Internal linking

Search engines find pages by following links, and they judge a page partly by what links to it. Internal linking is the part of that you fully control.

The Links view turns your own link graph into a score and a list of findings. This page explains what it measures, so the numbers mean something before you act on them.

It needs a content crawl

Link analysis reads the body of each page, not just the head. A metadata crawl collects titles and canonicals and nothing else, so it produces no link graph at all.

If the view is empty on a site you have crawled, that is usually the reason. See crawling and categories for the difference between the two crawl types.

Link analysis also runs on sites up to 10 000 pages. A larger site is told so, rather than shown a partial answer that would read as a complete one.

The score, and what it is built from

The number is a 0–100 summary of internal linking with a verdict beside it: Strong, Needs watching, or Urgent. Under it sits a count of how many of the findings currently have something in them.

Every one of those numbers comes from the last crawl. When the view says there is no new crawl data since the last analysis, it means you are reading a description of an earlier version of your site. Re-crawl before acting.

A broken link and a refused request are different things

A broken link is an internal link pointing at a page that answers with an HTTP error. That is a real defect on your site, and fixing it is work.

Four status codes are deliberately excluded: 429, 503, 508 and 509. In each of those the origin refused the request — too many requests, temporarily unavailable, resource limits. The page is not broken. Our crawler arrived at a bad moment.

Counting them would blame your site for our timing, and it would move the score every time a shared host was busy. So they are left out. A page that genuinely 404s is still counted, immediately.

Redirects cost you twice

A redirect link points at a URL that answers 301 or 302. The link still works, but the value it passes takes a detour. Updating it to the final URL is one of the cheapest fixes available.

A redirect chain is several of those in sequence — A goes to B goes to C. Each hop wastes crawl budget and dilutes the link further, and chains tend to grow quietly as a site is reorganised over the years.

Orphan and weak pages

An orphan page has no internal link pointing at it at all. It exists, it may be in your sitemap, and search engines will reach it late or not at all. The fix is a contextual link from a page that is already well linked.

A weak page has very few inbound internal links — measured against your site's own median, not a fixed number. A site where the typical page has forty inbound links has a different idea of "few" than one where the typical page has three. Both the threshold and the median are shown on the row so you can see which site you have.

Click depth is the one to read first

Click depth counts how many clicks from the home page it takes to reach a page by following links. Anything you care about ranking should sit within three.

Unreachable pages could not be reached from the home page at all. A large number there matters more than any single finding above it, because it says the structure — not individual links — is the problem.

Before you trust the counts

URLs still awaiting your review are counted nowhere. Approving one includes it in the dashboard, rejecting one excludes it entirely, and until you decide it sits outside every number on the page.

A site with a large pending backlog is not a site with few problems. It is a site that has not been asked yet.