How Site Structure Signals Topical Authority
A site organized around its subject looks authoritative to search engines for the same reason it feels authoritative to people: the structure itself is evidence of understanding.
You can learn a lot about someone from how they organize their bookshelf. Not from the books themselves — anyone can buy books — but from the arrangement. A shelf where the sailing books sit together, ordered from beginner to celestial navigation, belongs to someone who has thought about sailing. A shelf where they’re scattered among cookbooks and thrillers belongs to someone who has merely acquired them. The arrangement is evidence, and it’s hard to fake, because producing it requires the very understanding it signals.
Websites work the same way, and search engines read them the same way. When SEO people talk about “topical authority,” they mean roughly this: Google seems more willing to rank a page about, say, sourdough hydration if it lives on a site with substantial, organized coverage of baking than if it’s an isolated post on a site about everything and nothing. People sometimes describe this as a mysterious ranking factor to be gamed. I think it’s simpler and less mysterious than that. The structure of a site is a claim about how its subject decomposes, and a good decomposition is hard to produce without knowing the subject. The signal is trustworthy for the same reason the bookshelf is.
Consider what structure actually consists of, mechanically. It’s three things: URLs, navigation, and internal links. A URL path like /guides/fermentation/starters says this page is a leaf on a branch of a tree, and names the branch. Navigation says which branches the site considers primary. And internal links — the links from one of your pages to another — say which pages the site itself considers related, and which it considers central. A page that many of your other pages link to is one your site treats as important, and search engines have always read internal linking this way, because it’s how the original PageRank insight applies within a single site.
Now imagine two sites with identical articles, differently arranged. On the first, forty articles about email deliverability all live under one section; the overview page links to each specific article; each article links back to the overview and sideways to its neighbors; the piece on SPF records links to the piece on DKIM where it mentions DKIM. On the second, the same forty articles sit in an undifferentiated chronological blog, connected only by a shared tag nobody clicks. The first site has made its understanding legible. A crawler landing anywhere in the section can discover the whole of it and infer the shape of the topic from the shape of the links. On the second site that inference is impossible, not because the knowledge is absent but because it was never written down in the only language structure speaks.
The pattern that falls out of this is usually called hub and spoke, or pillar and cluster: a broad overview page linking down to narrow deep pages, which link back up and across. The names make it sound like an invention of content marketers, but it’s really just the shape knowledge already has. Subjects decompose into subtopics; subtopics relate to each other. A site that mirrors this decomposition isn’t performing authority for Google’s benefit. It’s exhibiting the ordinary structure of understanding, which Google rewards because it correlates with sites that are actually useful.
Which suggests a test I find clarifying: structure fails for machines almost exactly where it fails for humans. If a visitor who lands on your DKIM article can’t find your SPF article without using the search box, neither can a crawler, and both draw the same conclusion — that the site doesn’t know its own contents. The orphaned page, reachable from nowhere, is invisible to crawlers and undiscoverable by readers in one stroke. This convergence is convenient. It means you don’t have to choose between organizing for people and organizing for search engines. They’re the same task, and when GazeSite audits a site’s structure, most of what surfaces — pages no other page links to, sections that don’t cohere, hierarchies that contradict the navigation — reads equally as an SEO finding and as a usability finding. That’s not a coincidence in how I built it. It’s a fact about what structure is.
There’s a corollary about breadth that deserves saying plainly, because the content industry points the other way. Topical authority argues for depth over sprawl. Ten more thin posts on a topic you’ve barely covered don’t make you authoritative on it; they make you sprawling. Whereas thoroughly covering a narrow territory — every real question, linked into a coherent whole — produces authority on that territory even if the site is small. A tightly organized site about one thing outranks a loosely organized site about many things within that one thing, and “within that one thing” is the only arena a small site should be fighting in anyway.
The practical upshot fits in a sentence: before writing the next article, spend the hour connecting the ones you have. Decide the decomposition. Fix the URLs to reflect it, build the hub pages, add the sideways links where one article already mentions another’s subject. This feels like housekeeping, and nobody’s excited by housekeeping. But it’s really the act of writing down what your site knows about how its subject fits together — and that, not the word count, is the thing that was always being measured.
Test your own website
Type your website address and press Test. In about ten seconds you see a grade for your homepage. Then pick what to order. The first option is free.
More articles
The biggest obstacle to accessibility isn't difficulty, it's the belief that you have to fix everything to fix anything.
Read →The same finding means different things on different pages, so a useful report has to change shape with the page it describes.
Read →Social proof works when it answers a doubt at the moment the doubt arises, and placement is what determines whether it does.
Read →