Skip to content

Core mechanics · module 7 of 13 · 7 min read

Links, entities and reputation

How does a search engine decide a site is worth trusting?

  • backlinks
  • anchor text
  • nofollow/sponsored/ugc
  • E-E-A-T
  • entities
  • link spam

Read first: What "ranking" actually means

Links were the original insight behind Google: a link is a person vouching for a page, and counting votes beats counting keywords. That idea is still in the system, heavily modified, and heavily gamed.

What a link is measuring

Not volume. The useful mental model is: would this link exist if search engines did not? A link inside relevant editorial content, on a site with its own audience, is evidence. A link in a footer on 40,000 pages, in a directory nobody reads, or bought, is not — and detecting the difference has been an explicit priority for two decades.

Anchor text — the visible words — tells Google what the linked page is about. Natural anchors vary; a thousand identical commercial anchors is a pattern, and patterns are what spam systems detect.

The attributes

  • rel="nofollow" — do not pass a vote.
  • rel="sponsored" — paid or compensated.
  • rel="ugc" — user-generated, comments and forums.

Use them honestly. Paid links without sponsored (or nofollow) are a link-scheme violation, not a clever tactic.

Beyond links: entities and reputation

Modern systems reason about entities — people, organisations, products as things rather than strings — and about the reputation attached to them. Being mentioned, reviewed, cited and referenced consistently across the web builds a picture that is not purely link-based.

E-E-A-T — experience, expertise, authoritativeness, trust — is not a score in the system. It is the vocabulary Google's quality raters use, and it describes what the ranking systems are trying to approximate. It matters most where being wrong hurts people: health, finance, safety, legal.

Where the line sits

Google's spam policies name link schemes explicitly: buying or selling links that pass ranking signals, excessive exchanges, large-scale guest posting with keyword-rich anchors, and automated link building. The consequences range from links being ignored — the common case — to a manual action.

The 2024 site reputation abuse policy extended this to a related trick: publishing third-party content on a trusted domain purely to borrow its reputation. Ranking well because a strong host site vouched for content it has nothing to do with is the thing being targeted.

What trips people up

  • Buying links and calling it PR.
  • Treating a third-party "authority" metric as a goal. It is a vendor's model, not Google's.
  • Ignoring unlinked mentions, which are often easier to earn and increasingly useful.

You have got this when

You can look at a link opportunity and answer honestly whether it would exist if search did not — and you are comfortable saying no to the ones that would not.

Go to the source

What has changed since

Stories from the briefs that touch this module.