Transcript
Intro
Someone matched 23 million internal links across 1,800 websites against Google clicks. Pages with zero to four internal links pointing at them averaged about 2 clicks. Pages with 40 to 44 links averaged four times that. And then, somewhere around 45 to 50 links, the line turned around and traffic started falling.
So more links help, until they don't. Today we find out why, which internal linking rules are real and which are left over from 2009, and then we audit a site and find its most important page hiding six clicks deep.
Why call internal links the lever you fully control? Backlinks have to be earned, bought or begged for, and each one takes weeks. Internal links you can change this afternoon, on every page, with nobody's permission. Which also means every mistake in them is yours, and in our audits they are some of the most common problems we find.
One caveat on that chart, and it's the only one this episode gets. It is a correlation across many sites, so pages with more links may just be the pages their owners cared about most. We treat it as a map of where returns flatten, and we back the direction with controlled tests later on.
What a link does
So what does an internal link actually do? Google says it uses links as a signal when determining the relevancy of pages, and to find new pages to crawl. Two jobs, then. A link passes some importance from one page to another, and its anchor text tells Google what the target is about.
Not every link passes the same amount. A Google patent from 2010 describes what people call the reasonable surfer: links weighted by how likely a person is to click them, judged from where the link sits, how big it is and what it says. A patent shows an idea Google had, not proof it runs today. It does line up with what the tests show.
The 2024 leak adds one detail. Among the quality signals Google stores, there is a field called anchor mismatch demotion. The leak shows what Google stores, not how much it weighs or whether it's live. The name alone suggests that when an anchor and its target disagree, something gets demoted. One more reason to keep your anchors honest.
Real links only
Rule one is boring, and it breaks more sites than any other. Google says it can generally only crawl a link if it is an a element with an href pointing at a real URL. A span with a click handler, a button that swaps content with JavaScript, a router link with no href: all of them look like links to a visitor and may not be links to Google.
ReelRadar, our fictional film site, had exactly this. Each genre page listed twenty films and then a button labelled More films. Google's crawlers don't click buttons. So every film past the first screen had no crawlable link from its own genre page. The fix was dull and it worked: numbered pages with real links. Page two, page three, and so on.
How do you check your own site? Run an important page through Search Console's URL Inspection tool and look at the rendered HTML Google saw. Search it for the link you expect, inside an a element with an href. Then crawl the site twice, once with JavaScript rendering off and once on, and compare the URL counts. A big gap means your links depend on scripts.
Anchor text
Rule two: anchors should describe the page they point to. Google's own bad examples are click here, read more, and the word website. Its test is simple. Read only the anchor, out of context, and ask whether you know what the next page is about. And remember the words around the link count too.
An easy one to miss: image links. When an image is the link, Google uses the image's alt text as the anchor. So a row of film posters with empty alt attributes is a row of links that say nothing at all. Give each poster the film's name, and the links start describing their targets again.
Variety matters as well. In that same big study, the number of different anchors pointing at a page was strongly correlated with Google traffic, so strongly the authors ran the numbers three times. Link a film page as Harbour of Glass in one place, the Harbour of Glass release date in another, and Rowan Ellery's lighthouse thriller in a third, and you've described it from three sides.
Then a twist arrived in February 2026. The same researchers added a note: new data suggests some sites with very high anchor diversity may suffer. So don't build a spreadsheet that rotates synonyms. Variety should come from writing each link naturally for the sentence it sits in. Verdict on maximising variety: contested.
Which brings us to the first myth. Exact match internal anchors, where the link text is the exact query a page targets, are risky and can get you penalised.
The fear comes from backlinks, where bulk exact match anchors are a classic spam pattern. Inside your own site, linking to your Harbour of Glass page with the words Harbour of Glass is just accurate. In the big study, pages with at least one exact match internal anchor had at least five times the traffic. Google warns against stuffing, not against accuracy. Verdict: placebo.
Menus vs in-content links
Second myth. More internal links is always better, so link every page to every other page you can.
Look at that opening chart again. The authors traced the downturn past 50 links to menus and sitewide links: pages linked from every menu on the site tended to have lower traffic. Google's own line is that there's no magical ideal number of links, and if you think it's too much, it probably is. Verdict: placebo.
That doesn't make menu links worthless. In one controlled test, adding more internal links to a homepage footer lifted the destination pages by 5 percent. In another, a site with about 8,000 regional pages linked each one to its nearest neighbours, and the pages receiving those links gained 7 percent. Both are links with an obvious reason to exist.
