Flat vs. Deep Site Architecture: A Practical Decision Framework

TL;DR — A flat site keeps most pages close to the homepage; a deep site nests them through many category layers. Neither is automatically better — the right amount of depth depends mostly on how clearly your categories divide up, and secondarily on how big your site is. A small brochure site can usually stay flat. A giant online store usually needs more layers. The practical goal is to give important pages short, clear, crawlable paths, and to test changes rather than assume they’ll work; there is no universal three-click rule, and no change guarantees crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor., ranking, or traffic gains.

What “flat” and “deep” actually mean

Every site has some hierarchy — homepage at the top, then sections, then individual pages. The useful definition isn’t a fixed layer count, though — it’s your site’s click-depth distribution (what share of pages sit at depth 1, 2, 3, 4+) and, more specifically, how many clicks your priority pages need. Flat describes a distribution where most pages, especially priority ones, sit one or two clicks from the homepage. Deep describes one where reaching typical pages means passing through several layers: homepage → category → subcategory → sub-subcategory → page. The page-count and level bands later in this article are labeled heuristics for calibrating that distribution — not the definition itself.

The neighboring articles in this cluster (site architectureSite architecture is how a website's pages are organized, categorized, and interlinked. It controls how crawlers discover pages, how link equity flows, and how clearly search engines understand each page's topical context. Silo structure, hub and spoke, and topic clusters are the three common models. and website structureWebsite structure (site architecture) is a site's visible hierarchy, navigation, breadcrumbs, and URL organization — how pages relate and how people and search engines move between them. Internal linking is the primary signal Google reads to understand that structure, not URL folders.) already explain why a sensible pyramid beats both a pancake-flat site and a needlessly deep one. This article is the practical follow-up: how do you decide, for your actual site, how flat or deep it should be — and how do you check and fix it?

The short answer

There’s no universal “correct” number of levels. It comes down to two questions:

1. How many pages do you have that need to be foundA 302 (\"Found\") is a temporary redirect: it forwards users to a new URL while telling search engines the original URL should stay in the index. It's a weak canonicalization signal, not the zero-equity dead end of SEO folklore. on their own? A little business site with 40 pages can keep everything one or two clicks from the homepage. A store with 20,000 products can’t — it needs categories and subcategories, or the navigation becomes an unusable wall of links.
2. How clearly do your categories divide up? If your sections are obvious and don’t overlap (Shoes vs. Shirts), you can stay flatter. If they blur into each other, you’ll usually need a little more structure to keep things organized.

The one rule worth remembering

Keep important pages easy to reach through crawlable links. Google gives no universal click-count threshold. Treat “three clicks” as a diagnostic heuristic, not a search-engine requirement.

If a page you care about is buried five or six clicks down, you don’t have to rip your whole site apart. Usually you just add a hub page (a category-style page) that links straight to it from higher up — that alone can pull it from six clicks to three.

Want the real decision framework — page-count thresholds, how to measure your site’s depth, and how to fix it — switch to the Advanced tab, or jump to the Decision Tree.

TL;DR — This is the “how do I decide and act” companion to the conceptual pyramid-vs-flat coverage already in this cluster — I’m not re-arguing why a pyramid beats both extremes. Depth should scale mostly with (a) how distinct the natural categories, templates, and update patterns are, and secondarily with (b) how many pages need their own findable URL. Rough shape, as heuristics you calibrate rather than fixed cutoffs: ~50-page brochure site → flat; a few thousand pages → shallow pyramid, 2–3 levels; 10,000+ SKUs → multi-level hierarchy with hub pages, or you get the mega-menu link-dump failure mode. Google does not set a maximum number of clicks. Measure by crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor. for a segmented click-depth distribution (by template and priority, not a sitewide average), then cautiously cross-reference GSC Crawl StatsA Google Search Console report (under Settings) that shows how Google has crawled your site over the last 90 days — total requests, download size, and average response time, broken down by response code, file type, Googlebot type, and purpose. It's only available for root-level properties (a Domain property or a URL-prefix property verified at the site's root). — aggregate first-party data, not per-URL proof — to find the depth cliff where crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor. drops off. Fix with hub pages (the highest-ROI move), related-content modules, and breadcrumbsBreadcrumbs are a secondary navigation trail (Home > Category > Page) that shows where a page sits in a site's hierarchy. They create internal links that pass PageRank, and when marked up with BreadcrumbList structured data they can drive the path Google shows in desktop search results., testing each change against a comparison segment with a monitoring window and a rollback trigger. URL-folder depth ≠ click depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops. — link a buried page closer rather than rewriting its URL. None of this guarantees crawling, indexingStoring a crawled page in the search index so it can appear in results. Crawled is not the same as indexed — Google selects what to keep, and indexing isn't guaranteed., ranking, traffic, or AI-citation gains.

