Most sites do not lose rankings because of one bad page. They lose because their pages sit in a pile instead of a structure. Site architecture SEO is the discipline of arranging URLs, hubs, and internal links so that both crawlers and readers can see the full shape of your topical coverage in a few clicks.
This guide takes a different angle from the usual “keep it three clicks deep” advice. We are going to plan the architecture before a single article is published, because that is what separates sites that build authority in months from sites that end up with 400 orphaned posts and a flat traffic line.
What site architecture actually means for SEO
Site architecture is the arrangement of your URLs into groups and connections: which pages exist, how they are nested, how they link to each other, and how link equity flows between them.
It sits on three layers that people often confuse:
- Information architecture: the logical grouping of topics and subtopics (the map).
- URL structure: how that map is expressed in folders and slugs (the address system).
- Internal linking: how crawlers and users actually move between pages (the roads).
You can have a clean URL structure and still have terrible architecture if nothing links to your money pages. You can also have messy URLs and strong architecture if your hubs and internal links are disciplined. Links beat folders when the two disagree, but the cleanest sites make them agree.
Why architecture drives topical authority
Topical authority is not a metric inside Google. It is the practical result of covering a subject completely, consistently, and in a way that is machine readable. Architecture is what makes coverage readable:
- Crawl efficiency: shallow, well linked pages get discovered and re-crawled faster.
- Context: a page about “cold email deliverability” means more when it sits inside a hub about cold email and links to 8 sibling pages on the same subject.
- Equity distribution: external links hitting your hub can be passed down to commercial and supporting pages.
- Intent separation: a clear structure prevents two pages from competing for the same query.

Map clusters before you publish (the step almost everyone skips)
The single biggest architecture mistake is publishing first and organizing later. Retrofitting a silo onto 200 existing posts costs redirects, lost equity, and weeks of internal link cleanup. Mapping first costs one afternoon in a spreadsheet.
Step 1: Define your entity and your money pages
Start from what you sell, not from a keyword export. Write down:
- The 3 to 6 core services or product categories you monetize.
- The audience each one serves.
- The problems that lead someone to buy.
Those core categories become your top level silos. Everything else supports them. If a proposed cluster does not support a money page or a strategic topic, it belongs on the backlog, not in the architecture.
Step 2: Build the cluster sheet
Create one row per planned URL before writing anything. A minimal cluster sheet looks like this:
| Working title | Target query | Intent | Role | Parent hub | Planned URL | Links out to |
|---|---|---|---|---|---|---|
| Link building guide | link building | Informational | Hub | Home | /link-building/ | all children |
| Guest posting outreach | guest post outreach | Informational | Spoke | /link-building/ | /link-building/guest-post-outreach/ | hub + 3 siblings |
| Email templates | outreach email templates | Informational | Spoke | /link-building/ | /link-building/outreach-email-templates/ | hub + service page |
The “Links out to” column is the one that prevents orphaned pages. If you decide the internal links at planning time, the writer implements them on day one instead of you auditing them six months later.
Step 3: Set a minimum viable cluster
Do not publish a hub with two children. A cluster reads as authoritative when it answers the obvious follow-up questions. A workable minimum:
- 1 hub page targeting the head term or category.
- 6 to 12 spokes covering subtopics, comparisons, how-tos, and definitions.
- 1 to 3 conversion pages (service, tool, product) linked from the hub and relevant spokes.
Publish clusters in full batches rather than dripping one article per topic across five topics. Depth in one silo beats a thin layer across ten.
Flat vs deep architecture: examples that make the difference concrete
“Flat” means most pages sit within a few clicks of the homepage. “Deep” means users and crawlers must pass through several intermediate levels. Flat is generally better for SEO, but flat does not mean unstructured.
Deep architecture example (avoid)
Home
└ /resources/
└ /resources/blog/
└ /resources/blog/seo/
└ /resources/blog/seo/link-building/
└ /resources/blog/seo/link-building/guest-posting/
└ /resources/blog/seo/link-building/guest-posting/templates/
Six levels deep. The final page is five or six clicks from the homepage, receives almost no internal equity, and its URL is bloated with filler folders like resources and blog that carry no topical meaning.
Flat architecture example (preferred)
Home ├ /link-building/ (hub) │ ├ /link-building/guest-posting/ │ ├ /link-building/outreach-email-templates/ │ ├ /link-building/broken-link-building/ │ └ /link-building/anchor-text/ ├ /technical-seo/ (hub) │ ├ /technical-seo/site-architecture/ │ └ /technical-seo/crawl-budget/ └ /services/ (money pages)
Every content page is two clicks from the homepage, the folder names carry topical meaning, and siblings link to each other horizontally.
