,

A platform migration that did not lose traffic

2,400 URLs, a documentation subdomain move, and a shift to client-side rendering on the templates carrying most of the organic traffic.

Client
Nordvik Software
Period
Sep 2023 – Aug 2024
My role
Technical SEO lead, alongside a four-person engineering team
Services
Site Migration, Technical SEO
0%Organic traffic lost90 days post-launch vs 90 days prior, Search Console
2,400URLs migratedIncluding ~300 with traffic but no internal links
+38%Organic trial signups12 months post-launch, GA4 attributed

Context

Nordvik Software sell project management tooling to mid-market construction firms. They were moving from a five-year-old WordPress marketing site onto a headless Next.js stack, and organic accounted for just under half of trial signups.

The problem

There was no problem yet. That was the point — I was brought in before the migration rather than after one had gone wrong, which is rare and is the only reason this went the way it did.

The risks were ordinary and severe: 2,400 URLs changing shape, a documentation subdirectory moving to a subdomain, and a rendering change from server-rendered HTML to client-side hydration on templates that carried most of the organic traffic.

What I did

I built the redirect map from crawl data and Search Console rather than from the sitemap alone, which caught around 300 URLs that still earned traffic but were no longer linked from anywhere on the site.

For the rendering change I pushed for server-side rendering on the templates that mattered and accepted client-side hydration everywhere else. That was a compromise — the engineering cost of SSR everywhere was not justified by the traffic on the remaining templates.

We staged the whole thing behind authentication and crawled it as though it were live, then compared internal link graphs before and after. That comparison found a navigation change that would have orphaned the entire resources section.

The results

The migration completed without measurable organic loss. Rankings wobbled for about three weeks, which is normal, and recovered without intervention. The documentation subdomain took longer, roughly ten weeks, which was expected and was communicated up front.

What I would do differently

Very little on the technical side. I would have pushed harder for a longer pre-launch freeze on content changes — editors kept publishing during the final fortnight, which made it harder than it needed to be to tell migration effects apart from ordinary fluctuation.

Got a similar problem?

Tell me what you are working on and what has already been tried. I reply to everything within two working days.

Start a conversation