Internal Links: Make Important Pages Findable
Key takeaways
- Google generally needs a real HTML link with an
hrefattribute to crawl a destination reliably: Google's link guidance. - Every important page should have a link from at least one other page on the site, according to the same guidance.
- Descriptive, relevant anchor text explains the destination; keyword stuffing and a fixed links-per-page quota are not substitutes for usefulness.
- Google's AI search guidance includes internal discoverability among ordinary SEO fundamentals, but eligibility does not guarantee inclusion: AI features and your website.
- An internal-link audit should record source pages, exact destinations and verification evidence, not just a rising total of links.
What are internal links supposed to achieve?
Internal links connect pages on the same website and help people find the next useful piece of information. Google says it also uses links to discover pages and as a signal when determining relevance.
For a business website, the useful question is not how many links an article contains. It is whether someone with a specific problem can reach the page that answers their next question or explains the relevant service.
A buyer reading a materials guide may need the matching product category. Someone checking installation requirements may need the service scope or enquiry route. Someone comparing approaches may need a detailed explanation before a sales conversation makes sense.
Our judgement is to map those transitions before adding links. Turning every paragraph into a doorway to the same contact form is technically navigation, but not necessarily hospitality.
Which pages should you audit first?
Start with commercially important pages and the articles that naturally lead to them. This is a prioritisation recommendation, not a claim that Google assigns a special status to your sales targets.
Build a short list of current service pages, relevant categories and useful comparison or decision guides. For each destination, identify the reader question it resolves and the action it legitimately supports.
Then inspect likely entry pages. Can a reader reach the next relevant page from the body, a genuinely useful related-reading block or clear navigation? Does the destination still offer what the link promises?
For a UK service business with local pages, our preference is to link to the location or service that actually applies. Do not send every regional article to a generic national page simply because that is the easiest URL to remember. Equally, do not manufacture near-identical city pages just to create more destinations.
What makes a link crawlable?
Use a real anchor element with an href pointing to the intended destination. Google's documentation says it generally cannot reliably extract destinations from elements without the recommended HTML link structure.
A button-like visual treatment is not itself a problem. What matters is the underlying link. Ask the developer or content team to inspect the rendered page rather than assuming that something clickable by a person is necessarily implemented as a crawlable link.
Google also says JavaScript-inserted links can be crawlable when they use the recommended markup. The distinction is not simply JavaScript bad, static page good. Check what is actually present after rendering.
Our acceptance test is to record the source page, rendered link element and exact destination, then follow that destination. Keep links intended for discovery separate from controls that perform actions, such as submitting a form.
How should you write the anchor text?
Write a concise description that sets the right expectation for the destination. Google recommends relevant, descriptive anchor text and natural wording rather than cramming related keywords into a link.
Try reading the linked words without the rest of the paragraph. Would a reader have a reasonable idea of what opens next? Then read the whole sentence to check that the link belongs there.
Our editorial rule is to link the specific concept, service or resource being discussed. Avoid a chain of adjacent links with no explanation between them. The surrounding words provide useful context too, a point Google explicitly makes in its guidance.
There is no need to rewrite a perfectly clear sentence into a geographical keyword parade. Your navigation should not sound as though it is being paid by the town name.
How do you find orphaned or misleading destinations?
Compare your known important pages with the pages reached through internal links, then inspect the gaps. For this audit, an orphan page means a page in your known inventory with no internal incoming link found in the crawl you reviewed.
That is an operational definition. A missing link in one crawl does not prove there is no link anywhere, especially if the crawl excluded sections or did not render the relevant content. Record the crawl's scope before treating its findings as complete.
Compare crawl results with your content inventory and published sitemap, then check suspected gaps manually. Our XML sitemap guide covers the discovery file; this audit asks whether readers also have a useful route through the site.
Check misleading links as well as missing ones. A live destination can still be wrong: an old service, an irrelevant locale or a page whose content no longer matches the anchor text. A successful response code is not proof of editorial relevance.
If duplicate URLs are complicating the destination list, review the canonical-tag guide before adding more links to competing versions.
What should a practical link audit record?
Record each proposed change as a source-to-destination decision with a reason and a verification result. Our suggested worksheet is deliberately small enough for a content team to maintain.
| Field | What to record |
|---|---|
| Reader need | The next question or task the link helps with |
| Source | Exact live page where the link belongs |
| Destination | Exact intended URL and its current purpose |
| Anchor | Proposed visible wording in its sentence |
| Technical evidence | Rendered link, response and redirect check |
| Acceptance | Destination relevance, working navigation and review date |
Fix incorrect or dead destinations before adding decorative extras. Then address important pages without a useful incoming route. Finally improve vague anchors and inconsistent destination choices.
Do not set an arbitrary quota. Google explicitly says there is no magical ideal number of links a page should contain. The spreadsheet can count links; it cannot decide whether a reader needed them.
Do internal links help AI search visibility?
They belong to the foundations Google recommends for its AI search features, but they do not guarantee a citation. Google's AI guidance specifically includes making content easily findable through internal links.
The same guidance says a page must be indexed and eligible for a search snippet to appear as a supporting link in AI Overviews or AI Mode. It also says meeting requirements does not guarantee crawling, indexing or serving, and no special AI markup is required.
Keep two kinds of evidence separate. Implementation evidence shows that the intended links exist and work. Outcome evidence shows what happened later in search visibility, observed citations, relevant visits or qualified enquiries.
Google reports traffic from its AI features within the overall Web search reporting in Search Console, rather than providing the audit described here with a clean causal result. Treat any improvement after a link change as an observation to investigate, not automatic proof that the change caused it.
What should you do next?
Pick one commercial journey and verify it from useful entry page to appropriate next step. Use the SAGEO framework to connect technical accessibility, clear content and the business purpose of the destination.
If you want to discuss the wider approach, use the conversation routes on that page with your priority-page list and the gaps you found. Ask for evidence against that brief, not a guaranteed number of links or AI mentions.
Limitations
This is an audit method grounded in Google's published guidance, not a study measuring ranking gains. Google's rules for its own AI features do not establish how every other answer engine discovers or selects sources. Crawl coverage, rendering and analytics configuration affect what you can verify, and no traffic or lead outcome is promised.
Generated illustrations are conceptual scenes, not real client dashboards, measured results or testimonials.