Comparison table
| Criteria | Flat architecture | Deep architecture |
|---|---|---|
| Click depth | 1 to 3 clicks | 4+ clicks |
| Crawl frequency | High and even | Low on deep pages |
| Equity distribution | Spread across the cluster | Diluted at each level |
| Best for | Blogs, SaaS, service sites, sites under 5,000 URLs | Large ecommerce and marketplaces with faceted catalogs |
| Main risk | Becoming an unsorted pile without hubs | Orphaned and rarely crawled pages |
Rule of thumb: aim for a maximum of three levels of hierarchy and no important page more than three clicks from the homepage. Large ecommerce catalogs can justify a fourth level, but everything below that should be handled with pagination logic, filtering rules, and cross-links rather than more folders.

How to plan URL structure
URLs are not a ranking lever on their own, but they are a permanent commitment. Changing them later means redirects, lost equity, and broken external links. Decide the rules once:
- Use topic folders, not format folders.
/link-building/anchor-text/beats/blog/anchor-text/. - Keep slugs short and stable. Drop years, stop words, and campaign names.
- Use lowercase and hyphens. No underscores, no uppercase, no parameters in canonical URLs.
- One page, one URL. Decide trailing slash, HTTPS, and www once and enforce it with redirects.
- Do not encode depth you do not need. Two folder levels is plenty for most sites.
- Reserve top level folders for silos and money pages so the root does not become a dumping ground.
If you inherit a legacy structure, remember that reorganizing internal links is cheap while reorganizing URLs is expensive. Fix the linking first, then migrate URLs only if the gain is clearly worth the risk.
Building hub pages that actually work
A hub page (pillar page, category page, guide) is the entry point to a silo. Weak hubs are just a list of links. Strong hubs rank on their own and pass authority downward.
Anatomy of a strong hub
- A real answer above the fold. Define the topic and give the summary a reader could use immediately.
- Section-by-section coverage of the subtopics, each with a descriptive link to the spoke page that goes deeper.
- Descriptive anchor text matching the spoke’s target query, not “read more” or “click here”.
- Contextual conversion links to the relevant service, tool, or product page.
- Visible navigation placement: hubs should be reachable from the main menu or footer, not just from a blog index.
- An update discipline: every new spoke gets added to its hub the day it is published.
Hub types by site model
| Site type | Hub format | Typical spokes |
|---|---|---|
| SaaS | Ultimate guide + use case pages | How-tos, comparisons, integrations, templates |
| Ecommerce | Category page with editorial intro | Subcategories, buying guides, product pages |
| Local service | Service page per core offer | City pages, pricing, FAQs, case studies |
| Publisher | Topic landing page | Explainers, news, evergreen deep dives |
Silo depth and internal linking rules
A silo is a themed group of pages that link tightly to each other and to their hub. Strict siloing (never linking outside the silo) is outdated and hurts users. Modern practice is themed clustering with intentional cross-links.
The linking rules that keep a silo coherent
- Hub to spoke: the hub links to every spoke it owns.
- Spoke to hub: every spoke links back to the hub with a consistent anchor, ideally in the first 200 words.
- Spoke to sibling: each spoke links to 2 to 5 siblings where genuinely relevant.
- Spoke to conversion: at least one contextual link to the related money page.
- Cross-silo links: allowed, but only when the connection is real and the anchor makes the relationship obvious.
- No page without an inbound internal link. Ever.
Orphaned pages: how they happen and how to prevent them
An orphaned page has no internal links pointing to it. It may exist in the sitemap, but it receives no equity and gets crawled rarely. Common causes:
- Posts published to a blog feed that pushes them off the index after a few weeks.
- Landing pages built for campaigns and never linked from the site.
- Tag and archive pages that were noindexed without replacing their linking role.
- Content migrated from an old CMS without link rebuilding.
Prevention checklist: before hitting publish, confirm the page has (1) a link from its hub, (2) at least two links from siblings, and (3) an entry in your cluster sheet. That is a 60 second check that prevents a permanent problem.

Where backlinks fit into your architecture
Architecture decides what happens to authority after it arrives. If you earn links to scattered pages with no internal outlets, that equity dead-ends. If you point external links at hubs, and your hubs link cleanly to spokes and money pages, one strong link benefits an entire cluster.
Practical approach:
- Choose 2 or 3 link targets per silo: usually the hub plus one flagship asset (data study, tool, template).
- Direct outreach and guest post links to those targets rather than spreading them thin.
- Make sure those targets link down to the pages you want to rank commercially.
- Re-audit quarterly: pages that accumulate links but link out to nothing are wasted assets.
Auditing an existing site: a practical order of operations
If your site already exists and looks like a pile, do not start with a replatform. Work in this order:
- Crawl the site and export URL, click depth, inbound internal links, and indexation status.
- List orphaned and deep pages (zero internal links, or depth of 4+).