So here is how we split the work. Menus and footers carry the handful of hub pages every visitor needs. Links inside the main text carry the specific connections, like a film page pointing to its director's other films, with a sentence saying why. That is where the anchor and the words around it do the describing.
Here's how that looks on a ReelRadar film page. Below the synopsis, three sentences link out: the director's first film, the novel it's adapted from, and this week's streaming list. Each link sits in a sentence that explains it. Now picture the usual widget, twelve thumbnails called You may also like. One reads like a guide. The other reads like a vending machine.
Google adds one practical rule for in-content links: don't chain several links next to each other. Each link needs its own words around it, so readers can tell them apart and the surrounding text can describe the target. A sentence with five links in a row describes none of them. It mostly describes a hurry.
Nofollow sculpting
Third myth, and the oldest one in the building. Put nofollow on internal links to pages you don't care about, like the login page, and the saved authority flows to the pages you do care about. People call it PageRank sculpting.
Google's head of webspam at the time explained in June 2009 that it had stopped working. Authority sent to a nofollowed link no longer flows on to the other links on the page. His advice was to let PageRank flow freely within your site. And since 2019, nofollow is only a hint anyway. Verdict: placebo, for seventeen years and counting.
Use nofollow for what Google documents it for: links you don't vouch for, paid links and links your users add. For your own pages, the way to show what matters is the same advice Google gave back in 2009. Link to your important pages from prominent places, and don't bury them.
The audit
Now the fun part, the audit. We crawled ReelRadar, about 1,200 pages, and lined up three things for every URL: how many internal links point to it, how many clicks it sits from the home page, and its clicks in Search Console. Everything in this section is example data from our made-up site. The steps are the ones we run on real ones.
First finding: orphans. The sitemap listed 46 film pages the crawl never reached, because nothing linked to them. They existed only in the sitemap and in site search. Google says every page you care about should have a link from at least one other page on your site. These had none. Forty six pages, waiting for a visitor who could never find the door.
Each orphan needs a decision before it needs a link. If the page has value, link it from its genre page and from related films. If it duplicates another page, merge them and redirect. If it has no value at all, let it go with a 404 or 410. What you shouldn't do is leave it in the sitemap and hope.
Next, join links with clicks. Plot internal links against impressions for every page, and the interesting dots sit in one corner: pages that get seen in search but have almost no internal links. On ReelRadar, most of them were new release film pages, linked only from a single date list.
Then the finding that hurt. ReelRadar's where to stream this week page is its money page: it updates every Friday, and it's the page the site wants people to bookmark. The crawler found it six clicks from home: home, films, browse, by service, archive, this week. One internal link pointed at it. One.
We added a link from the home page, one in the main menu under Streaming, and a line on every film page with a streaming date, pointing to this week's list. The page went from depth six with one inlink to depth one, linked from the strongest pages on the site. Its anchors now say what it is: what's streaming this week.
We are not showing you a traffic chart for this fix, because ReelRadar doesn't exist and we don't invent numbers. On real sites, this is a change we read in Search Console over the following weeks, against the same weeks before. When you run it yourself, measure the target page, and also the pages you took links from.
Why watch the donor pages? One controlled test added related article links, and the pages giving the links improved, while the pages receiving them showed nothing conclusive. Another cut the number of links in a link block, and the pages still linked ranked better. Links aren't free. Moving them changes both ends.
So our rule for money pages is about distance and company. Keep them within a couple of clicks of the home page. Link to them from the pages that already earn links and traffic, inside the content where the link makes sense. And give each one a small set of descriptive anchors you use consistently.
That last point is what the anchor check in our free script does. It lists every anchor pointing to each important page, flags the empty ones and the click heres, and shows you where one page gets described five different ways while another gets no description at all.
The final flag matters more than it sounds. If the words Harbour of Glass trailer link to the film page in one place and the trailer page in another, you are telling Google two different pages answer the same question. Pick one target per phrase and stick to it.
If you do one thing this week, run a crawl and sort by two columns: click depth, deepest first, and inlinks, fewest first. Any page you would be upset to lose that shows up near the top of either list gets new links from your strongest pages, in the content, with honest anchors.
Wrap up
Scorecard. More internal links always being better: placebo, returns flatten. Exact match internal anchors being risky: placebo, accuracy isn't stuffing. Nofollow sculpting: placebo since 2009. Maximum anchor variety: contested since that February note. What works is real links, descriptive anchors, and important pages kept close to strong ones.
The internal link audit script and the anchor check are free on the episode page at irongoo.com. Run them this week, find your orphans and your deepest important page, and fix those two first. Your money page will thank you, quietly, in Search Console.
