Internal Linking: The Ranking Lever You Fully Control

What internal links actually do

Two jobs, and it helps to keep them separate because they fail in different ways.

Discovery. Crawlers find pages by following links. A sitemap helps Google learn a URL exists, but it is a hint, not a guarantee of crawling or indexing. A page that is genuinely reachable by following links from your homepage is far better positioned than one that only appears in an XML file.

Signal distribution. Links pass value between pages. When something links to your homepage, that value can flow onward through your internal links to the pages you actually want people to find. If your internal structure is a shallow archive listing, most of it stops there.

Google's link best practices documentation sets out the baseline requirements, and they are less demanding than most people assume:

• Links must be <a> elements with an href attribute to be crawlable. A <div> with a click handler is not a link, and JavaScript-inserted links are fine provided they produce proper markup.
• Anchor text should be descriptive, concise and relevant.
• Link to pages that help readers understand the current page.
• Every page you care about should have at least one internal link pointing at it.

That last line is the whole audit, and most sites fail it somewhere.

Find your orphan pages first

Before designing anything, find out what is currently unreachable. On a content site the answer is usually "more than you think," because articles get written and then never referenced again.

The manual version takes ten minutes and needs no tooling. Take your list of published URLs, then for each one search your own site for links pointing at it. Anything with zero inbound internal links is an orphan.

If you write in Markdown or MDX, one pass over your content directory will do it — count occurrences of each slug across all other files. Any slug appearing nowhere but its own file is orphaned. If you publish through a CMS, a crawler such as Screaming Frog will produce the same list by comparing crawled URLs against sitemap URLs.

Then fix them in this order:

1. Pages with zero inbound links — every one of these is pure upside.
2. Pages with exactly one inbound link, if that link is only the archive listing. Functionally still orphaned.
3. Your most important pages, regardless of current count. These should be the best-connected pages on the site, and frequently are not.

How to structure the links

The pattern that survives contact with a growing site is hub and spoke.

A hub is a page that introduces a subject and links out to everything specific about it — a category page, or a comprehensive overview article. A spoke is a page answering one narrow question in depth. Spokes link up to the hub, the hub links down to every spoke, and spokes link across to siblings where a reader would genuinely want the detour.

The result is that no page is more than a couple of clicks from any related page, and value arriving anywhere in the cluster can reach the rest of it. This is the concrete structure underneath what people call "topical authority" — the mechanism doing the work, rather than the label.

A few practical rules that hold up:

• Link forward and backward. A new article should link to relevant older ones, and you should go back and add a link from the relevant older ones to it. The second half is the step everyone skips, and it is the half that gets the new page crawled.
• Link from strong pages to weak ones. Whichever pages draw the most impressions have the most to pass along. Check your Search Console Pages report, then make sure your best performers link to the pages you want to lift.
• Keep important pages shallow. If reaching an article takes five clicks from the homepage, that is a structural statement about how much you value it.
• Link where the reader is already thinking about it. A link inside the sentence that raises the topic is worth more, to both readers and crawlers, than the same link in a list at the bottom.

Anchor text: the part people waste

The clickable text is a description of the destination. That is information you are giving away for free every time you write "read more."

Instead of	Write
Click here	How Core Web Vitals are measured
Read more	Why CrUX and Lighthouse disagree
This article	Google's spam policy on scaled content
https://example.com/blog/x	The name of the thing at that URL

Two refinements worth applying:

Vary the phrasing. Linking to the same page from twelve places with the identical anchor text reads as mechanical. Natural variation — sometimes the title, sometimes a description of the specific point being referenced — is both better writing and a richer signal.

Describe the destination, not the relationship. "See our guide" tells a reader nothing about whether they want to click. "See how the 28-day CrUX window delays your results" tells them exactly.

Common mistakes

Boilerplate related-post blocks as a substitute for real links. An automated "you might also like" row is fine as an addition. It is not a cluster. Every page carries the same block, so it says nothing specific about how two pages relate — the information content is close to zero.

Linking everything to everything. If a 1,500-word article contains thirty internal links, none of them means much, and the reading experience suffers. Three to eight genuinely useful links in a long piece is a healthy range.

Nofollowing internal links to "sculpt" flow. This has not worked the way people imagine for well over a decade and mostly just discards signal. Use nofollow for its actual purposes, and use rel="sponsored" on paid or sponsored links — which is a policy requirement, not an optimisation.

Forgetting the retrofit. Adding links from new pages to old ones is easy because you are already in the file. Adding links from old pages to new ones requires going back. Budget for it — every time you publish, spend ten minutes adding two or three links from existing articles to the new one.

Breaking links when slugs change. If you rename a URL, redirect the old one and update the internal links that pointed at it. Internal links that resolve through a redirect chain still work, but they are lossy and they accumulate.

How to know it worked

Do not expect a same-week result. Internal linking changes take effect as pages are re-crawled, which runs on a scale of weeks.

What to watch, roughly in the order it moves:

1. Coverage in Search Console. Pages moving out of "Discovered — currently not indexed" into indexed is the first sign that discovery improved.
2. Impressions on previously orphaned pages. A page nobody linked to often has almost no impression history; any sustained non-zero trend is new.
3. Average position on the cluster, read per-page rather than site-wide. The aggregate hides too much — see reading position data for why.

The honest expectation: internal linking rarely moves a page from position 70 to position 5 on its own. What it reliably does is stop pages being invisible, spread signals you already have to where they are useful, and make the rest of your work count for more. On a site with orphan pages, it is the cheapest improvement available — and unlike almost everything else in search, no one else has a vote.

Sources: Google Search Central — Link best practices · Google Search Central — Sitemaps overview · Google Search Central — Qualify outbound links