Internal linking: the structure that decides what ranks, and the audit that finds the gaps

Internal links decide which of your own pages gets found, crawled often and treated as the main page on a topic. On a small site the navigation does that job by itself. Past a few hundred URLs, the links you place inside the body copy become the strongest instruction you give a search engine about what matters, because they pass signals, they create the paths a crawler follows, and their anchor text describes the destination. The rule to build on comes from Google itself: every page you care about should have a link from at least one other page on your site. The work is finding the pages that do not, and linking to them from somewhere that already has attention.

This page covers internal link structure, anchor text and the audit that finds gaps. For links from other websites, see a guide to link building.

Why internal links decide what ranks on a large site

• They are how pages are discovered. A URL with no internal link pointing at it relies on a sitemap entry and luck. A page linked from three well visited pages gets crawled on a regular rhythm.
• They distribute signals. The authority your site has earned flows along internal links. Which pages you link to, and from where, is how you decide where it lands.
• They settle which page owns a topic. When three of your pages could answer the same query, the one your other pages link to with that phrasing is the one treated as the answer.
• They describe the destination before the crawler arrives. Anchor text is a label you control, and on a large site it is the cheapest content signal available.
• They reduce waste. On a big catalogue, crawling is finite. The Google guide to managing crawl budget explains why link structure matters once a site is large enough for that to bite.

Make the links crawlable in the first place

The Google reference on crawlable links is specific, and worth quoting in any front end code review. Google can only crawl a link that is an HTML anchor element with an href attribute whose value resolves to a real web address. Four consequences:

• A span or a div with a JavaScript click handler is not a link. It is a button, and nothing follows it.
• An href containing a JavaScript call rather than a path does not resolve to an address.
• Absolute paths, root relative paths and query strings are all fine.
• For an image used as a link, the alt text acts as the anchor text, so write it as a description of the destination.

The patterns that work

Pattern	What it does	Use it for
Hub and spoke	One overview page links to every detail page on a topic, and each links back	Service lines, product categories, guide series
Breadcrumbs	Gives every page a stated place in the hierarchy	Any site more than two levels deep
Contextual body links	Passes the strongest signal, because the surrounding text gives it meaning	The three to six most relevant links per page
Next step links	Moves a reader from learning to buying	End of every guide, pointing at the service or product page
Curated related items	Connects pages a reader would genuinely want next	Better than an automated "related posts" block, which links by tag rather than by sense
Footer and mega menu	Site wide reach for a small number of important pages	Key pages only. A footer with 200 links tells a crawler nothing

Breadcrumbs are worth marking up with structured data so they can appear in results. The Google documentation on breadcrumb structured data has the schema, and the SEO starter guide covers the basics of organising a site so links reflect how people actually move through it.

Anchor text, in five rules

• Describe the destination. The Google guidance is that good anchor text is descriptive, reasonably concise, and relevant to both the page it sits on and the page it points to.
• Retire "click here" and "read more". They describe nothing, and they are poor for anybody using a screen reader.
• Vary the phrasing naturally. The same exact phrase repeated from four hundred pages reads as a template rather than a recommendation.
• Three to six words. Long enough to be specific, short enough to sit inside a sentence.
• Link once per destination per page, at the first natural mention. Three links to the same URL from one page adds nothing.

Finding the pages that need links

This is an audit you can finish in an afternoon, and it is where the gains are.

• Crawl the site with a desktop crawler such as Screaming Frog and export inlinks per URL. Sort ascending. Everything with zero or one internal link is your worklist.
• Find orphans properly by comparing your XML sitemap against the crawl. A URL in the sitemap that the crawl never reached is unreachable by a human too.
• Pull the near miss queries from Search Console. Filter the Performance report to positions roughly 5 to 20 with real impressions. Those pages are close, and an internal link from a strong page is the cheapest push available.
• Use site search to find unlinked mentions. Search your own domain for the topic phrase to list every page that already discusses it, then add a link from the three best to the page that should own it.
• Check click depth. Any page more than three clicks from the home page needs either a link from a hub or a reason to exist.
• Look at your strongest pages and count their outgoing links. A page with external authority and no links out to your commercial pages is a dam.

A worklist format that gets done

Target page: /services/maintenance-audit   Position 11, 420 impressions/mo
  Link from: /blog/why-audits-fail      anchor: maintenance audit service
  Link from: /blog/downtime-costs       anchor: book a maintenance audit
  Link from: /services (hub)            anchor: maintenance audit

Target page: /guides/pump-selection     Orphan, 0 inlinks
  Link from: /guides (hub)              anchor: how to select a pump
  Link from: /blog/pump-failures        anchor: pump selection guide

Rule applied: 3 inlinks minimum for every commercial page,
              1 hub link minimum for every guide.

Set a standing rule so the gap does not come back. Every new page gets three internal links added on the day it publishes, from pages chosen by hand, and every guide links on to the commercial page it supports.

The internal linking questions worth an answer

How many internal links should a page have?

Enough to serve the reader, which in body copy is usually three to eight. The useful rule is on the receiving end: three internal links minimum for any page that is meant to earn money, and at least one for everything else.

Do internal links need nofollow?

No. Use nofollow on links you have no control over, such as user submitted content. Internal links to your own pages are an editorial decision you are making deliberately.

Are footer links as good as body links?

Body links are stronger, because the surrounding text gives them meaning and readers actually click them. Footers are useful for giving a handful of important pages site wide reach, and they lose value as the list grows.

What is the ideal site structure?

A hub per topic or service, detail pages under it, every detail page linking back to the hub and sideways to its genuine siblings, and nothing important more than three clicks from the home page.

How do I find orphan pages?

Crawl the site, export the URLs with no inlinks, then compare the crawl against your XML sitemap and your server logs. Anything in the sitemap that the crawl never reached is an orphan.

Does changing internal links work quickly?

It is among the faster things in search work, because you are not waiting for anyone else. Pages already sitting in positions 5 to 20 tend to show movement within a few weeks of getting better internal links.

Where those links point now matters twice

Internal links have always told search engines which of your pages owns a topic. They now do something extra. When somebody asks ChatGPT, Gemini, Perplexity, Claude or Google AI Overviews a question in your category, the answer is a short paragraph built from a handful of pages those systems fetched, and one of your URLs is either cited there or not. The page you have made the clear, well linked owner of a topic is the one most likely to be the one they reach and quote. It is worth knowing which of your pages, if any, are being cited in those answers today, because that is a different list from your Search Console top pages. AI Knows Us runs the buyer questions across the assistants and reports which sources come back.