Quick Answer
Internal linking builds topical authority by telling Google which of your pages is the authoritative one on a subject and which pages support it. A site with 200 unconnected posts reads as 200 separate opinions. The same 200 posts organised into clusters, each pointing at one money page and at each other, read as coverage of a subject. The architecture is the difference, not the word count.
Written by Cody Wise, founder of Wise Media. This article describes the exact internal link architecture Wise Media runs on client sites and on this one. Every claim about Google’s behaviour below is sourced to Google’s own documentation rather than to agency commentary.
Summary
- Google says every page you care about should have a link from at least one other page on your site. That is the minimum bar, and a surprising number of sites fail it.
- Topical authority is earned per cluster, not per post. One pillar page plus the supporting posts that answer every question a buyer has, all linked.
- Anchor text is the signal. Google says good anchor text is descriptive, reasonably concise and relevant, and that keyword stuffing in anchors violates its spam policies.
- Orphan pages and deep pages are the two most common structural faults, and both are findable in an afternoon.
- More links is not better. Google states there is no magical ideal number, and chaining links together strips the context that gives them meaning.
Table of Contents
- What Google actually says about internal links
- The pillar and cluster architecture
- Anchor text: the part most sites get wrong
- Link depth and orphan pages
- When internal linking makes things worse
- Nofollow, and why it is almost never the answer
- How to audit your internal link graph
- Common mistakes
- FAQ
What Google Actually Says About Internal Links
Most internal linking advice circulating online is inference dressed as fact. Google publishes a surprising amount on this directly, so start there. Four statements from Google’s documentation do most of the work.
- Links are how pages are found. In its guidance on crawlable links, Google states that every page you care about should have a link from at least one other page on your site, and describes this as how Google discovers pages.
- Only real anchors count. Google can generally crawl a link only when it is an HTML anchor element with an href attribute. Span elements with href, framework router attributes and onclick-only navigation are listed as not recommended.
- There is no magic number. Google states there is no ideal number of links a given page should contain.
- Discovery through linking reduces the need for other crutches. In the sitemaps overview, Google says that if your site’s pages are properly linked, it can usually discover most of your site, and defines properly linked as every page you deem important being reachable through some form of navigation.
That last one is worth sitting with. A sitemap is a list. Internal links are a structure. Google states plainly that a sitemap does not guarantee that the items in it will be crawled and indexed. Submitting a URL is not the same as giving it a reason to exist.
Everything else in this article is the practical architecture built on top of those four facts. If you want it designed and implemented rather than explained, that is the work inside our SEO and AI search service.
The Pillar and Cluster Architecture
A cluster is one page you want to rank commercially, plus the supporting pages that answer every question a buyer asks before they are ready to contact you, all linked to that page and to each other.
The reason this works is not mystical. A buyer researching a purchase asks a predictable sequence of questions. If your site answers all of them and the answers are visibly connected, a search engine has strong evidence that this site covers the subject. If the same answers exist on the site but nothing connects them, each page competes alone against sites whose pages have support.
What belongs in a cluster
| Layer | Page type | What it has to do |
|---|---|---|
| Pillar | The commercial page: a service page or a definitive guide | Convert, and receive links from every supporting post |
| Cost and comparison | What it costs, option A versus option B | Catch the buyer who has decided to buy and is choosing how |
| Process and timeline | How it works, how long it takes | Remove the uncertainty that stalls an enquiry |
| Objection and risk | What goes wrong, legal or compliance angles, common mistakes | Answer the fear that stops a form submission |
| Definition and basics | What is X, do I even need X | Catch the top of funnel and funnel it inward |
Five to eight supporting pages per cluster is usually enough to be credible. Twenty-eight posts loosely about the same service, written one at a time with no structure, is not a cluster. It is a cannibalisation problem that happens to be large.
The three link rule
Every supporting page in a cluster carries three kinds of internal link. This is the rule we run on every site we manage, and it is deliberately small enough to actually get followed.
- One link to the cluster’s pillar page, placed in the first third of the article, in a contextual sentence rather than a banner. Early placement matters because a reader who leaves at 30 percent never sees a link in the conclusion.
- At least two links to sibling pages in the same cluster, placed wherever the article naturally defers to the sibling. The test is whether a reader would want that page next.
- At least one link to a page in a different cluster, which is what stops your site becoming a set of isolated silos with no connective tissue.
That is three to five links minimum per page, all of them earning their place. Compare that to the usual approach, which is a related posts widget at the bottom generating links nobody chose and nobody clicks.
A worked example
Take a hypothetical Alberta plumbing company with a commercial page for emergency plumbing. The cluster that should surround it:
- Pillar: the emergency plumbing service page.
- What does an emergency plumber cost in Calgary, and what drives the price.
- How fast should a plumber actually arrive, and what a response window means.
- What to do in the first ten minutes of a burst pipe.
- Licensing and insurance: how to check a plumber is qualified in Alberta.
- Emergency call-out versus scheduled repair: which you actually need.
Each of those five posts links to the service page early, to two of the others where the text naturally defers, and out to something in a different cluster, perhaps a post about seasonal maintenance. That is a cluster. It took six pages, not sixty.