This isn’t the “why” article

Two other pieces in this cluster already own the conceptual case. Site architectureSite architecture is how a website's pages are organized, categorized, and interlinked. It controls how crawlers discover pages, how link equity flows, and how clearly search engines understand each page's topical context. Silo structure, hub and spoke, and topic clusters are the three common models. covers flat vs. deep hierarchies as one of the failure modes at the extremes, and website structureWebsite structure (site architecture) is a site's visible hierarchy, navigation, breadcrumbs, and URL organization — how pages relate and how people and search engines move between them. Internal linking is the primary signal Google reads to understand that structure, not URL folders. makes the pyramid-beats-both-extremes argument with Mueller’s quotes about context, crawling, and mega menusA mega menu is a large, categorized navigation panel — usually opened by hover or click on a top-level nav item — that surfaces many links at once, grouped into columns instead of the single list a standard dropdown shows.. I’m not going to re-derive any of that here or re-run those quotes as the centerpiece. If you want the why, read those first.

What neither covers — and what people planning a new site, replatforming, or auditing an existing one actually need — is the operator’s manual: for my specific site, how many levels should I have, how do I measure where I currently sit, and how do I fix it when it’s wrong? That’s this article. The centerpiece is the decision tree in the Decision Tree tab; everything below is the reasoning and workflow behind it.

Depth is a function of category distinctness and size

There is no fixed right answer, and anyone who hands you one (I’ve seen “always flat,” “always three clicks,” “never more than two subcategory levels”) is selling a house style as a law. The honest framing is that hierarchy depth should scale with two independent variables — lead with the first, since it’s the one SEO-only advice usually skips:

1. How distinct or overlapping the natural categories are, and what task the user is on. Nielsen Norman Group’s research on flat vs. deep website hierarchies nails it: “Flat hierarchies tend to work well if you have distinct, recognizable categories, because people don’t have to click through as many levels,” and “Categories that are specific and do not overlap are the easiest to understand.” Their bottom line matches the whole spirit of this framework: “Like most design questions, there’s no single right answer, and going too far to either extreme will backfire.” Also weigh which template a page uses, how important it is to the business, and how often that section updates (Illyes’ /news/ vs. /archives/ point below) — these matter as much as a raw page count.
2. How many distinct items need their own findable page. Forty pages and 20,000 products are different problems, but page count on its own is a rough signal, not the deciding factor — a small site with badly overlapping categories can need more structure than a larger one with clean, distinct sections.

Google’s own guidance points the same direction on the size axis. Gary Illyes has said hierarchy should scale with the site — that for a large site it’s “likely better to have a hierarchical structure” because it lets search engines “treat different sections differently, especially when it comes to crawling,” and that if you “put everything in one directory, that’s hardly possible.” (Reported by Search Engine Journal; I’d treat the exact wording as trade-press transcription rather than a canonical doc.) That’s the load-bearing point for the large-catalog end of the framework: size forces hierarchy.

Page-count rules of thumb

Nobody — not Google, not Bing — publishes a page-count-to-levels table, and no dated study ties the ~50-page or 10,000+-page bands below to a specific site population. Treat these as practitioner heuristics you calibrate to your own site, not verified universal cutoffs — category distinctness, templates, and update patterns (above) should move you off these bands in either direction:

• Small / brochure sites (roughly under 50–100 pages, few natural categories): stay flat. Homepage → one level of section/category → pages, with most content one or two clicks deep. Bing’s three-click figure is your outer bound, and you’ll almost never hit it. NN/g’s “distinct categories work well flat” applies directly here.
• Mid-size content sites (a few hundred to a few thousand pages, several genuinely distinct topic areas): a shallow pyramid — homepage → category → optional subcategory → page, most content within three clicks. This is the sweet spot the cluster’s conceptual pieces describe, and it’s where this very site sits.
• Large ecommerce catalogs, enterprise sites, publisher archives (10,000+ pages or SKUs): hierarchy stops being an aesthetic choice and becomes a necessity. You need enough levels — top-level category → subcategory → (sometimes a filtered/facet layer) → product — that you’re not trying to expose the entire catalog from one navigation surface. The classic failure here is over-flattening a huge catalog into a mega menuA mega menu is a large, categorized navigation panel — usually opened by hover or click on a top-level nav item — that surfaces many links at once, grouped into columns instead of the single list a standard dropdown shows. that dumps hundreds of links one click from the homepage; the mega-menu topic in the ecommerce cluster covers why that strips the grouping signals search engines rely on. The corrective isn’t “add infinite depth” — it’s “add enough hierarchy to organize the catalog, then use hub pages to keep priority items reachable in three-to-four clicks anyway.”

You’ll see competitor guides assert crisp numbers here (“flat = 3 clicks or fewer,” “keep subcategories to 2–3 levels,” “8 top-level categories × 4–8 subcategories”). Those are useful as industry consensus — the SEO field’s de facto defaults — but they’re not sourced to any search engine, so label them that way in your own head. The uncontroversial, well-supported shape is simply: small catalog → flat, large catalog → more hierarchy.

URL depth is not click depth (a reminder, not a re-derivation)

The url-structure and website-structure articles in this cluster already establish the key fact: Google reads the link graph, not the slashes in your URLs. I’m not re-arguing it. The reason it belongs here is purely operational, because it changes how you fix a depth problem.

