Site Architecture

Every website has an architecture, whether anybody designed it or not. Some sites have one because someone sat down and decided what the topics were and how pages would connect. Most have one because pages were added over the years, each where it seemed to fit at the time. You can usually tell which is which within a minute of browsing: on the first kind, every page leads naturally to the next; on the second, the best article on a subject is three clicks deep and links to nothing.

This guide explains what site architecture is, why it matters, and how to plan one for a site that is meant to grow. Three articles go deeper into its main tools: pillar pages, internal linking and URL design.

What site architecture means

It has four parts, which work together:

1. Hierarchy: which pages sit above which. Typically the home page, then a page per main topic, then the articles within each topic.
2. Internal links: the links between pages, both in navigation and inside the text.
3. Addresses: the URLs, and what they tell people about the page.
4. Navigation aids: menus, breadcrumbs, related articles, sitemaps.

Why it matters

Search engines find pages through links

Google’s SEO starter guide is direct about this: “the vast majority of the new pages Google finds every day are through links”. Its guidance on links adds that every page you care about should have a link from at least one other page on your site. A page that nothing links to, an orphan, may be found through a sitemap, but it is sending a clear message: nobody on this site thinks it matters.

Links tell search engines what pages are about

Google says it uses links “as a signal when determining the relevancy of pages”. The words in a link, its anchor text, tell users and Google something about the page being linked to. A site where articles about the same subject link to each other with descriptive words is telling search engines, page by page, what each topic is and which pages cover it.

Readers need a path

Most visitors do not arrive at your home page. They land on an article from a search, a link or a share, and decide within seconds whether to read on or leave. A good structure gives them a next step that makes sense: the guide this article belongs to, the article that answers their next question.

AI answer engines need clear pages

AI assistants and answer features increasingly quote web pages directly. They do best with pages that answer one question clearly, sit within a coherent topic and are described accurately by structured data. A well-structured site produces exactly those pages. A disorganized one forces every page to explain everything, and usually explains nothing well.

The topical model

There are many ways to organize a site, but for sites built around content, one has proved its worth: organizing by topic.

• Each main topic gets a pillar page: a comprehensive guide that introduces the subject and links to every article on it.
• Each topic has articles that cover one aspect in depth, and link back to the pillar and across to their closest neighbors.
• The home page and the main navigation link to the pillars.

The model was set out by HubSpot Research in 2017 as the “topic cluster” or pillar and cluster approach, and it has become standard practice since. Its strength is that it scales. When each topic has a pillar, every new article has an obvious home and an obvious page to link to, and the site gets clearer as it grows instead of messier. We explain how to build the central piece in our article on pillar pages.

Hierarchy and depth

A useful principle is that important pages should be close to the home page. A pillar page linked from the main navigation is one click away from anywhere; an article linked only from page six of a blog archive is effectively hidden. Google does not publish a rule on how many clicks is too many, so treat this as practical sense rather than a number to hit: the more a page matters, the fewer steps it should take to reach it.

The opposite mistake is just as common. A site where every page links to every other page has no hierarchy at all, and neither readers nor search engines can tell what matters. Structure is as much about the links you do not add as the ones you do.

Addresses

Google recommends descriptive addresses with meaningful words, rather than long strings of numbers, and notes that grouping similar topics in directories can help it learn how often the addresses in each directory change.

That raises a real choice: should an article’s address include its topic, as in /site-architecture/internal-linking/, or should it stand alone, as in /internal-linking/? Both work. Nested addresses show the hierarchy in the URL. Flat addresses never have to change when you reorganize, because an article’s place in the structure is expressed by its links, its breadcrumbs and its navigation rather than its address. On sites that expect to keep evolving, we prefer flat addresses for that reason. What matters most is that the relationship between pages is real: the pillar links to the article, the article links back, and the breadcrumbs show where the reader is.

Whatever you choose, choose once. Changing addresses means redirects, and every redirect is a small risk. We go deeper into slugs, patterns and stable addresses in URL Design: How to Build URLs That Last.

Navigation, breadcrumbs and sitemaps

Menus should lead to the pillars and the handful of pages every visitor needs. A menu with forty items is not navigation; it is a list.

Breadcrumbs show each page’s position in the hierarchy. Google’s documentation describes them as indicating “the page’s position in the site hierarchy”, which helps users understand and explore a site, and Google can use breadcrumb markup to show where a page sits when it appears in search results.

XML sitemaps list every page you want indexed. Google says that if a site’s pages are properly linked, it can usually discover most of the site without one, and that sitemaps matter most for large sites, new sites with few links from elsewhere and sites with a lot of media. They are a backup to good linking, not a substitute for it.

How to plan an architecture

1. List what the site is about. Five to ten main topics is plenty for most sites. If you have thirty, some of them are sub-topics.
2. Assign every existing page to one topic. A page that fits nowhere is either a gap in your topics or a page you do not need.
3. Create or choose a pillar for each topic. The pillar is the page you would want someone to read first.
4. Plan the links. Each article links up to its pillar and across to two or three neighbors; each pillar links down to all its articles.
5. Fix the addresses if they need fixing, once, with redirects.
6. Put the pillars in the navigation and add breadcrumbs.
7. Write down the rules so that every future page follows them.

Common problems

• Orphan pages with no links pointing to them.
• Several pages on the same question, competing with each other instead of one page answering it well.
• Category and tag sprawl: dozens of thin archive pages that each list two posts.
• Buried pillars: the best guide on a topic reachable only through search.
• Links with empty words, such as “click here” and “read more”, that tell nobody where they lead.
• Structure by date: a blog whose only organization is the order posts were published.

Where our products fit

Architecture is the idea our company is named after, so this is where we have the most to say and the most interest in saying it. Every one of our themes is built on the pillar and article structure described here, and our free plugin TWA AEO connects articles to their pillars and prints the matching structured data. For a site that already exists and grew without a plan, WebsiteSecondLife proposes a topical structure and carries it out once you approve it.