The short answer
Click depth is the smallest number of internal-link steps from the homepage to a target page. The guideline I have worked to on Estonian projects since 2017: priority commercial pages and main sections no more than 3 steps away, everything else as the catalogue logic allows.
It is not a Google requirement. The guide "Managing crawl budget for large sites" (developers.google.com, checked 21.09.2026) lists server response time, duplicates, soft 404s, redirect chains, sitemap freshness and HTTP 304 support — there is not a line about clicks from the homepage. The three-click rule comes from nineties usability writing, not from search engine documentation, and most articles ranking for this query present it as a Google rule without naming a source.
Click depth and URL depth are two different metrics
They get confused constantly, and that confusion is expensive: people rewrite page addresses instead of adding one menu entry, then spend another six months cleaning up after the 301 redirects.
| Metric | What it measures | How to change it |
|---|---|---|
| Click depth | Steps from the homepage along links | A menu link, a homepage block, an internal link |
| URL depth | Path segments after the domain | Only by changing the address plus a 301 |
| Depth in the sitemap | Nothing — a sitemap defines no hierarchy | Not applicable |
The product page for the Lambert 220 sofa at /shop/sofas/corner/lambert-220 sits at URL level 4 and can still be 2 clicks from the homepage if a bestsellers block links to it. The opposite case is more common: the contact page /contact lives at the root, but the only link to it is in the footer of product pages, putting it 7 steps away.
In a webmaster hangout on 9 June 2018, reported by Search Engine Journal the same day, John Mueller said Google leans on the number of clicks from the homepage more than on the shape of the address when judging how important a page is. Three years later, on 28 March 2021, he answered that flattening URLs for SEO has no benefit: showing depth in the address is fine. Both statements point the same way — fix links, not addresses.
How to measure depth without producing a junk report
One mistake makes the whole report meaningless, and it is very common: the crawler is started from a list of URLs taken out of sitemap.xml. Every sitemap URL then becomes a start page with depth 1, and the output is an evenly beautiful report where nothing is wrong.
- Exactly one start URL — the homepage of the locale in question. On a multilingual site that means three separate crawls, otherwise
/et/gets depth 2 and drags the whole Estonian branch down with it. - JavaScript rendering on, if the menu or the internal-link blocks are assembled by script. Without it the crawler sees an empty
<nav>and reports a depth the site does not have. - Links marked
nofollowand links inside non-indexable blocks are counted separately: the step exists, the weight does not travel along it. - Export the distribution by level, not the average: how many URLs sit at depth 1, 2, 3 and so on.
- Cross the list of pages at depth 4 or deeper with the list of priority pages — categories, services, articles you expect traffic from. The intersection of those two lists is the job.
A report saying "average depth 3.8" tells you nothing. A site where 90% of URLs are product pages at depth 5 while every category sits at depth 2 is healthy. A site with the same average but a category at level six is not.
Where depth actually comes from
On a 40-page service site depth is rarely a problem: everything is in the menu. It grows in catalogues, and almost always for one reason.
The arithmetic is simple. A category of 200 products showing 20 per page produces 10 listing pages. If they are joined only by next and previous links, a product on the tenth page sits 12 steps from the homepage. Replacing that pagination with a numbered block cuts the path to 4 steps straight away, because all ten pages are visible from the first one.
Next by contribution come filters, where every combination opens a new URL, and CMS archives by date, author and tag. In WordPress those archives are on by default and almost nobody needs them. The rule is short: a page you have no intention of promoting should not appear in the depth report at all — its place is outside the index.
There is one more source of depth that a click report will never show. A page with no incoming internal link has no depth at all: the crawler never reaches it and it stays out of the export. Those are orphan pages, and finding them means cross-checking the crawl against the sitemap and Search Console.
What to do with pages at depth 4 and deeper
The first decision about each such page is not "how do I lift it" but "is it needed at all". A page with no keyword cluster of its own that plays no part in sales does not get lifted: it gets removed from the index with a 410 or a noindex tag.
For the ones that stay, this order works for me:
- Check the menu. A priority section four steps away has usually just missed navigation. That is a half-hour fix and the cheapest of them all.
- Rebuild pagination. Numbered pages instead of a next chain, and for large categories a higher number of products per page. It works better than any link block.
- Add links from neighbouring pages. Related-products blocks, same-category blocks, links from blog articles. How to do this without turning it mechanical is in my piece on internal linking for SEO.
- Rebuild the structure itself, if depth comes down to a redundant subcategory level. This is the most expensive option and I take it on only when the structure has drifted away from the keyword map; what it should look like is in the article on site structure built from keyword research.
The order matters: cheap fixes on top, expensive ones at the bottom. In practice points 1 and 2 cover most cases and the fourth is rarely reached.
Where depth stops being harmless
On a 300-page site the difference between depth 3 and depth 6 is barely visible: Googlebot will cover the whole site in a few days either way. It starts to matter where crawling stops being complete, and that is exactly the scale Google names in its guide: from 10,000 pages with daily changes, or from a million with weekly ones. At those volumes depth turns into a question of crawl budget — pages on the lower levels simply never reach the queue.
I will state the limits honestly. I have no measurements of my own in the form of "we lifted the depth, rankings went up", and I have never seen an experiment where depth was changed in isolation from everything else. Moving a page from level six to level two means adding links to it, which changes click depth and the amount of internal weight at the same time. Those two effects cannot be separated in the field, so I treat figures like "depth-2 pages rank N positions higher" as correlation.
Depth shows that navigation has drifted away from what you sell. The drift is what gets treated, not the 3.8 sitting in the crawler report.
A separate note on new sites. With fewer than 20 sections, measuring depth is premature: sketch the structure on paper and check that every commercial section is visible from the menu. Screaming Frog is not what you need here, the wireframe is — and that is usually what I talk about at the web development stage, before the catalogue gets filled.