A URL like /category/subcategory/product/ looks three levels deep, but if a hub page links to it directly, it’s one click from the homepage. Conversely, a page with a short, tidy URL can be buried six clicks deep with no hub linking to it. So two rules fall out for remediation:

• Don’t “fix” click depth by rewriting URLs flatter. Stripping folders out of the address while the link graph stays deep does nothing.
• Do fix it by linking the page closer. Add or strengthen a hub link from a shallower level. The URL can stay exactly as it is.

That distinction is what makes the worked example later (six clicks → three, without renaming anything) possible.

Auditing your current depth, in order

You can’t decide where to go without knowing where you are. Three steps:

1. Get the click-depth distribution from a crawl, segmented, not averaged. Run Ahrefs Site Audit (its Structure Explorer / depth view) or Screaming Frog (the Site Structure tab and the Crawl DepthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops. column) and look at what percentage of your pages sit at depth 1, 2, 3, and 4+. Don’t stop at the sitewide number — break it out by template, page purpose, business priority, and discovery source (internal linkAn internal link is a hyperlink from one page on a website to another page on the same website. Internal links help search engines discover your pages and pass ranking signals (PageRank and anchor-text context) between them. vs. sitemapA sitemap is a file that lists the pages, images, videos, and other files on your site so search engines can discover them. It helps discovery, but submitting a sitemap doesn't guarantee crawling or indexing.-only). See the crawl depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops. article for the URL-level measurement mechanics. This is your ground-truth map of how deep the site actually is in the link graph — not in the URLs. A healthy site has its priority pages, specifically, concentrated in the shallow buckets; a good overall average can still hide a buried revenue template.

2. Cross-reference with GSC Crawl Stats — cautiously. The crawl tells you how deep pages are; Google’s Crawl Stats report tells you what Google is actually choosing to crawl, and how often. It splits requests into Discovery (URLs Google hadn’t crawled before) and Refresh (recrawls of known pages). Treat Crawl Stats as aggregate, sitewide, first-party data — it isn’t a per-URL log, so it can’t by itself prove that any single page or segment is suffering from depth specifically. The practical read — this is practitioner interpretation, not a Google statement: if Discovery stays near zero while you keep publishing, your internal linkingLinks between pages on the same site. isn’t surfacing new or deep pages to the crawlerA crawler — also called a spider or bot — is an automated program that fetches web pages, extracts their links, and queues new URLs to visit. Search engines use crawlers to discover and download content for their index.; if Refresh drops sharply without pages being removed, something structural is suppressing recrawls. Almost no competitor guide connects the crawler’s depth data to GSC’s crawl data, and the join is where the insight lives — but treat it as a hypothesis to test (compare a segment you changed against a similar segment you didn’t), not a proof on its own.

3. Find the depth cliff. Overlay the two: the depth level where crawl frequencyCrawl frequency is how often a search engine comes back to re-fetch a page it already knows about. Popular pages that change often get refreshed many times a day; stable pages can go weeks or months between crawls — and you influence it indirectly, not by setting a dial. and coverage fall off a cliff is the practical, measured definition of “too deep” for your site — far more useful than any generic number. You’re not guessing that depth 5 is bad; you’re observing that on your site, crawling collapses past depth 4 for a specific segment. (Note: you’ll find secondary content claiming things like “pages get crawled 5–10× less at depth 5.” I couldn’t source that to any primary Google material, so I won’t present it as a stat — measure your own cliff instead, and hold it to a comparison group before you act on it.)

Fixing depth problems

In rough priority order:

• Hub / category pages — the single highest-ROI fix. Add or strengthen a mid-level page that both humans and crawlers can use to reach buried content in fewer clicks. It shortens click depth without touching URLs. This is almost always the first move.
• Related-content modules. End-of-article and sidebar links create additional paths to deep pages. They’re a weaker signal than body-content links (Google tries to identify a page’s primary content area and weights contextual links above navigational “module” links — I’d treat that as a well-attested read of Mueller’s position rather than a verbatim quote), but at scale, more real paths to deep pages genuinely help.
• BreadcrumbsBreadcrumbs are a secondary navigation trail (Home > Category > Page) that shows where a page sits in a site's hierarchy. They create internal links that pass PageRank, and when marked up with BreadcrumbList structured data they can drive the path Google shows in desktop search results.. They reinforce hierarchy for users and search engines, and the machine-readable version is BreadcrumbList structured dataStructured data is a standardized way of labeling page content (using the schema.org vocabulary in JSON-LD, Microdata, or RDFa) so search engines can understand its meaning. It's not a direct ranking factor — its value is rich results and entity understanding.. Google’s own line: “A breadcrumb trail on a page indicates the page’s position in the site hierarchy, and it may help users understand and explore a site effectively.” (See the breadcrumbs article in this cluster for the markup.)
• PaginationPagination splits a large set of content — product listings, blog archives, search results — across multiple sequentially numbered URLs. For SEO, each paginated page should be crawlable, indexable, and self-canonical; Google no longer uses rel=prev/next, but Bing still does. — mind how it interacts with depth. The pagination article in this cluster covers the mechanics; the depth-specific point is that a paginated listing is itself a crawl path (page 1 → 2 → 3 …), so a product or article that only appears on page 4+ of a category listing inherits real extra click depth. Don’t rely on deep pagination to carry your priority items — surface them via a hub or related module instead, and keep the pagination self-canonicalizing so the path stays open.