Anchor Text: The Part Most Sites Get Wrong
The link tells Google two pages are related. The anchor text tells it what the destination is about. Get the anchor wrong and you have spent the link and sent no signal.
Google’s published rules are short and specific. Good anchor text is descriptive, reasonably concise and relevant to both the linking page and the linked page. Generic phrases like click here and read more are called out explicitly. Overly long anchor text is discouraged. And Google states that keyword stuffing is a violation of its spam policies, which applies to anchors as much as body copy.
The out of context test
Google offers a test worth adopting as a standing rule: read the anchor text alone, with no surrounding sentence. If it does not make sense on its own, make it more specific.
| Anchor | Verdict | Why |
|---|---|---|
| click here | Fail | Named by Google as a phrase to avoid. Carries no information |
| this article | Fail | Meaningless out of context |
| emergency plumber cost Calgary emergency plumbing prices Alberta | Fail | Stuffed. Reads as manipulation, not description |
| what an emergency plumber costs in Calgary | Pass | Descriptive, concise, makes sense alone |
| our emergency plumbing service | Pass | Clear destination, natural in a sentence |
Vary anchors without engineering them
Twelve pages linking to one destination with the identical exact-match anchor is a pattern, and patterns produced by a spreadsheet look like what they are. The fix is not a rotation schedule. It is to write each link inside its own sentence so the anchor naturally reflects the angle that paragraph is taking.
One more Google rule that almost nobody follows: surrounding words give context, so do not chain multiple links together. Space them out. A paragraph containing four adjacent links passes no context to any of them. For image links, Google uses the image’s alt text as the anchor, which is one more reason empty alt attributes are a mistake.
Link Depth and Orphan Pages
Two structural faults account for most of what internal link audits find. Both are mechanical, and both are fixable without writing anything new.
Orphan pages
An orphan is a page with no internal links pointing to it. It usually exists because it was published, shared once and never referenced again, or because it was built for a campaign and left behind.
Google’s bar is explicit: every page you care about should have a link from at least one other page on your site. One link is the floor, not the target. A page in a cluster should have several.
Finding orphans takes a crawl plus your sitemap. Crawl the site following only internal links, export the URL list, then compare it against the full list of published URLs from your CMS or sitemap. Anything in the second list and not the first is an orphan. For each one, decide honestly: link it into a cluster, merge it into a stronger page, or remove it. Keeping a page nobody links to because deleting feels wasteful is how sites accumulate dead weight.
Click depth, and what actually matters about it
Click depth is the number of clicks from the homepage to a page. The familiar three click rule is a usability heuristic, not a Google requirement, and it is often quoted as though it were the latter. Google publishes no depth threshold.
What Google does publish, in its crawl budget guidance, is more useful. Crawl demand depends on perceived inventory, popularity and staleness, and Google notes that duplicates and unwanted URLs waste crawling. The guidance also states directly that crawl budget is an advanced concern, relevant mainly to sites with over a million unique pages changing weekly, or over ten thousand changing daily, or sites with many URLs sitting in the Discovered but currently not indexed state in Search Console.
For a typical Canadian business site with a few hundred pages, crawl budget is not your problem and you should stop reading advice that says it is. That third condition, though, is worth checking. A pile of pages stuck in Discovered but currently not indexed is often a linking problem wearing a crawling costume: Google found the URL in a sitemap, saw nothing on the site treating it as important, and deprioritised it.
When Internal Linking Makes Things Worse
Linking does not fix a structural problem. It can entrench one.
If two pages target the same query, linking them to each other with similar anchors reinforces the ambiguity rather than resolving it. You are telling Google both pages are about the same thing, which is precisely the state you need to exit. The fix is to merge one into the other and redirect, then point the links at the survivor. We cover the detection and repair of that in full in our guide to finding and fixing keyword cannibalization.
Three other ways internal linking backfires, worth watching for:
- Linking into redirect chains. Internal links should point at the final URL. Google’s crawl guidance lists avoiding long redirect chains as a best practice, and internal links are the easiest place to get this right since you control both ends.
- Bulk automated linking. Plugins that insert a link every time a keyword appears produce chained links with no surrounding context, which is the exact thing Google’s anchor guidance warns against.
- Linking to thin pages to prop them up. If a page is not worth a reader’s time, pointing more links at it does not make it worth Google’s. Merge it or cut it.
Nofollow, and Why It Is Almost Never the Answer
Applying nofollow to internal links to sculpt where authority flows is a tactic that stopped working a long time ago and never worked the way people believed. Google’s current link documentation describes the attributes by purpose: nofollow for links you do not vouch for, sponsored for paid links, and ugc for user-generated content. None of those describe your own navigation.
If you do not want a page crawled or treated as important, the honest answers are structural: do not link to it prominently, noindex it if it genuinely should not appear in search, or remove it. Google’s crawl guidance is explicit that noindex is the wrong tool for saving crawl budget specifically, because Google still requests those pages, and that robots.txt is the tool for blocking unimportant URLs. Pick the tool that matches the intent rather than reaching for nofollow by habit.
