Most SEO audits treat internal linking as a checkbox. Fix the broken links, add a sitemap, move on to the thing that "actually matters" — backlinks, content volume, technical crawl budget.
That ordering is backwards.
I've been going through a fair number of site structures lately — some client work, some just sites I'm curious about — and the pattern that keeps showing up isn't broken links. It's disconnected ones. Pages that exist, rank for something, and sit there with almost nothing pointing to them from inside the site itself.
1. The audit nobody runs
Everyone runs a broken-link check. Almost nobody runs an orphan check — a straight list of every page with zero or near-zero internal links pointing at it.
When I do run that check, the results are rarely subtle. Service pages that back up a homepage's biggest claims, sitting completely unlinked except from one blog post's footer. Case studies that never get referenced from the About page, even when the About page is narrating the exact story the case study proves. Blog posts published in the same category, never once linking to each other.
None of this shows up in a broken-link report, because nothing is broken. The links that should exist were just never written.
2. What orphaned pages actually cost you
A page with no internal links pointing to it isn't just harder to find. It's a page your own site is telling search engines and AI systems doesn't matter much.
Crawlers use internal link density as a proxy for importance. So does the "entity verification" layer inside AI answer engines — when a model is deciding whether to trust a claim on your homepage, it often follows the links from that page to see if the claim holds up elsewhere on the site. If there's nowhere to follow, there's nothing to verify.
That's the part most site owners miss. It's not "orphaned pages rank worse." It's "orphaned pages can't be corroborated," and corroboration is increasingly the whole game.
3. Anchor text is a semantic signal, not a chore
Generic anchor text — "click here," "read more," "learn more" — tells a crawler almost nothing about the page it points to. Specific anchor text does the opposite: it tells the destination page's topic before the destination page even loads.
The sites that get this right tend to hyperlink the actual claim, not a vague call-to-action wrapped around it. If a homepage says "I work on GEO and AI Overviews," the words "GEO and AI Overviews" should be the link — not a separate "learn more" button underneath. The anchor text carries the topical signal. Wasting it on filler words is leaving free semantic clarity on the table.
4. The two-click rule isn't about UX anymore
"Every important page should be two clicks from the homepage" used to be a usability guideline. It's now also a crawl-and-citation guideline. The deeper a page sits in your architecture with no shortcuts pointing to it, the less confident a crawler — human or AI — is that the page matters enough to represent your expertise.
This is where topic hub pages earn their keep. A category page that gathers every related post and links out to the relevant service page does two things at once: it shortens the path to every individual post, and it tells the crawler explicitly which pages belong to the same topic cluster.
5. What I'd check first
If you're auditing your own site this week, skip the broken-link scan and start here instead:
- List every page with fewer than two internal links pointing to it
- List every specific expertise or service claim on your homepage and check whether it's hyperlinked to proof
- Check whether pages in the same content category link to each other at all
- Check whether your highest-authority page (usually the homepage or a cornerstone guide) is the source of most of your internal links, or just a dead end
Backlinks are still worth earning. But a site can't make the most of external authority if its own internal structure doesn't distribute it anywhere. Fix the architecture first. The rest compounds faster once it's there.