One nuance worth holding onto: depth is a risk factor, not an automatic penalty. A technically deep page can still get crawled fine if it has strong external links or lives in a frequently-updated, well-organized directory (Illyes’ /news/ example). That’s exactly why the depth-cliff measurement beats assuming by depth number alone.

None of these fixes can promise crawling, indexingStoring a crawled page in the search index so it can appear in results. Crawled is not the same as indexed — Google selects what to keep, and indexing isn't guaranteed., rankings, traffic, or AI-citation gains — depth is one input among many. Treat each change as reversible and testable: pick a comparison segment you’re not changing, set a monitoring window (a few crawl/recrawl cycles is usually enough to see a directional shift in Discovery/Refresh or the depth distribution), and define a rollback trigger up front — if a hub page adds clicks for some users without shortening the crawl path you were targeting, unwind it rather than layering more changes on top. See the Validation Tests tab for the fuller test/ monitor/rollback table.

Worked example: six clicks to three

Take a product buried like this: Home → top category → subcategory → sub-subcategory → listing page 3 → product. That’s six clicks. It’s a genuinely important product, but the crawler has to traverse a deep listing to reach it, and it’s getting recrawled rarely.

The fix isn’t a re-architecture and it isn’t a URL change. Add a hub / landing page one level below the homepage — say a “Best Sellers” or seasonal collection page — and link the product directly from it: Home → collection hub → product. Now it’s three clicks. The URL never changed (reinforcing that click depth, not URL depth, was the problem), and you’ve added a shallow, high-value path that both users and crawlers can follow. That single hub page can do the same job for a whole batch of buried priority items at once.

When “get everything crawled” isn’t the goal

A closing reframe I keep coming back to in my crawl-budget writing: more crawling doesn’t mean better rankings. But a page that never gets crawled can’t rank at all — and the pages that go uncrawled tend to be exactly the newer, poorly-linked, deep ones this framework is about. Most sites don’t need to obsess over crawl budgetThe number of URLs an engine will crawl in a timeframe.; Google’s own guidance says the sites that do are roughly the 1M+ page sites changing weekly or ~10K-page sites changing daily. For everyone else, depth is a findability and signal-passing concern, not a crawl-budget emergency — but the fix (link priority pages shallower) is the same either way.

The neighboring topics — how internal linksAn internal link is a hyperlink from one page on a website to another page on the same website. Internal links help search engines discover your pages and pass ranking signals (PageRank and anchor-text context) between them. pass signals, how crawl depth affects discovery, breadcrumbs, pagination, and the site-architecture models — all plug into the decisions in the tree below.

AI summary

A condensed take on the Advanced version:

• This is the “how do I decide” companion, not the “why a pyramid wins” article — that’s already covered by the site-architecture and website-structure pieces in this cluster.
• Depth should scale mostly with category distinctness (templates, priority, update patterns), and secondarily with page count. NN/g: distinct, non-overlapping categories work well flat; there’s no single right answer. Page count alone is a rough signal, not the deciding factor.
• Page-count bands are practitioner heuristics, not verified cutoffs: ~50-page brochure site → flat; a few thousand pages → shallow pyramid (2–3 levels, within 3 clicks as an operational target); 10,000+ SKUs → multi-level hierarchy, or you get the mega-menu link-dump failure. Illyes: large sites are “likely better” with hierarchical structure.
• Only Bing states a number (important pages within ~3 clicks, an operational target, not a guarantee); Google deliberately states none — treat competitor “3-click / 2-level” numbers as industry consensus, not official guidance.
• URL-folder depth ≠ click depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops.. Google reads the link graph. Fix depth by linking a page closer, not by rewriting its URL flatter. See the crawl depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops. article for URL-level measurement mechanics.
• Audit in order, segmented not averaged: (1) crawl for the click-depth distribution by template/priority/discovery source (Ahrefs Site Audit / Screaming Frog); (2) cautiously cross-reference GSC Crawl StatsA Google Search Console report (under Settings) that shows how Google has crawled your site over the last 90 days — total requests, download size, and average response time, broken down by response code, file type, Googlebot type, and purpose. It's only available for root-level properties (a Domain property or a URL-prefix property verified at the site's root). (Discovery vs. Refresh) — aggregate first-party data, not per-URL proof — for what Google actually crawls; (3) find the depth cliff where crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor. drops, treated as a hypothesis to test against a comparison segment, not a proof on its own.
• Fix, in priority order: hub/category pages (highest ROI), related-content modules, breadcrumbsBreadcrumbs are a secondary navigation trail (Home > Category > Page) that shows where a page sits in a site's hierarchy. They create internal links that pass PageRank, and when marked up with BreadcrumbList structured data they can drive the path Google shows in desktop search results. (+ BreadcrumbList schemaBreadcrumbList (schema.org/BreadcrumbList) is structured data — usually JSON-LD — that marks up a page's breadcrumb trail (Home > Category > Page) so search engines can read the page's position in the site hierarchy, drive a readable breadcrumb path in the search snippet, and use it to categorize the page.), and mind how paginationPagination splits a large set of content — product listings, blog archives, search results — across multiple sequentially numbered URLs. For SEO, each paginated page should be crawlable, indexable, and self-canonical; Google no longer uses rel=prev/next, but Bing still does. adds depth to items on page 4+. Treat every change as reversible: set a monitoring window and a rollback trigger, and don’t promise crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor., indexingStoring a crawled page in the search index so it can appear in results. Crawled is not the same as indexed — Google selects what to keep, and indexing isn't guaranteed., ranking, traffic, or AI-citation gains.
• Worked example: a product at 6 clicks (Home → category → subcategory → sub-subcategory → listing page 3 → product) pulled to 3 clicks via a new collection hub — no URL change.
• Depth is a risk factor, not an automatic penalty: a deep page with strong links or in a fresh directory can crawl fine. Measure the cliff; don’t assume by number.

