The first thing our own link checker told us about our own blog was that the guide we had published the previous week was an orphan. Nothing linked to it. It was in the sitemap, it rendered, it was indexable, and no other page on the site mentioned it existed.
That is the normal state of a site that publishes regularly, and it is the reason this class of tool exists. Below is the full result of a 50-page crawl of toolz.dev, run with the three checkers on this site, with every number as it came back.
TL;DR: 50 pages, 4,534 internal links. 15 pages had nothing linking to them within the crawl. Zero dead ends, because the template puts about 80 links on every page, which is exactly the result that proves a clean dead-end score can be meaningless. Six anchor conflicts, four of them serious, including the phrase "video compressor" pointing at both the tool and its guide. Run the orphan page checker, the dead-end page checker and the anchor text checker on one crawl and read them together, because each one alone misleads.
What does an internal link audit actually measure?
Three questions about the same graph, which is why one crawl answers all three.
What links in? A page with no inbound internal links is an orphan. Search engines find it through the sitemap if at all, and it receives none of the authority that flows along internal links from pages that already rank.
What links out? A page with no outbound internal links is a dead end. A reader who arrives has nowhere to go, and the page passes nothing on to the rest of the site.
What do the links say? Anchor text is the phrase a link is written as, and it is one of the clearer signals about what the linked page is for. One phrase pointing at two different pages splits that signal.
The three fail in different directions, and reading one without the others is how an audit produces a confident wrong answer. Our own crawl is a good demonstration of that.
What 50 pages of our own blog looked like
| Measure | Result |
|---|---|
| Pages crawled | 50 |
| Internal links found | 4,534 |
| Average outbound internal links per page | 80.2 |
| Pages with no inbound link in the crawl | 15 |
| Dead ends (no outbound links) | 0 |
| Low connectivity (one or two) | 0 |
| Anchor phrases pointing at more than one page | 6 |
| Of those, three or more links involved | 4 |
Two of those rows are worth more than the rest.
Zero dead ends is not a good score here, it is a non-result. The average page carries 80 outbound internal links, and no article has 80 contextual links in its body. They come from the navigation, the category menu, the related-tools rail and the footer, which are on every page of the site. A dead-end check against a site with a large template tells you about the template. If that is your site too, the useful version of this check is to paste a list of article URLs and look at the number relative to the others rather than against the three-link floor.
Fifteen orphans out of fifty is the number to be careful with, in the other direction. Every one of them is orphaned within the crawl, and the crawl was 50 blog posts. On the live site most of them are linked from the blog index, from tag pages, and from the tool page each guide belongs to. That is the single most common way to misread this tool: a crawl boundary is not a site.
What it does tell us is real anyway. Those pages have no link from another article, which means they get nothing from the posts that already rank, and a reader finishing a related guide is never pointed at them.
Which conflicts did the anchor check find?
Six, and the shape of them is more instructive than the count.
| Anchor | Targets | Links | What it is |
|---|---|---|---|
toolz.dev |
/ and /tools |
51 | A sitewide element and one article disagreeing |
video compressor |
/tools/video-compressor and /blog/video-compressor-guide |
5 | The real thing this check exists for |
calculators |
/blog?tag=calculators and /tools/category/calculators |
4 | Two different pages with one name |
its guide |
three different guides | 3 | A generic anchor doing identity work |
The video compressor row is the textbook case. Four links use that phrase for the tool and one uses it for the guide about the tool. Both are defensible decisions in isolation, and together they tell a search engine that two different URLs are the thing called "video compressor". The fix is to pick one, which here is the tool, and give the guide an anchor describing what the guide is: something closer to "how video compression works".
The calculators row is the same problem wearing a different hat. A tag listing and a category hub are genuinely different pages, and calling both of them "calculators" in links means neither owns the word.
its guide is the generic case, and it needs a different fix. Three articles each link the words "its guide" to their own guide. No search engine learns anything from that, and a reader scanning the page learns nothing either. The repair is not to pick one target, it is to write the link text properly in all three places.
toolz.dev at 51 links is the one to leave alone. Fifty of those are a sitewide element pointing at the home page and one is a body link pointing at the tools index. It is a real inconsistency, worth a minute to fix, and it is not costing anything measurable.
How should you read your own numbers?
In this order, because each step changes how you read the next.
Set the boundary first. Crawl a section, not a sample. Fifty pages of one blog category, or one documentation tree, gives an answer you can act on. Fifty pages scattered across a site produces orphans that are artefacts of what you did not include.
Read the dead-end average before the dead-end count. If the average outbound link count is 60 or 80, the template is carrying the site and the count of zeros will be zero whatever the articles do. Compare sections against each other instead. A section averaging 85 and one averaging 81 are the same; a section averaging 12 is a different kind of page.
Treat every orphan as a question. Open it. If nothing should link to it, it is fine where it is. If something should, the link belongs in the body of a related page, not in a list at the bottom of the home page. Google's own guidance on link best practices is specific about this: a crawlable <a href> in content is what counts.
Fix anchors last and in batches. They are the fiddliest change and the one most likely to be reintroduced by the next article. Fixing them one at a time as they appear is how the same conflict comes back in three months.
Why do orphan pages keep appearing?
Because nothing in a normal publishing workflow creates the inbound link. You write a post, you publish it, and the only places that link to it are whatever lists posts automatically. Linking it from an older, related article is a separate editorial act that nobody is assigned, and the older article is the one that already ranks, which is exactly where the link is worth something.
Three other causes show up repeatedly:
Unpublishing a category or a tag. Every post that lived there loses its only inbound route at once.
Migrations. New templates are usually cleaner than old ones, and cleaner means fewer links. A redesign that removes a "related posts" block strands every post that had no body links.
Campaign landing pages. Built to be linked from an ad, never linked from the site, and still in the sitemap two years later.
Does any of this change rankings?
Indirectly, and the honest framing is that internal linking moves what you already have rather than creating something new.
An orphaned page is not penalised and is not invisible: a sitemap is enough for discovery. What it lacks is the authority that internal links pass and the context the anchor text supplies. On a site where the good pages are well linked, an orphan is competing without either.
A dead end costs the pages it should have linked to rather than itself. Anchor conflicts cost the clarity of both targets.
None of these are the reason a page is not ranking when the content is thin or the term is owned by three sites with a decade of links. They are the reason a page that should rank slightly better does not, and they are cheap to fix, which is the whole argument for doing them.
When you have the structure straight, the other half of the picture is what a given page actually says. The content analyzer covers the on-page side, and the category hubs list the rest of the measurement tools.



