the banner skin:

  1. First paint: HTML, CSS, fonts you host yourself, and strictly necessary session cookies for features the user explicitly opened (for example a logged-in portal path on Next.js).
  2. Consent layer: Clear information and granular choice before non-essential cookies or similar storage, aligned with GDPR consent rules Datatilsynet enforces.
  3. Only after opt-in: Analytics, remarketing, A/B frameworks and social embeds that read or write the device.

Next.js root layouts that import analytics in _app or global providers often violate that order because hydration runs third-party code before the CMP finishes. On Astro marketing routes I keep third-party scripts out of the static HTML entirely and inject them from one guarded loader triggered by the consent callback. On remaining Next.js routes I apply the same loader so Copenhagen’s Danish and English editions behave identically.

I do not promise audit outcomes. I document network captures on staging that show zero non-essential requests before consent, and I list every script in the privacy notice appendix your DPO signs off.

#Core Web Vitals: only the web.dev thresholds

Marketing slides sometimes invent performance targets. I do not. For Copenhagen staging sign-off I use Google’s public thresholds from web.dev/articles/vitals:

MetricGood threshold (field data, 75th percentile)
Largest Contentful Paint (LCP)within 2.5 seconds
Interaction to Next Paint (INP)within 200 milliseconds
Cumulative Layout Shift (CLS)at or below 0.1

I compare CrUX or your RUM export before and after the island cutover on the same URL set. Laboratory scores are supporting evidence only; field data at the 75th percentile is the gate.

Removing global hydration on article and documentation templates usually improves INP because the main thread is not busy parsing a large React runtime on pages without interactive islands. I state what changed in the build, not a guaranteed Search Console ranking outcome.

#URLs, Danish and English locales, and staged cutover

Many Copenhagen sites run Danish (da-DK) and English (en) pairs. Before templates move, I build a URL inventory from:

  • XML sitemaps and internal crawl of the live Next.js site
  • Google Search Console export for the Danish property
  • Server access logs for PDF and legacy .aspx or Drupal paths still receiving traffic

Each row gets a decision: keep path identical, 301 once to a new slug, or retire with your written approval. Google’s redirect documentation recommends permanent server-side 301 responses when the move is lasting (developers.google.com/search/docs/crawling-indexing/301-redirects). I deploy rules at the CDN edge and test that no URL returns a chain longer than one hop.

hreflang is rebuilt from translation groups in the CMS, not copied from brittle plugin output. Each Danish page lists its English sibling and itself; each English page returns the link. I validate the rendered HTML, not the database field alone.

Cutover is staged by route family: documentation first, then marketing, then the remaining Next.js app paths. Edge rules send one family to Astro while the rest stays on the old deployment until redirects and consent checks pass on staging.

#How the written project runs from inventory to cutover

  1. Access and crawl. You grant read access to the Next.js repo, the hosting dashboard and Search Console. I crawl the live site and export every indexable URL, attachment and locale prefix.
  2. Route and island map. I deliver a table listing each route family, the target framework, CMS data sources and interactive islands. You approve or adjust in writing.
  3. Consent and tag plan. I list every third-party script, its category (strictly necessary vs marketing), and the load hook after consent. Legal review on your side stamps the privacy notice.
  4. Template build on a branch. Astro layouts and islands land on a feature branch with preview URLs. Next.js routes that stay receive the same consent wrapper.
  5. Redirect file and hreflang QA. I run automated checks for status codes, redirect depth and reciprocal hreflang on built HTML for da-DK and en.
  6. Staged edge cutover. One route family at a time moves behind the CDN. I watch Search Console coverage and server logs for unexpected 404 responses.
  7. Handover note. Editors receive a short guide for publishing through the headless CMS into Astro collections.

#Risks I check before switching production traffic

  • Hidden API routes in the Next.js monolith that marketing pages still call after the split. I trace fetch usage per template before cutover.
  • Locale cookies that set language before consent. Necessary language switching stays; tracking pixels tied to locale do not.
  • PDF and annex URLs hosted outside the Next.js app but linked from ranked articles. They stay in the inventory with the same path or a single 301.
  • Structured data generated by a React helper in the old layout. I compare JSON-LD output per template against the legacy HTML archive.
  • Editorial shortcuts such as embedding unvetted iframes in MDX. Those are either removed or gated behind consent on staging review.

#What I deliver and how you can start

At handover you receive the Git repository, deployment configuration, the URL map with redirect file, a short editor note for the headless CMS, and staging evidence for vitals and cookie order. You also get a list of routes that still require Next.js so your team can plan the next release without guessing.

Send me your public site URL, repository access or export, languages in production, and which routes need login or payment via the contact page. Mention if cross-border campaigns use separate tracking domains or consent banners per market. I reply in writing on working days with an island map, consent loading order, and a phased cutover plan you can approve before any production DNS change.

See also in Denmark

Need this service: Next.js / Astro migration in Copenhagen?

Describe your project by e-mail. I reply within 48 hours.

Describe your project

Explore other WordPress services and knowledge base