Moving a company website off WordPress

Who it is for

Customers of a healthcare company.

What it solves

Leaving WordPress without losing the Google rankings that bring in enquiries.

Our role

Moving the content, and a new site with an editing panel

What was built

• The move of all content from WordPress, with a completeness check
• A new platform with a panel for editing content in-house
• A map of 82 redirects and a command that verifies them
• Restored data for Google, and addresses of their own for the offer and contact

Technologies

Next.js, Payload, PostgreSQL, TypeScript, Tailwind CSS, Docker

Where it started

The website of a healthcare company ran on WordPress. Enquiries came through search, and Google had 158 addresses indexed. The main requirement was not to lose rankings. That kind of loss is silent, because nothing breaks. Enquiries just stop coming.

We collected all the content

Through the WordPress interface we exported 44 pages, 28 posts, 28 case studies and 247 files, and made a static copy of the site alongside. Each record kept its old address, which later became the redirect map.

The export had two traps. The server cut responses below the requested page size, and a crawler only sees what links lead to. So we compared the copy with the export and filled the gaps.

We checked completeness

We compared the export with a copy of six content tables, without form submissions or accounts. Pages, posts and case studies matched. Comparing addresses recovered one lost file.

We built the new platform

Next.js, Payload CMS and PostgreSQL, with a panel for editing in-house. Running the import again updates records without duplicating them. Database structure changes go through migrations, and publishing in the panel refreshes the site.

We restored the structure and content

• The three-level menu, missing from the export, was rewritten from the live site.
• 39 links to a server being switched off became internal.
• 76 PDF files moved to the media library.
• We kept 19 videos that the converter silently removed. Nothing goes to Google before play is clicked.
• We fixed an import that cut images out of articles.

We built the redirect map

It has 82 rules: 57 generated from data and 25 by hand (menus, archives, plugin leftovers). A checking command goes through every address of the old site and shows where each one leads. The run of 29 October 2026 on the test version: 99 addresses, 67 permanently redirected, 32 unchanged, 0 dead.

We kept what Google sees

• Every page has a canonical address, a description and sharing cards.
• We restored the structured data the plugin used to create: organisation, articles and breadcrumbs.
• The offer and contact have their own addresses, because Google does not treat the #contact fragment as a separate page.
• robots.txt and a sitemap were added. Old WordPress addresses return a real 404.

Security

The form is protected against abuse, and the test version stays out of Google. Publishing is described by scripts that check the result.

Design and panel

The home page and subpages got a new layout. In the panel, lists are grouped and show where each piece of content appears.

What we deliberately did not do

Old image addresses have no redirects, so Google will find the images at their new addresses. We also don’t publish change dates on imported content, because each one would show the import date.