How to Audit Your Internal Link Graph
This takes an afternoon on a few hundred pages. Work it in order.
- Crawl the site following internal links only. Export every URL found, with its inbound internal link count and its click depth from the homepage.
- Export every published URL from your CMS or sitemap, independently of the crawl.
- Diff the two lists. Anything published but not found by the crawl is an orphan. Write them down.
- Sort by inbound link count ascending. Every page with zero or one inbound link is a candidate for attention.
- Map your clusters on paper. For each commercial page, list the supporting pages that should surround it. Do this before looking at what exists.
- Mark every page that belongs to no cluster. These are the merge or remove candidates.
- Check anchor text on links to each pillar page. Flag any that are generic or identical across many sources.
- Check Search Console’s page indexing report for pages in Discovered but currently not indexed, and cross-reference them against your orphan list. The overlap is usually large.
- Fix redirect chains in internal links so every link points at the final URL.
- Implement the three link rule on the highest-value cluster first, then work down. Do not try to fix the whole site in one pass.
Re-crawl after the first cluster is done and confirm the numbers moved before repeating. Internal linking is one of the few SEO jobs where you can verify the work mechanically rather than waiting on rankings.
Where Internal Linking Sits Against Other Levers
| Lever | Control | Speed | When it is the right next move |
|---|---|---|---|
| Internal linking | Total. You own both ends | Fast to implement, weeks to read | You have content and no structure. Most sites |
| New content | Total | Slow | A cluster has real gaps a buyer asks about |
| Merging duplicates | Total | Fast | Two pages target one query |
| Earned links and mentions | None. Someone else decides | Slow and unpredictable | The cluster is built and you need external validation |
The order matters. Chasing external links into a site with no internal structure sends authority to pages that cannot pass it anywhere useful. We argued that case at length in our comparison of link building versus topical authority, and the internal architecture in this article is the half of that argument you can act on today without anyone else’s permission.
Common Mistakes
- Treating a related posts widget as internal linking. Machine-chosen links at the bottom of a page carry no editorial signal and little context.
- Linking only from new posts to old ones. The link graph has to be maintained in both directions. When you publish a new post, go back and link to it from the two most relevant existing pages.
- Putting every link in the conclusion. Most readers never arrive there.
- Using the same exact-match anchor everywhere. Descriptive and varied beats optimised and identical.
- Building silos with no cross-links. Clusters that never reference each other read as separate small sites.
- Linking through JavaScript that produces no real anchor element. Google documents that it generally crawls a link only when it is an anchor element with an href.
- Measuring links instead of outcomes. The number to watch is pages indexed and ranking per cluster, not links added.
One cross-cluster note worth making, since this article is itself an example of the rule. Structure is not only an organic concern. The same discipline of one page, one job, one clear next step is what separates a paid landing page that converts from one that does not, which we worked through in detail in the Google Ads landing page diagnostic.
FAQ
How many internal links should a page have?
Google states there is no magical ideal number. A practical floor for a cluster page is three to five editorial links: one to the pillar page, two to siblings, one cross-cluster. Beyond that, add a link only when a reader would genuinely want that page next.
Do internal links help topical authority?
They are the mechanism that makes a set of related pages legible as coverage of a subject rather than as unrelated articles. Google uses links to discover pages and anchor text to understand what they are about. Without internal links, each page argues its case alone.
What is an orphan page and does it matter?
An orphan page has no internal links pointing to it. It matters because Google’s own guidance says every page you care about should have a link from at least one other page on your site, and because orphans frequently show up in Search Console as discovered but not indexed.
Is the three click rule a real Google requirement?
No. It is a usability heuristic. Google publishes no click depth threshold. What matters is that important pages are reachable through navigation and are linked from pages that are themselves important, not the raw click count.
Should I use nofollow on internal links?
Almost never. Google describes nofollow as being for links you do not vouch for. If a page should not be prominent, link to it less, noindex it, or remove it. For crawl control specifically, Google’s guidance points at robots.txt rather than noindex or nofollow.
Does a sitemap replace internal linking?
No. Google states a sitemap does not guarantee that its items will be crawled and indexed, and says that if a site’s pages are properly linked it can usually discover most of the site. A sitemap lists URLs. Internal links tell Google which of them matter.
How long does internal linking take to show results?
The implementation is immediate. The effect depends on recrawling, which varies with how often your site is crawled. Expect to read results over weeks rather than days, and verify the mechanical work with a recrawl rather than waiting on rankings to tell you whether it was done.
The Short Version
Most Canadian business sites do not have a content problem. They have an architecture problem: real answers to real buyer questions, published one at a time, connected to nothing. Internal linking is the cheapest lever available because the content already exists and you control both ends of every link.
Pick your highest-value commercial page. Map the five to eight questions a buyer asks before contacting you. Link what exists, write what does not, and apply the three link rule. Then do the next cluster.
Want the Architecture Built for You?
Wise Media audits internal link graphs, maps clusters against commercial intent, merges what is cannibalising and builds the structure that makes existing content rank. It is part of our website growth packages, alongside the content that fills the gaps the audit finds.
Tell us what you have published and what you want ranking: start with the intake form.