How many hierarchy levels should your site have?

This is the centerpiece. Walk it top to bottom: it branches first on site size, then on category distinctness, then on content-update patterns, and lands on a concrete recommendation with a target click depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops. and a first remediation move.

Whatever leaf you land on, the audit and remediation are the same: crawl for your click-depth distribution, cross-reference GSC Crawl StatsA Google Search Console report (under Settings) that shows how Google has crawled your site over the last 90 days — total requests, download size, and average response time, broken down by response code, file type, Googlebot type, and purpose. It's only available for root-level properties (a Domain property or a URL-prefix property verified at the site's root). to find the depth cliff, and shorten the path to buried priority pages with a hub page first.

Official documentation

Primary-source guidance relevant to deciding, measuring, and fixing site depth.

Google

• SEO link best practices (Make your links crawlable) — links must be <a href>, and “every page you care about should have a link from at least one other page” (the bar deep/orphan pagesAn orphan page is a page on your site that no other page links to internally. Because crawlers discover pages by following links, an orphan page is effectively invisible to search engines unless it's in an XML sitemap or linked from an external site. fail).
• Optimize your crawl budget — how wasted crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor. on low-value URL patterns starves the rest of a large, deep site.
• Crawl Stats report (Search Console Help) — the Discovery vs. Refresh breakdown that powers the depth-cliff diagnosis.
• Breadcrumb (BreadcrumbList) structured data — the remediation markup that signals a page’s position in the hierarchy.
• Help Google understand your ecommerce site structure — hierarchical linking for large catalogs, and how Google infers importance from links.

Bing / Microsoft

• Bing Webmaster Guidelines — the one place either engine states a hard number: keep important pages within roughly three clicks of the homepage, with sitemapA sitemap is a file that lists the pages, images, videos, and other files on your site so search engines can discover them. It helps discovery, but submitting a sitemap doesn't guarantee crawling or indexing.-first discovery.

Quotes from the source

On-the-record statements relevant to deciding depth (rather than re-litigating why a pyramid wins — those quotes live in the site-architecture and website-structure articles). Where the source page supports it, each link jumps to the passage.

Gary Illyes, Google — hierarchy scales with site size

• “For a large site it’s likely better to have a hierarchical structure… that will allow you to do funky stuff on just one section, and will also allow search engines to potentially treat different sections differently, especially when it comes to crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor..”
• “Having a /news/ section for newsy content and /archives/ for old content would allow search engines to crawl /news/ faster than the other directory.”
• “If you put everything in one directory, that’s hardly possible.” — Gary Illyes, Google, SEO Office Hours, reported by Search Engine Journal. Relayed via trade-press transcription of an office-hours video, not a canonical doc page — confirm the exact wording against the original before treating it as final.

Google Search ConsoleA free Google service that reports how a site performs in Google Search and surfaces problems with how Google crawls, indexes, and serves it. It's first-party data straight from Google — but you don't need it to appear in results. Help — the Crawl StatsA Google Search Console report (under Settings) that shows how Google has crawled your site over the last 90 days — total requests, download size, and average response time, broken down by response code, file type, Googlebot type, and purpose. It's only available for root-level properties (a Domain property or a URL-prefix property verified at the site's root). mechanism behind the “depth cliff”

• Discovery: “The URL requested was never crawled by Google before.” / Refresh: “A recrawlCrawl frequency is how often a search engine comes back to re-fetch a page it already knows about. Popular pages that change often get refreshed many times a day; stable pages can go weeks or months between crawls — and you influence it indirectly, not by setting a dial. of a known page.” — Crawl Stats report. From a JavaScript-rendered help-center page; confirm the phrasing on the live page before treating it as final.

Google Search Central docs — breadcrumbsBreadcrumbs are a secondary navigation trail (Home > Category > Page) that shows where a page sits in a site's hierarchy. They create internal links that pass PageRank, and when marked up with BreadcrumbList structured data they can drive the path Google shows in desktop search results. and hierarchy

• “A breadcrumb trail on a page indicates the page’s position in the site hierarchy, and it may help users understand and explore a site effectively.” — Breadcrumb (BreadcrumbList) structured data. The page uses JS-rendered tabs; confirm the deep-linked passage resolves on the live page.

