What an internal link does
Search engines model a site as a graph of pages and links. Three documented or well-established effects follow from each link:
Discovery and recrawl. Crawlers find pages through links; pages with more links from important pages are found sooner and refetched more often. The site architecture guide covers the structural side; at the page level, a new page that no existing page links to may wait a long time to be indexed.
Context through anchor text. Google's guidance recommends descriptive anchor text — text that tells the reader and the engine what the linked page contains. Internal anchor text is a strong, controllable signal of what a page is about, which is why "click here" wastes it and why identical anchors pointing at different pages confuse it.
Authority distribution. The site's authority, earned largely through external links (BacklinkA link from another site to yours.), flows through internal links. A page linked from the homepage and from many relevant pages receives more than a page linked once from a footer. The model behind this is the family of link-analysis algorithms Google has described historically; the exact mechanics are not published, but the direction — more and better internal links to a page raise its standing — is consistent practitioner experience and the basis of hub-and-spoke structures.
Kinds of internal links
| Kind | Where | Job | Weight |
|---|---|---|---|
| Navigation | Header, menus | Discovery of sections; hierarchy | Followed; repeated on every page so individually light |
| Breadcrumb | Below the header | Path back up the hierarchy; BreadcrumbList schema | Followed; cheap, consistent |
| Hub lists | Hub and category pages | Listing every page in a section | Strong for discovery; the primary parent link |
| Contextual | Inside body text | Related concept at the point of relevance; anchor carries meaning | Strongest per link |
| Related blocks | End of page | Curated next reads | Useful when curated; noise when automated |
| Footer | Site-wide | Utility pages; a few pillars | Followed; light |
| Sidebar / widgets | Site-wide | Recent, popular | Often noise; rarely worth the clutter |
The pattern that works: hubs for breadth, contextual links for depth, curated related blocks for next steps, and navigation and footer kept to what every page needs.
Rules
- Every page links up, sideways and down. Up to its hub (breadcrumbsA trail of links showing a page's position in the site hierarchy, from the home page down to the current page. do this), sideways to the pages most related to it, down to its children if it is a hub. A page with only the "up" link is a dead end.
- Descriptive anchors. The target's topic in the anchor text, varied naturally: "the crawling and indexing guide", "how canonical tags work", "crawl budget on large sites". Never "here", "this page", "read more". Google's guidance and the site's own link validator both treat generic anchors as defects.
- Link at the point of relevance. A contextual link belongs where the concept comes up in the text, where a reader might want it. Lists of links at the bottom with no context are weaker in both senses.
- One primary target per concept. When several pages could receive a link for "keyword research", the anchor goes to the page that owns that query. Linking the same anchor to different pages is how cannibalisation is reinforced, and anchors that consistently name another page are a common reason Google ranks the wrong page for a keyword.
- Prefer links to the canonical URL. Linking to a redirecting or parameter URL sends signals through a detour (Canonical TagTells search engines the preferred URL.).
- Make links real links.
<a href="…">elements the crawler can follow; not click handlers, not buttons that navigate by script. The JavaScript SEO guide explains why. - Keep the count honest. Link what helps. A hub listing a hundred pages is a hub; a guide linking a hundred times is a page nobody can read and a signal spread thin.
- Do not nofollowA link attribute asking search engines not to treat the link as an editorial endorsement. internal links. Nothing is gained. Attributes belong on outbound links that are paid, user-generated or untrusted, as the outbound links guide explains.
- Fix what breaks. Internal links to removed pages, redirects and errors are found by every crawl and fixed at the source.
Finding under-linked pages
A site crawl lists inlinks per page. Three groups deserve attention:
- Pages that should rank with few inlinks. The priority pages from the keyword map that have fewer internal links than pages that matter less. These get links from the hub, from siblings and from contextual mentions across the site.
- Orphans and near-orphans. Pages with zero or one inlink. Either link them or remove them.
- Pages with many inlinks and no purpose. Tag pages, legacy pages, utility pages that accumulate links by template. Their weight could go to better targets.
