Architecture is decided before anyone opens a design tool
By the time templates exist, the URL structure is set, the navigation reflects the org chart, and the hierarchy has been inherited from whatever the previous site did. Changing any of it afterwards means redirects, lost equity and rework.
This is the cheapest week of the project to get right and one of the most expensive things to correct later.
Five decisions that compound
- URL structure. Short, stable, descriptive, and organised by how buyers think rather than how you are departmentally arranged. No dates unless the content is genuinely time-bound.
- One page, one primary intent. The discipline that prevents your own pages competing — and self-competition is more common than competitor pressure on most sites.
- Hierarchy depth. Anything more than three clicks from the homepage is treated as less important, by crawlers and by people.
- Internal linking as architecture, not decoration. Links from your strongest pages toward your commercial pages are the most underused lever most sites have.
- Where new content will go. A structure that cannot absorb next year's service line will be worked around, and the workaround becomes the mess.
Navigation is not architecture
Navigation is what a person sees; architecture is how everything relates. A site can have clean navigation and incoherent architecture — orphaned pages, three URLs covering one topic, commercial pages five clicks deep with nothing linking to them.
Crawl your own site and look at what links to what. The result usually differs from the sitemap someone drew at the start.
The self-competition test
Take your ten most commercially important queries. For each, name the one page that should own it. If two pages compete for one query, or one page carries two queries, you have found the architecture problem before it costs you anything.