Nielsen Norman Group — the UX-research counterweight (category distinctness)

• “Flat hierarchies tend to work well if you have distinct, recognizable categories, because people don’t have to click through as many levels.”
• “Categories that are specific and do not overlap are the easiest to understand.”
• “Like most design questions, there’s no single right answer, and going too far to either extreme will backfire.” — Kathryn Whitenton, Flat vs. Deep Website Hierarchies, Nielsen Norman Group. Not a search-engine source — cited as the strongest available UX research on tying depth to category distinctness and findability, which the SEO-only guidance tends to miss.

Depth decision & audit checklist

Work it top to bottom — decide the target first, then measure, then fix:

• Sized the site. You know roughly how many pages need their own findable URL (the first branch of the decision tree).
• Judged category distinctness. You know whether your natural categories are distinct (can stay flatter) or overlapping (need hubs / cross-links).
• Set a target. Priority pages have an explicit click-depth target (~3 clicks for most sites; 3–4 for large catalogs).
• Crawled for the click-depth distribution. You’ve run Ahrefs Site Audit or Screaming Frog and know what % of pages sit at depth 1 / 2 / 3 / 4+.
• Cross-referenced GSC Crawl StatsA Google Search Console report (under Settings) that shows how Google has crawled your site over the last 90 days — total requests, download size, and average response time, broken down by response code, file type, Googlebot type, and purpose. It's only available for root-level properties (a Domain property or a URL-prefix property verified at the site's root).. You’ve checked Discovery vs. Refresh and response patterns for what Google is actually crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor..
• Located the depth cliff. You’ve identified the depth level where crawl frequency/coverage drops off — your measured “too deep,” not a generic number.
• Priority pages are shallow. No money page is buried 5–6 clicks deep with no hub linking to it.
• No mega-menu over-flattening. A large catalog isn’t dumped into one navigation surface (grouping signals intact).
• Hub pages in place. Buried priority segments have a shallow hub/collection page linking straight to them.
• BreadcrumbsBreadcrumbs are a secondary navigation trail (Home > Category > Page) that shows where a page sits in a site's hierarchy. They create internal links that pass PageRank, and when marked up with BreadcrumbList structured data they can drive the path Google shows in desktop search results. present with BreadcrumbList structured dataStructured data is a standardized way of labeling page content (using the schema.org vocabulary in JSON-LD, Microdata, or RDFa) so search engines can understand its meaning. It's not a direct ranking factor — its value is rich results and entity understanding..
• PaginationPagination splits a large set of content — product listings, blog archives, search results — across multiple sequentially numbered URLs. For SEO, each paginated page should be crawlable, indexable, and self-canonical; Google no longer uses rel=prev/next, but Bing still does. isn’t the only path to priority items that fall on page 4+ of a listing.
• You didn’t “fix” depth by rewriting URLs. Click depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops. was shortened by linking closer, not by flattening the URL path.

The mental models

1. Depth is a dial set by two inputs. Turn it based on (a) how many findable pages you have and (b) how distinct your categories are. Small + distinct → flat. Huge + overlapping → hierarchy plus hubs. Everything else is between.

2. Only Bing gives you a number. Bing’s ~3-click target is the one hard figure either engine states; Google deliberately talks link-graph proximity, not level counts. Treat competitor “2–3 levels / 3 clicks” numbers as industry consensus, not law.

3. Measure the cliff, don’t assume the number. “Too deep” isn’t a universal constant. Crawl for your depth distribution, overlay GSC Crawl StatsA Google Search Console report (under Settings) that shows how Google has crawled your site over the last 90 days — total requests, download size, and average response time, broken down by response code, file type, Googlebot type, and purpose. It's only available for root-level properties (a Domain property or a URL-prefix property verified at the site's root)., and find the depth where crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor. actually falls off. That’s your site’s “too deep.”

4. Fix with links, not URLs. URL-folder depth ≠ click depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops.. To pull a page shallower, link it from a hub closer to the homepage. Rewriting its URL flatter changes nothing.

5. Hub page first. Of every remediation, adding or strengthening a mid-level hub page is the highest ROI — it can drop a batch of buried items from six clicks to three at once, with no URL changes.

6. Depth is a risk factor, not a penalty. A deep page with strong external links or in a fresh, well-organized directory can crawl fine. Depth raises risk; it doesn’t guarantee harm. The measurement, not the number, tells you whether it’s actually a problem.

Worked example: pulling a product from 6 clicks to 3

Before — 6 clicks deep

The product is a priority item, but a crawlerA crawler — also called a spider or bot — is an automated program that fetches web pages, extracts their links, and queues new URLs to visit. Search engines use crawlers to discover and download content for their index. has to traverse a deep, paginated listing to reach it — so it’s recrawled rarely and inherits little internal importance.

After — 3 clicks deep

What changed: one new hub page, linked from the homepage or main nav, that links directly to the product (and its peer priority items). The product’s URL never changed — it can still be /footwear/running/trail/model-x/. That’s the whole point of the URL-depth-vs-click-depth distinction: the fix is a shallow link path, not a URL rewrite. One hub page can do this for an entire batch of buried priority items simultaneously.

