Internal linking: the structure nobody plans

People talk a lot about external links — who links to me, how do I get more of them. Almost nobody talks about internal links. Yet they are the only part of your linking that you fully control yourself.

They decide which of your pages get found at all, how quickly, and which of them count as important.

What internal links do

Three things at once:

They distribute weight. Whatever strength a page has built up, it passes on through its links. A page that nothing internal points to gets nothing — no matter how good it is.

They guide the search engine. A page that no other page links to is hard to find. It can be in the sitemap and still be fetched rarely.

They guide the visitor. This is the part that affects enquiries most directly, and the one most often forgotten when optimising.

The three-click rule

The core statement on structure: Important content should be reachable from the home page in at most three clicks.

What happens if it sits deeper: it is fetched less often and therefore updated more slowly. And for visitors it practically does not exist — nobody clicks through five levels.

Too flat is wrong as well, though. If everything sits on one level, there is no hierarchy, and search engines cannot recognise the relationships between the pieces of content. It is then not clear which page carries the main topic and which one works out a detail.

The right structure lies in between: a clear outline with few levels.

Topic areas instead of a jumble

The organising idea for this is grouping by topic: content that belongs together is gathered into one area, and the pages support each other.

For a website like mine that means: everything on loading time belongs together. Everything on accessibility belongs together. Everything on relaunches belongs together.

What that achieves:

• Search engines recognise what the website is focused on.
• Pages within one area strengthen each other.
• The visitor following a topic finds where it continues.

In WordPress, specifically:

• Categories as topic areas — each category is one clearly bounded topic.
• One category per post. Not three. A post in five categories belongs to none of them properly.
• Tags connect across category boundaries — they are not a second category system.
• Give category pages their own text, do not leave them as an automatic list.

The mistake I see most often: dozens of tags, each with one or two posts, existing as pages of their own and containing nothing but one or two preview texts. That creates many thin pages and organises nothing.

(Atlas, seo/site-architecture.md)

How many links, and where

As a guideline: about ten internal links to an important page. That is no magic number but an order of magnitude — a page with one incoming link is undersupplied, a page with a hundred buys itself nothing.

More important than the number is where they come from:

The links have to come from suitable pages. A link from a thematically related page carries weight. A link from some random context does not.

Placed naturally in the text. A link in the middle of running text, at the point where a reader would actually want to read on, is worth more than a “You might also be interested in” box at the end of the page.

The first link counts

A detail that is hardly known and changes the practice:

When a page links to the same target several times, only the first link is evaluated. All further ones do not count — and above all: only the link text of the first link is taken into account.

What follows from that:

• The most important link to a target and its most meaningful link text belong early in the content.
• If a navigation link at the top of the page already points to the same target, the later link in the text is already the second one — and its text is no longer evaluated.

That is one more reason to use meaningful labels in the navigation. On every page they are the first link to their target.

Vary the link texts

With several links to the same page from different pages: Use different wording.

The same link text on every link looks manufactured. And the link text is a signal for what the target page is about — different wordings cover different search wordings.

Example for the same target page: “understanding loading time”, “why your site loads slowly”, “what the Core Web Vitals measure”.

That only applies to links from the running text of different pages, though. In the navigation the opposite applies: there the same page must have the same name everywhere.

(Atlas, seo/site-architecture.md)

The quality rules

Always link to the canonical address. Not to a variant, not to one that redirects.

The number that goes with it is sobering: in analyses, 57 percent of the websites examined link internally to non-canonical addresses. That is one of the most common mistakes of all — and one of the most unnecessary, because it happens when the link is set and then stays forever.

Typical causes: the address was copied with or without the trailing slash. The page has moved since. The link was taken over from an old version.

Fix broken internal links at once. They cost crawl budget and lead visitors into nothing.

Avoid redirect chains. Every hop loses something. Link directly to the final target.

Images as links carry their alternative text as link text. That means: if an image is a link, its alternative text has to describe where the link goes — not what is shown in the image. It is the same point that also applies to accessibility: one change, two effects.

Orphan pages

A page without incoming internal links is an orphan. It can be in the sitemap, it can be online, it can be well written — it is still hardly found.

How they come about:

• An old campaign page whose links were removed.
• A blog post that only the archive page lists — and the archive page is set to noindex.
• A page that dropped out after the menu was rebuilt.
• A page that can only be reached through the internal search.

The decision is always the same: Should the page be found? Then link to it from a suitable page. If not, it goes — or at least out of the sitemap.

How to check this yourself

The three-click test. Take your five most important pages. Start at the home page and count the clicks. More than three is a finding.

The orphan test by hand. Take an important subpage and ask: which other pages link to it? If none comes to mind, that is the answer.

Search Console, the internal links report. It shows which pages have how many internal links — sorted. At the very bottom are your undersupplied pages. That is the quickest overview you get without an extra tool.

The link text pass. Go through a page and read only the link texts. Do they say where they lead?

The category count. Look at how many categories and tags you have and how many posts are in each. Categories with one post and tags with two are not a system of order.

For the complete overview you need a crawling tool that outputs which page is linked from which, and with what link text.

What to do

1. Check the click depth of your most important pages. If a service page is four clicks away, it belongs in the menu or on the home page.

2. Tidy up the category system. One post, one category. Tags only where they actually connect.

3. Link new content from existing content. This is the step that is almost always forgotten when publishing: a new post appears, and nobody goes through the three older posts that ought to point to it. That belongs in the publishing routine.

4. Correct links to redirecting addresses.

5. Set links in the text, not only in boxes at the end.

6. And add the internal link pass to the regular check. As content grows, the structure changes, and what was well linked two years ago is hanging in the air today.

What I tell clients about this: Internal linking is the area with the best ratio of effort to effect, because it depends neither on Google nor on third parties. You alone decide which of your pages get weight. It is just that nobody does it.

Sources

• Atlas, seo/site-architecture.md — grouping by topic areas, with the example of an agency website and pages within one area strengthening each other; the structural rule of neither too flat nor too deep, with the three-click rule and the consequence of less frequent fetching at greater depth; the WordPress implementation (categories as areas, one category per post, tags as cross-connections rather than a second category system, category pages with their own content); the three purposes of internal linking; the guideline of about ten internal links to an important page and the requirement on origin and natural placement; the priority of the first link, with only the first link text evaluated and the consequence for navigation links; varying link texts with examples; the quality rules (always link to canonical addresses, 57 percent of websites link internally to non-canonical addresses, fix broken internal links at once, avoid redirect chains); images as internal links with the alternative text as the signal for the link target and the double benefit for accessibility; and the recommendation of regular link audits with the note to link new content from existing content.
• Udemy, Technischer SEO-Audit mit Screaming Frog (May 2025), section 3 — orphan addresses as a check of their own, with the decision rule to either link them internally or take them out of the sitemap, and correcting internal links that point to redirecting addresses.
• Own audit practice — how orphan pages come about, the three-click test and the orphan test by hand, the internal links report in Search Console as the quickest overview without an extra tool, counting categories and tags as a sign of a system of order that does not hold, and the note that linking new content from existing content belongs in the publishing routine.