Internal linking as structure

What internal links are for

Three things, in order of importance. They let a reader move to the thing they now want. They tell a crawler that a page exists and is reachable. And collectively they express a structure: what is central, what supports what, what belongs together.

The third is the one clinics neglect and it is the reason internal linking belongs in the technical part of this course rather than the writing part. A site whose links contradict its navigation is a site making two different claims about what it is.

The hub pattern

For each treatment you compete on, the treatment page is the hub. Supporting pages are the concern page for the problem it addresses, the answer pages for questions about it, the relevant comparison page, and the fees page. Every supporting page links to the hub. The hub links to every supporting page, in context, where the reader would want it.

In context matters. A block of related links at the foot of a page is worth much less than a link inside the sentence that raises the question, because the reader who has just formed the question is the reader who will follow it.

Link text

Link text should describe what is at the other end, in the words a reader would use. Three rules cover almost every case.

• Describe the destination, not the action. "What recovery from this treatment involves" rather than "click here" or "read more".
• Vary it naturally. The same page linked from ten places will be described ten slightly different ways, which is normal writing. Identical text every time reads as mechanical.
• Do not over-optimise. Repeating a target phrase as link text everywhere is a pattern search engines have recognised for many years and it does more harm than good.

Google's documentation on link best practices is short and covers both the crawlability requirement and the descriptive text point.

Orphans and dead ends

An orphan is a page nothing links to. A dead end is a page that links to nothing. Both are symptoms of the same neglect and both are easy to find.

Orphans are found by comparing your sitemap against a crawl that starts at the homepage: anything in the first list and not the second is orphaned. Dead ends are found by looking for pages with no outbound internal links, which are usually pages imported from an old site or produced by a form builder.

Fix orphans by giving the page a parent, per Module 09, Information architecture: the site plan, and linking it from there. Fix dead ends by asking what the reader of that page wants next and linking to it.

Two patterns to avoid

The footer full of links. A footer listing every page on the site makes every page equally linked, which tells a crawler nothing about what is important and gives a reader an unreadable wall. Keep the footer to genuine site-wide navigation.

Automated related-content blocks. A widget that inserts links based on shared words produces links a reader would not want, from sentences that do not raise the question. If you use one, check what it produces and be prepared to turn it off.

A link is a recommendation you are making to your own reader. If you would not make it aloud, do not make it in HTML.

Maintaining it as the site grows

The rule that keeps this manageable: when you publish a new page, before you publish it, link to it from at least two existing pages and link from it to at least two existing pages. Four links, decided at the moment when you understand the page best.

Doing this at publication takes two minutes. Doing it retrospectively across forty pages takes a day, which is why it never happens. Put it at the foot of your publication checklist alongside the compliance check from Module 05, The rules that constrain everything you publish.