Reading a depth distribution

A crawl of a mid-size site might come back like this:

The number that matters isn’t “15% at depth 5+” in the abstract — it’s which pages are in that bucket. Cross-reference GSC Crawl StatsA Google Search Console report (under Settings) that shows how Google has crawled your site over the last 90 days — total requests, download size, and average response time, broken down by response code, file type, Googlebot type, and purpose. It's only available for root-level properties (a Domain property or a URL-prefix property verified at the site's root).: if those deep pages show little Refresh crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor. and you care about them, that’s your worklist for hub pages. If they’re low-value tail content crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor. fine, leave them alone — depth is a risk factor, not an automatic problem.

Forcing every page to depth one

A homepage or mega-menu that links to everything creates a low depth number but no useful hierarchy. Keep the highest-value sections prominent, then use category and hub pages to provide clear paths through the rest of the inventory.

Adding categories that do not help anyone choose

Extra layers are not automatically organization. A subcategory with one child, vague labels, or heavy overlap makes users guess and adds clicks without narrowing the task. Merge weak layers or redesign the taxonomy around real distinctions.

Treating URL-folder depth as architecture depth

Moving /shop/shoes/trail/ to /trail/ does not make the page easier to discover if the same links still lead to it. Measure link paths and change URLs only when there is a separate information-architecture or migration reason.

Using the sitewide average as the decision

A pleasant average can hide a buried revenue template and a pile of shallow low-value pages. Segment depth by template, importance, indexability, and organic role before deciding whether the architecture is too flat or too deep.

Removing hierarchy instead of strengthening hubs

When a deep section underperforms, deleting category levels can create a link dump. First test whether better hub content, contextual modules, paginationPagination splits a large set of content — product listings, blog archives, search results — across multiple sequentially numbered URLs. For SEO, each paginated page should be crawlable, indexable, and self-canonical; Google no longer uses rel=prev/next, but Bing still does., and cross-links can shorten important paths while preserving useful grouping.

The crawl looks flat but important pages are still hard to find

Symptom: median depth is low, yet priority pages remain buried. Likely cause: navigation favors many low-value URLs or the average hides template differences. Fix: join depth to a priority list, inspect shortest paths for those pages, and move links from generic sitewide clutter into relevant shallow hubs.

A new category layer caused crawling or traffic losses

Symptom: child pages moved deeper and performance fell after taxonomy expansion. Likely cause: the new parent is itself hard to reach, links were removed, or old URLs redirectA redirect sends browsers and crawlers from a requested URL to a different one. An HTTP redirect specifically is a 3xx status code paired with a Location header; meta refresh and JavaScript redirects achieve a similar navigation without being a 3xx response themselves. Permanent redirects (301/308) are Google's signal the target should be canonical; temporary ones (302/303/307) aren't. through extra hops. Fix: compare before/after link paths, repair navigation and redirects, and keep the layer only if it improves real grouping.

Flattening created an unusable navigation wall

Symptom: the menu contains hundreds of choices and engagement suffers. Likely cause: depth reduction was implemented as sitewide links instead of better paths. Fix: restore scannable categories, expose the most important destinations, and use contextual links or related modules for the long tail.

The architecture is different for crawlers and users

Symptom: a rendered user can navigate to pages a raw crawlerA crawler — also called a spider or bot — is an automated program that fetches web pages, extracts their links, and queues new URLs to visit. Search engines use crawlers to discover and download content for their index. cannot discover. Likely cause: client-only controls, hidden states, or links without usable href values. Fix: use crawlable anchors for destinations and rerun the graph in the renderingTurning HTML, CSS, and JavaScript into the final visual page and DOM. mode appropriate to the production implementation.

Prompt: evaluate a proposed hierarchy

Evaluate this proposed site hierarchy without applying a universal click-depth rule.
For each template and priority group, identify the shortest expected path from the
homepage, the parent that gives the path meaning, and any layer that has too few,
too many, or overlapping children. Flag mega-menu flattening, single-child categories,
and priority pages that are less prominent than comparable pages. Recommend changes
using only the supplied inventory and business priorities; do not invent categories.

Inventory:
[PASTE URL | TEMPLATE | PROPOSED PARENT | PRIORITY | PRIMARY USER TASK]

Prompt: interpret a crawl-depth distribution

Analyze this crawl export by template and business priority. Compare depth distribution,
orphan status, and shortest-path source. Find cases where the sitewide average hides a
buried important group or where many low-value links make the graph artificially flat.
Return evidence rows, likely architecture cause, and the smallest link or hub change to
test. Treat URL slashes as descriptive data, not click depth.

Crawl export:
[PASTE URL | TEMPLATE | DEPTH | SHORTEST-PATH SOURCE | INDEXABILITY | PRIORITY]

Validate an architecture-depth change

Test yourself: Flat vs. Deep Architecture

Five quick questions on deciding, measuring, and fixing site depth. Pick an answer for each, then check.

Resources worth your time

My related writing

• When Should You Worry About Crawl Budget? — when depth actually starts to hurt crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor., and why most sites don’t need to obsess over it.
• Internal Links for SEO: An Actionable Guide — how hub pages and internal linksAn internal link is a hyperlink from one page on a website to another page on the same website. Internal links help search engines discover your pages and pass ranking signals (PageRank and anchor-text context) between them. shape the link graph that is your architecture.
• How to Structure Your Website Architecture for SEO — the broader structure walkthrough this depth framework fits inside.