Which of those generated pages should keep their links and stay indexable is a per-template decision, set out in the index rules for tags, archives and utility pages.
Search Console's links report shows the same picture from Google's side (internal links by target page), and the two views together find the gaps. Then the work is editorial: open the pages that discuss the under-linked page's topic and add the link where the topic comes up, with the right anchor.
Linking a new guide into a site
A site publishes a guide to hreflang. Before launch it has one inlink: the international SEO hub's list. The editor searches the site for every page that mentions international versions, language targeting or duplicate content across countries — the site migrations guide, the canonicalisation section of the crawling guide, the e-commerce category guide — and adds a contextual link in each, with anchors that describe the guide. The international hub's "start here" block pins it. The pillar's body mentions it under the international discipline. Six inlinks from relevant pages, with descriptive anchors, before the page is a day old; the crawler finds it through all of them.
Anchor text in detail
Anchor text is read as a description of the target. The practical guidance:
- Describe the target, not the action. "the budget allocation guide" rather than "click here to read about budgets".
- Vary naturally. The same page linked from ten places with ten slightly different descriptive anchors reads as natural; the same exact phrase ten times reads as manufactured. Internal anchors are not subject to the manipulation concerns of external ones, but they are still signals of meaning, and exact repetition carries less meaning than varied description.
- Match the target's topic, not the source's. The anchor is about where the link goes.
- Keep it short enough to read as a phrase. A full sentence as a link is hard to scan; three to eight words is typical.
The link quality and anchor text guide covers anchors on external links, where the rules are stricter.
Internal links and topic clusters
The topic clusters guide describes a content structure — a pillar page and the cluster of pages around it — whose whole mechanism is internal linking: every cluster page links to the pillar, the pillar links to every cluster page, and cluster pages link to each other where related. The structure tells the engine that the site has a body of work on a topic rather than scattered pages, and it concentrates authority on the pillar. This section of the site is built that way, and the link validator enforces the pattern.
Internal linking on large sites
Beyond a few hundred pages, contextual links cannot be placed by hand alone. The pattern is a layered system: structural links generated from the hierarchy (breadcrumbs, parent and child links, siblings); related-content modules generated from shared taxonomy, with rules that prefer the same section and the same intent; contextual links written by authors following the anchor-to-URL rules from the keyword map; and a periodic report of pages below the inbound-link minimum, which editors fix by hand. This site's content pipeline follows the pattern — every page declares related pages, hubs list their children, and a validator reports pages with too few inbound links before publication — and the same approach, with its rules written down, scales to any site whose content has structured metadata.
Measuring the effect
Internal linking changes are confirmed in three places: the crawl (inlink counts, depth, anchors), Search Console's links report (Google's view of internal links by page), and, with a lag, impressions and position for the target pages' queries. Because links are usually changed alongside other work, the cleanest measurement is to change linking alone for a set of pages and watch those pages against similar ones left unchanged. The SEO KPIs guide covers the tracking.
Internal links are one strand of a page's optimisation, so the on-page SEO audit is useful when you want to review linking alongside titles, headings and content on the same pages.
New pages launched with a single link from a blog index are a frequent cause covered in why new content is not ranking.
The internal linking test sets that comparison out as a plan: under-linked target pages split at random, links added to one group only, and source pages watched as a guardrail.
Common mistakes
- Generic anchors. "Click here" and "read more" waste the strongest controllable signal.
- Links only in navigation and footers. Discovery without context or weight.
- Automated related-posts blocks. Links by date or tag, unrelated to the content.
- Linking the same anchor to several pages. Reinforces cannibalisationTwo or more pages on one site competing for the same query and intent, splitting signals..
- Hundreds of links per page. Nobody reads them; each carries little.
- Links to redirects and dead pages. Found in every audit; fixed at the source.
- Nofollow on internal links. A misunderstanding of what nofollow is for.
The site architecture guide is the structural companion to this page; the on-page SEO expertise page lists internal linking among the author's core on-page practices.
