Internal linking for AI search: the pages your own site votes for
Why sitewide links tell you nothing, what an orphan page really costs, and how to find both — with real numbers from crawling our own site.
Internal links are the only ranking signal you control completely. No outreach, no waiting, no one else's approval. And on most sites they are quietly broken in a way nobody notices, because the thing that breaks is invisible in every dashboard: the pages your own site says are important are not the pages you would name.
Why this matters more for AI search than it used to
Engines use internal links for two things: finding pages, and judging which ones matter. For AI answers the first one dominates. An answer engine has to retrieve a page before it can quote it, and retrieval depends on the page having been crawled recently enough to be in the index at all. A page nothing links to gets crawled once, if the sitemap is read, and then largely forgotten.
That is a harsher outcome than it used to be. In traditional search a rarely-crawled page still sits in the index and can surface for a long-tail query. In AI search, a page that was not retrieved simply does not exist for that answer.
The measurement mistake almost everyone makes
Count inbound internal links naively and your privacy policy wins. It is in the footer of every page, so on a 200-page site it collects 200 inbound links — more than any article you have ever written. Your terms page comes second. This is not a quirk to work around; it is the reason most internal link reports are useless.
Our internal link checker does this split automatically: any target appearing on 80% or more of the crawled pages is classed as template furniture and reported separately, leaving a ranking of pages that earned their links.
Three problems worth finding
Orphan pages
A page nothing links to. It is usually in the sitemap, so it can be discovered — but discovery and inclusion are separate decisions, and a page with no inbound links gives an engine no evidence that it matters. Orphans are the single most common reason a page stays unindexed on an otherwise healthy site, and the fix is almost never technical. It is one contextual link from a page that already gets crawled.
Click depth
Crawl frequency falls off sharply with distance from your homepage. Two clicks gets revisited regularly. Four clicks might be fetched once and left for months. That matters most for anything whose value depends on being current — pricing, availability, comparison tables with dates in them.
Broken internal links
Every one spends crawl budget on nothing and lands a reader on an error. They accumulate silently after any restructure, because nothing warns you when a link you wrote two years ago stops resolving.
What our own site looked like
We ran the checker against rankvyze.com while building it, which is the only honest way to describe what the output is worth. At the time of writing the site had 93 URLs in its sitemap. A 25-page crawl found 670 internal links across 94 unique targets.
| Clicks from homepage | Pages | What it means |
|---|---|---|
| 0 | 1 | The homepage itself |
| 1 | 24 | Everything in the nav and footer |
| 2 | 69 | The long tail — glossary, guides, tools |
| 3 or more | 0 | Nothing buried |
Zero orphans and zero broken links, which is the boring answer you want. But the finding that mattered was in the ranking rather than the errors: our highest-intent commercial page was reachable only through one hub page, two clicks from anywhere, while eight lower-intent pages sat in the footer. Nothing was broken. The site was simply voting for the wrong things.
How to fix it, in order
- 1
Find the orphans
Run the internal link checker against your homepage. Anything in your sitemap that nothing links to is the first list to work through, because those pages are getting no consideration at all.
- 2
Give each orphan one real link
From a page that is already crawled well, inside a sentence, with anchor text that describes the destination. One contextual link beats being added to the footer.
- 3
Pull your commercial pages up to two clicks
List the pages that make money. If any is three or more clicks from the homepage, it needs a link from somewhere shallower — usually the nav, the footer, or a hub page that is already shallow.
- 4
Fix broken links and stray nofollows
Both are cheap to fix and pure loss otherwise. Internal nofollow is almost always unintentional and usually arrives as a plugin default.
- 5
Re-run and compare
The ranking should now put your commercially important pages near the top. If it does not, your site is still telling engines something different from what you intend.
Frequently asked questions
- How many internal links should a page have?
- There is no correct number, and targets like "100 links per page" are folklore. What matters is that every page you care about receives at least one contextual link, and that important pages receive more than unimportant ones. Ranking your pages by contextual inbound links and checking the order matches your priorities is a better test than any count.
- Do footer links count?
- They count for discovery — a footer link guarantees a page is reachable and crawled. They count for very little in judging importance, because every page has them. Use the footer to guarantee reachability, and body links to signal importance.
- What is a good click depth?
- Within two clicks of the homepage for anything commercially important. Three is acceptable for archive and reference content. Beyond three, expect infrequent crawling and slow updates.
- Does anchor text matter for AI search?
- Yes, and arguably more than for traditional search. Anchor text is one of the clearest statements on your site about what a page is, and engines building an understanding of your entity read it as a label. "Click here" wastes that; the destination's actual subject does not.
- Can internal linking fix a page that is not indexed?
- Often, yes — if the reason it is unindexed is that nothing links to it, which is common. It will not fix a page blocked by robots.txt, carrying a noindex tag, or whose content only exists after JavaScript runs. Rule those out first with the meta tag checker.
Tools used in this post
See how AI engines describe you right now.
The free scan checks the technical signals in this article against your homepage — schema, rendering, crawler access, entity clarity — and scores each one.
Analyze My Website