My speaking

• How Search Works (SlideShare) — my walkthrough of crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor., renderingTurning HTML, CSS, and JavaScript into the final visual page and DOM., indexingStoring a crawled page in the search index so it can appear in results. Crawled is not the same as indexed — Google selects what to keep, and indexing isn't guaranteed., and ranking — the pipeline that click depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops. feeds into. (Standing disclaimer applies: this is my understanding of these systems, not an official spec.)

From around the industry

• Flat vs. Deep Website Hierarchies (Kathryn Whitenton, Nielsen Norman Group) — the rigorous UX-research take tying depth to category distinctness and findability.
• Why Google Recommends Hierarchical Site Structure For SEO (Search Engine Journal) — Illyes on why large sites need hierarchy and how directory grouping helps crawling.
• Crawl Stats report (Google Search ConsoleA free Google service that reports how a site performs in Google Search and surfaces problems with how Google crawls, indexes, and serves it. It's first-party data straight from Google — but you don't need it to appear in results. Help) — the Discovery vs. Refresh breakdown behind the depth-cliff diagnosis.
• How To Audit Crawl Depth & Improve Crawl Efficiency (Sitebulb) — a hands-on crawl-depth audit walkthrough.
• Structure Explorer (Ahrefs Academy) — using Site Audit to see your click-depth distribution.
• Site Architecture & Crawl Visualisations (Screaming Frog) — visualizing crawl depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops. and structure from a crawl.

Flat vs. Deep Architecture

Flat vs. deep architecture is the trade-off in how many hierarchy levels sit between a site's homepage and its individual pages — and, separately, how many clicks a crawler needs to reach them. Neither extreme is universally right: the correct depth is mostly a function of how distinct the site's categories are, and secondarily how much content it has.

Related: Site Architecture, Website Structure, Crawl depth, Internal Link

Flat vs. Deep Architecture

Flat vs. deep site architectureSite architecture is how a website's pages are organized, categorized, and interlinked. It controls how crawlers discover pages, how link equity flows, and how clearly search engines understand each page's topical context. Silo structure, hub and spoke, and topic clusters are the three common models. describes how many levels of hierarchy sit between a site’s homepage and its individual content pages — and, relatedly but not identically, how many clicks a crawlerA crawler — also called a spider or bot — is an automated program that fetches web pages, extracts their links, and queues new URLs to visit. Search engines use crawlers to discover and download content for their index. or visitor has to make to reach a given page. The useful definition is a site’s click-depth distribution and how many clicks its priority pages need, not a fixed layer count. A flat architecture keeps most pages — especially priority ones — within one or two levels of the homepage using few, broad categories. A deep architecture nests content through several layers of category → subcategory → sub-subcategory before it reaches an individual page.

Neither is universally correct. The right depth is mostly a function of how distinct (versus overlapping) the site’s natural categories, templates, and update patterns are, and secondarily of how many distinct items need their own findable page — page count alone is a rough signal, not the deciding factor. A 50-page brochure site can usually stay essentially flat; a catalog of tens of thousands of SKUs usually needs real hierarchy so the whole catalog isn’t dumped onto one navigation surface, but both are heuristics to calibrate, not verified cutoffs. Google has never published a numeric depth rule — its guidance is about the link graph, not a level count — while Bing recommends keeping important pages within roughly three clicks of the homepage as an operational target, not a guarantee.

Crucially, URL-path depth is not the same as click depthCrawl depth usually means click depth — how many clicks it takes to reach a page from the homepage by following internal links. It can also mean a crawler setting that limits how many levels deep a crawl goes before it stops.. A page whose URL looks three folders deep can still be one click from the homepage via a hub link, and a page with a short URL can be buried six clicks deep. What matters for crawlingCrawling is how search engines use automated bots (like Googlebot and Bingbot) to discover URLs and download pages. A page has to be crawlable to be indexed, but crawling on its own isn't a ranking factor. and signal-passing is the click depth in the link graph, not the number of slashes in the address. The practical work of getting depth right is measuring where a site currently sits (click-depth distribution from a crawlerA crawler — also called a spider or bot — is an automated program that fetches web pages, extracts their links, and queues new URLs to visit. Search engines use crawlers to discover and download content for their index., cross-referenced with what Google actually crawls in the GSC Crawl StatsA Google Search Console report (under Settings) that shows how Google has crawled your site over the last 90 days — total requests, download size, and average response time, broken down by response code, file type, Googlebot type, and purpose. It's only available for root-level properties (a Domain property or a URL-prefix property verified at the site's root). report) and then shortening the path to important buried pages with hub pages, related-content modules, and breadcrumbsBreadcrumbs are a secondary navigation trail (Home > Category > Page) that shows where a page sits in a site's hierarchy. They create internal links that pass PageRank, and when marked up with BreadcrumbList structured data they can drive the path Google shows in desktop search results..

Related: Site Architecture, Website Structure, Crawl depth, Internal Link

Changelog