How I restructured a managed IT site around services, not products

A multi-location managed IT provider had built its website around its tech stack. Not around what it sells, and not around what anyone types into Google.

I ran the information architecture migration that fixed it, through the agency I was working for at the time, across several months of 2023. I wrote the new service page copy, set the content strategy that ran alongside it, and worked through the redirects and broken links the move turned up.

What the site looked like

There was no service section. There were four of them.

The offering was spread across managed IT services, cloud services, business productivity and communications, as four separate top-level areas. Azure sat in one. Microsoft 365, Power BI and Teams calling sat in another. Voice and video and internet sat in a third.

Two details make the problem concrete. Power Automate had its own page in two of those areas at once, so the site competed with itself for the same search. And guides had been filed as though they were services: a page on how to write an IT budget, and another on maintaining your computer, both sitting inside service paths alongside the actual services.

The October 2023 crawl still held 41 URLs under the old path. This was not a tidy-up.

Underneath all of it is a simpler problem. Nobody searching for an IT support company searches for Power Automate. They search for IT support, or cyber security, or someone to move them to the cloud. The site was organised for the people who built it rather than the people buying from it.

What it became

One hierarchy, organised by service category. IT support. IT security, with cyber security and risk assessment beneath it. Cloud services, with Microsoft 365 beneath that rather than beside it as a peer.

The vendor products did not disappear. They stopped being the organising principle and became what they always were: things you talk about on the page for the service that uses them.

It ran for months, not overnight

This is the part worth knowing if you have one of these ahead of you.

A migration this size does not happen as a single cutover, and the evidence that it did not is sitting in the crawls. One Screaming Frog crawl from October 2023 holds both hierarchies at once, 41 URLs under the old path and five under the new one. That is what a restructure looks like caught mid-flight, and it is the honest shape of this kind of work: sections move as their copy is written and their redirects are ready, not when a plan says so.

Three things ran in parallel with the moves themselves.

The internal link graph was repointed. A 1,938 row inlink export shows links into the new paths from the home page, the contact page, the location pages, the case studies and dozens of blog posts. Redirects preserve a URL. They do not fix a site that still links to the old one, and internal links are the half people skip.

The old paths were cleaned up rather than abandoned. Broken links were traced from the retired pages, with a redirect map and three crawls across January, July and October to check the work rather than assume it.

The content strategy was built at the same time, on a pillar and cluster model. I audited the existing blog, grouped it into categories, checked Search Console for what already ranked, and built three clusters around the Essential Eight, cloud migration and business continuity planning, each with a pillar page carrying the primary keyword.

What the numbers show, and what they do not

Organic clicks went from 463 a month in January 2023 to 1,328 in November, peaking at 1,790 in August. Sessions moved from 2,197 to 3,219 across the same span. Two measurements from two systems, pointing the same way.

Now the caveats, because this is a case where they change what the numbers mean.

The growth sits on blog articles, not on the restructured service pages. That is what a cluster programme is designed to do, and it is not the same claim as “the restructure grew the traffic”. Average position was around 42 and click-through rate around 0.4% in the middle of the year, so this is long-tail informational reach rather than a ranking gain on commercial terms.

The restructure and the content programme ran together, so neither one owns the result. I could tell you a cleaner story by leaving one of them out. It would not be true.

I do not quote the year-on-year percentages, and they look excellent: clicks up 125% to 184% across five consecutive months. They are unusable. Over the same months, the all-channel figures in the analytics read +1%, +2%, −16%, −33%, and in November the two flip outright, one up 129% while the other reads down 27%. A months-long URL migration with both hierarchies live and the internal links being repointed is exactly the thing that breaks page-level continuity in Search Console. The comparison was never going to be readable, so the within-period absolutes above are the figures I stand behind.

Eight monthly reports survive out of thirteen.

What this one is

It is an architecture case study, not a results one. The claim is that I led an information architecture migration from vendor-product organisation to service categories, over several months, with the cluster content strategy and the new service page copy alongside it, and that the traffic rose while it happened.

The restructure is the part I would defend in an interview. The number is the part I would caveat before anyone asked.