Evidence brief · Information architecture

Internal linking for LLMs: functions and limits

Internal links have documented functions for users and search crawlers: they expose destinations, provide anchor context and connect site architecture. Their usefulness does not establish that every LLM computes a sitewide topical-authority score from link counts.

Updated 12 August 2026 6 min read

Use crawlable links

Google documents standard anchor elements with an href as the reliably crawlable link format. Navigation that depends only on scripted click handlers may not expose an equivalent crawl path.

Anchor text should describe the destination concisely and naturally. Generic labels can be appropriate in context, but descriptive wording gives readers and crawlers more information before they follow the link.

Design links around the next question

  • Link a definition to the method that applies it.
  • Link a recommendation to the primary evidence and stated limits.
  • Link an audit finding to the page that explains the remediation.
  • Remove or update links when the destination no longer resolves the promised question.

Measure outcomes, not link folklore

Track orphan pages, crawl depth, internal destinations and Search Console outcomes for a defined cluster. Do not prescribe a universal number of links per page. The correct number is the set of destinations that genuinely advances the reader task.

A traffic change after a linking update is an observation. It may justify further testing, but it does not isolate a hidden LLM authority mechanism.

Primary sources