- Group existing URLs into intended clusters in a spreadsheet. Anything that fits nowhere is a candidate to merge, redirect, or delete.
- Identify cannibalization: two or more URLs ranking for the same query. Consolidate into the stronger one with a 301.
- Create the missing hubs. Most sites have plenty of spokes and no entry points.
- Rebuild internal links hub to spoke, spoke to hub, spoke to sibling.
- Fix navigation and breadcrumbs so the hierarchy is visible to users and crawlers.
- Only then consider URL migration, and only for clusters where the folder path actively confuses the structure.
Metrics that tell you the architecture is improving
| Metric | Where to find it | Healthy direction |
|---|---|---|
| Average click depth | Site crawler | Decreasing toward 2 to 3 |
| Orphaned URLs | Crawler + sitemap comparison | Zero |
| Indexed vs published pages | Search Console | Gap shrinking |
| Crawl requests on new pages | Crawl stats / server logs | Faster first crawl |
| Queries per cluster | Search Console, filtered by folder | Growing keyword count |
| Cannibalized queries | Search Console page comparison | Decreasing |

Common site architecture mistakes
- Publishing before mapping. Every later fix costs ten times more.
- Mega menus with 200 links. They flatten everything and remove all prioritization signals.
- Date-based blog URLs. They add depth and signal staleness.
- Tag pages as pseudo hubs. Auto-generated tag archives are thin and rarely rank. Build editorial hubs instead.
- One spoke per silo. A hub with three links is not a topic, it is a placeholder.
- Ignoring pagination and facets. On ecommerce, uncontrolled filters can multiply URLs and drown the real pages.
- Breadcrumbs that do not match the URL path. Conflicting hierarchy signals confuse both users and parsers.
- Money pages isolated from content. If no article links to your service page, content is not doing its job.
A 7-day plan to structure a new site
- Day 1: List money pages and define 3 to 5 top level silos.
- Day 2: Research subtopics per silo and group them by intent.
- Day 3: Build the cluster sheet with planned URLs and internal link targets.
- Day 4: Design navigation, breadcrumbs, and footer to expose the hubs.
- Day 5: Write and publish the hub pages first, even in a lean version.
- Day 6: Publish the first batch of spokes for silo one, fully interlinked.
- Day 7: Crawl the site, confirm zero orphans, submit the sitemap, and plan the next batch.
Then repeat silo by silo. Finish one cluster before opening the next one. That single rule of sequencing does more for topical authority than any amount of on-page tweaking. This discussion raises a few points we skipped.
FAQ
What is the best site architecture for SEO?
A flat, themed structure: hubs at the top level, spokes one level down, no important page more than three clicks from the homepage, and every page linked from at least one hub and two siblings.
How deep should a silo be?
Two levels for most sites (hub then spokes), three levels maximum for large catalogs. Beyond that, use cross-linking and filtering rather than adding folders.
Is flat architecture always better than deep?
Flat is better for crawl efficiency and equity distribution on most sites. Very large ecommerce catalogs need more hierarchy for usability, but they should compensate with strong category hubs, breadcrumbs, and internal cross-links so deep products stay reachable.
Do URLs need to match the site hierarchy?
It helps for clarity and reporting, but internal links carry more weight than folder paths. If your URLs are already established, improve the linking structure first and only migrate URLs when the benefit clearly outweighs the migration risk.
What is the 80/20 rule in SEO?
Roughly 20% of your pages generate about 80% of your organic results. In architecture terms, identify that 20% (usually hubs and money pages), give them the strongest internal linking, and point your external link building at them.
How many pages does a content hub need?
Aim for a hub plus 6 to 12 spokes as a starting point. Fewer than that and the cluster reads as incomplete compared to competitors covering the same topic in depth. This reference is the one worth keeping handy.
How do I find orphaned pages?
Crawl your site, export all URLs with their inbound internal link count, then compare that list to your XML sitemap and your CMS export. Any URL that exists but has zero internal inbound links is orphaned and needs to be linked from its hub or removed.
Should silos be strictly separated?
No. Strict siloing that forbids all cross-links harms user experience and hides useful relationships. Keep the majority of links inside the cluster, and cross-link between silos when the connection genuinely helps the reader.
Key takeaways
- Map clusters in a spreadsheet before publishing, including planned URLs and internal links.
- Keep the structure flat: hubs at level one, spokes at level two, three clicks maximum.
- Use topic folders instead of format folders, and freeze your URL rules early.
- Publish complete clusters instead of scattering single posts across many topics.
- Every page needs an inbound internal link from its hub and its siblings, without exception.
- Point your backlinks at hubs and flagship assets so equity flows through the whole cluster.
Structure is the cheapest SEO leverage you have. It costs nothing but discipline at planning time, and it compounds with every page you publish afterward.
