Site Architecture: What Matters to Google
Site architecture is how pages are grouped, named and linked to each other. The most important part is not the number of clicks from the home page but three practical things: pages are reachable through ordinary links, every piece of content has one clear address, and visitors can tell where they are. Below: what Google’s documentation says, how to arrange sections, navigation and pagination, and which popular rules (“no deeper than four clicks” and the like) the documentation does not support.
What the documentation says and what it does not
Google’s SEO starter guide advises organizing your site logically, but immediately warns that you should not drop everything and reorganize for search engines: they will likely understand your pages as they are right now, however the site is organized. The main audience for structure is the visitor, who needs to find their way around and locate sections. So change the architecture when it gets in people’s way or breaks accessibility and creates duplicates, not for the sake of an “ideal” diagram.
The documentation has no rule that “every page must be within four clicks”, no numeric depth thresholds, no claim that value “fades” with every level, no data that “topical clusters” raise rankings, and no comparison of subdomain and subfolder strength. These are working heuristics from course authors. They can serve as a guide when designing a site, but not as a fact about Google and not as an argument for a migration.
What the documentation does say
1. Descriptive URLs
The help page on URL structure recommends descriptive addresses: words in a URL are useful to people, random identifiers less so. Google can also learn breadcrumbs from the words in a URL. On sites with thousands of URLs, grouping similar pages in directories helps Google estimate how often each directory changes; on small sites it is barely noticeable. Details are in the lesson on URL structure.
2. One piece of content, one address
The help page on duplicate URLs recommends that each piece of content be reachable through one URL. For architecture that means: one product should not sit in three categories under three addresses, filters and sort orders should not generate endless variants, and internal links should use a single canonical form of the address. If duplicates are unavoidable, see the lessons on canonical and duplicate content.
3. Links that can be followed
Google finds pages through links, and the help page on crawlable links explains which ones work: an <a> element with an href attribute pointing to a real address. Menus where navigation happens only through a script click, elements without href, or addresses inserted only after a user action are risky. More in the lessons on navigation, JavaScript SEO and how search engines find pages.
| Navigation element | Risk | What to do |
|---|---|---|
Menu built on <a href> links | low | leave as is |
Navigation via onclick or a button without href | Google may not find the address | replace with an ordinary link |
| Site search as the only way to pages | a crawler does not type queries into forms | add sections and links to the content |
| A page with no incoming internal links | it can only be discovered via the sitemap or external links | link it from a section; the sitemap is a supplement |
A sitemap helps to communicate addresses but does not replace links; see the lesson on the XML sitemap.
4. Breadcrumbs
According to the help page on breadcrumb markup, a trail indicates the page’s position in the site hierarchy and may help users understand and explore a site effectively; Google uses the markup to categorize the information from the page in search results. Breadcrumbs are both navigation for people and internal links up the hierarchy.
Pagination: what Google recommends
The pagination documentation says: each page of a sequence should have its own unique URL (for example ?page=2), pages are connected with ordinary links, fragments (#) are not suitable for splitting, and the first page should not be the canonical of the rest — each page has its own canonical. Google no longer uses rel="next" and rel="prev", although other search engines may. The advice to create a “view all” page and make it the canonical is not in the documentation. For online shops, the guide to designing URL structure recommends parameters such as ?page=2 and ?color=green and advises against session identifiers in internal links.
How to check your own architecture
Measurements work here, not diagrams. Run a crawler and collect, for the important pages, the number of incoming internal links, the server response and the depth in your own crawl (that is your metric, not a Google indicator). Then compare with Search Console: which weakly linked pages already get impressions and which get none. The anchors and placement of internal links to a specific page are shown by the internal link context tool, and general technical errors by the site audit.
Checklist
- Important pages are reachable through
<a href>links, not only via search or script. - URLs are descriptive and stable; session identifiers do not appear in links.
- Every piece of content has one address; filters and sorting do not create endless variants.
- Sections are grouped logically; very large sites group by directory.
- Breadcrumbs exist, with markup where appropriate.
- Pagination: its own URL and its own canonical for each page, ordinary links between pages.
- There are no pages without incoming internal links.
Mistakes
- Rebuilding a site for a “four click depth” that is not in the documentation.
- Pointing the first page of pagination as the canonical for all.
- Building navigation on scripts without
href. - Placing one piece of content under several addresses.
- Changing the URL structure without a plan for domains and redirects.
What next
Next comes search-friendly navigation: how to build menus and filters through which crawlers and people reach the right pages. How architecture relates to topics and content strategy is covered in the lesson on information architecture.