Internal linking: why search engines never find your pages

A sitemap states an address, links state importance

Two things people constantly conflate. A sitemap is a list: here are the addresses that exist. Internal links are an assessment: here are the pages I myself consider worth reaching.

A search engine reads both and believes the second. The logic is simple and quite human: if the site's owner does not think a page worth linking to from anywhere else, it is probably not important.

Hence a situation that looks paradoxical: the address is submitted, the page opens, it returns 200 — and it stays out of the index for months. Not because it is bad, but because nobody claimed it mattered.

Below is how internal linking works, where the hole usually is, and how to check your own in an hour.

Our case: twenty-five links to three hundred and one pages

While working out why 738 of our 3310 pages were indexed, we reached the product section and counted links.

The section page linked to twenty-five product pages. There were three hundred and one product pages in total.

The other two hundred and seventy-six existed, opened and sat in the sitemap — but the only way to reach them was through eleven pagination pages. Those pagination pages were not in the sitemap at all.

So we were asking the crawler to cover nine hundred addresses while telling it, through the structure of the site, that twenty-five of them mattered. It believed the structure.

This is not a rare fault. It appears on its own the moment a catalogue outgrows the first page of a list, and it almost never gets noticed, because to a person with a working site search everything looks fine.

Orphan pages and how to find them

An orphan page is one that no internal link points to. It exists, but the only way in is the direct address.

There are two ways to find them. First: export the list of all addresses on the site, separately collect every internal link, and subtract the second from the first. The difference is your orphans.

Second, without a tool to hand: take the sitemap and check a sample of twenty addresses by hand — try to reach each one from the home page using links only. However many of the twenty are unreachable is roughly the percentage of orphans on your site.

Check separately the pages created outside the normal flow: old landing pages, promotion pages, one-off publications. Those are orphaned most often.

And remember the order of operations: an orphan does not necessarily have to stop being one. Sometimes the right conclusion is that the page is not needed and should go.

The three-click rule is about people, not robots

The old rule says every page should be within three clicks of the home page. As a formal requirement it does not work — on a site with a thousand pages three clicks is physically unreachable for all of them.

But there is a true idea behind it. The longer the path from the home page, the less weight arrives and the less often a crawler looks in. Depth matters, even if the particular number is arbitrary.

It is more practical to count not clicks but a question: is there a natural path to this page that a real person would take? If the only way in is paging through a list, there is no path.

So long lists are broken up by meaning rather than by pagination: by curated groupings, categories and filters that have addresses of their own. Pagination is a way of not showing everything at once, not a way of navigating.

Four kinds of internal link

They do different jobs, and it is worth not confusing them.

Navigational — menu, footer, breadcrumbs. Identical on every page, so they pass little, but they set the frame. Breadcrumbs are underrated here: they are cheap and they show a search engine the hierarchy clearly.

Contextual — links inside the text. The most valuable, because they sit in content that explains where the link leads. One of these is worth several navigational ones.

Block links — related products, read also, other services. Cheap to implement and good for depth, provided they are chosen by meaning rather than at random.

And cross-links — direct connections between neighbouring pages at the same level. Almost nobody builds them, and they are precisely what pulls deep pages out of isolation.

Anchors: what is wrong with read more

An anchor is the text a link sits on. It tells both a person and a search engine what is on the other side.

Read more, here, click here tell them nothing. The page receives a link but no explanation of what it is about. For somebody using a screen reader it is meaningless as well: they hear a list of twenty read mores.

A working anchor names the thing: ventilation installation in a flat, not our service. It does not have to be an exact search phrase — what matters is that the anchor makes the content clear.

The same anchor on every link is also bad, for a different reason. A natural site links to a page in varied words, and a range of anchors looks more normal than a hundred identical ones.

A practical test: read only the anchors on a page aloud, without the surrounding text. If it is clear where they lead, the anchors are fine.

Pagination: the commonest hole

A list split into pages is where internal linking breaks most often, and almost always silently.

The first mistake: pagination pages are missing from the sitemap. Then the only route to deep items in the list is not declared at all, and a crawler reaches it last.

The second: pagination built in script without ordinary links — a load more button that fetches items. Convenient for a person; to a crawler it means there is nothing past the first batch.

The third: pagination pages carrying a canonical address pointing at page one. This is common advice and it is harmful: the search engine stops treating pages two onward as pages in their own right and does not follow the links on them.

The right way is simpler: ordinary links, each pagination page canonical to itself, all of them in the sitemap. And in parallel, a way to reach items that is not paging.

How many links belong on a page

A question people love to answer with a number, and there is no number.

The old limit of a hundred links was dropped long ago. The problem is not the count but that the more links a page carries, the less significance each one receives.

A practical guide: if a page has two hundred links and half are menu and footer, a hundred substantive ones remain, and each weighs half what it would at fifty.

The conclusion is not cut your links but do not link to everything. A service page linking to nineteen other services dilutes its own weight and says nothing about priority.

Three to five meaningful contextual links work better than twenty automatic ones.

How to build linking inside a catalogue

The commonest case and the most neglected. This is the order we are applying to our own section.

Curated groupings first. Not by formal categories but by how a person chooses: by task, by budget, by occasion. Each grouping has its own address and links to ten or twenty items.

Then cross-links between neighbours. Three to five links from every item page to similar ones. This is generated automatically, but the selection has to follow meaning: same category, same price level, a complementary item.

Then pagination into the sitemap, with ordinary links inside it.

And only then the item pages back into the sitemap — in tranches, starting with those that already have links pointing at them. The reverse order repeats the same story a month later: addresses submitted, no links, no crawl allocated.

Checking your own linking in an hour

Five actions, each answering something different.

Count how many links go from a section page to its items and compare that with the total number of items. A gap of several times over is your hole.

Take twenty random addresses from the sitemap and try to reach each from the home page using links only.

Check whether pagination pages are in the sitemap and whether they carry canonical addresses pointing at themselves.

Read the anchors on three key pages without the surrounding text. Is it clear where they lead?

And look at whether anything links to your commercial pages from substantive text — from articles, descriptions, answers. If only the menu and footer point at a service page, the search engine knows exactly that it is in the menu.

What not to do

Three tactics that look logical and do damage.

Do not auto-link every occurrence of a word in the text. You end up with a page where half the words are links, the anchors are meaningless, and reading is impossible.

Do not build an all services block linking to a hundred pages in the footer of every page. That is not internal linking but dilution: a hundred links from every page on the site say all hundred matter equally, which means none does.

And do not block internal links from passing weight in the hope of hoarding it for important pages. The tactic has not worked the way it is described for a long time, and it makes working out what leads where noticeably harder.