Content authority

Content authority is the credibility a particular piece of content, and the person who made it, carries on a particular topic. It is a judgement about one page and one creator in one subject area, not a property a whole domain owns and lends to everything published on it.

That distinction is the whole term. A site can be authoritative on tax law and have no standing at all on cycling injuries, and the page about cycling injuries is judged on its own creator, its own sources and its own evidence.

What it is assessed against

Google's General Guidelines, the document its quality raters work from, dated 11 September 2025, sets out E-E-A-T — Experience, Expertise, Authoritativeness and Trust — and is explicit about the ordering: "Trust is the most important member of the E-E-A-T family because untrustworthy pages have low E-E-A-T no matter how Experienced, Expert, or Authoritative they may seem." Its own illustration is a financial scam written by a highly experienced expert scammer.

The guidelines also say where the assessment comes from, and it is not one place:

• What the site or creator says about themselves — the About page, the creator profile, taken as a starting point.
• What others say about them — independent reviews, references, news articles and other sources of credible information, checked in both directions: evidence that a creator is trustworthy, and evidence that they are not.
• What is visible on the page — the main content itself, plus reviews and comments, where experience is often demonstrable rather than asserted.

The guidelines add a warning that matters commercially: a source may not be trustworthy where there is a clear conflict of interest. A manufacturer's own "review" of its product, or a paid influencer's, is named as the example.

It is not a metric, and nothing publishes one

Google's Creating helpful content guidance states that "while E-E-A-T itself isn't a specific ranking factor, using a mix of factors that can identify content with good E-E-A-T is useful", and that its systems "give even more weight to content that aligns with strong E-E-A-T for topics that could significantly impact the health, financial stability, or safety of people".

So there is no authority number to raise. Third-party authority scores are vendor models built from link graphs; they are legitimate as heuristics and illegitimate as descriptions of what a search engine computes. Optimising a vendor's number is optimising a proxy for a proxy.

Why it matters to being cited by an agent

An assistant answering from retrieved documents is doing something a ranking system does not: attaching a claim to a source in front of a user. Whatever the retrieval scoring, the material that survives that step tends to share the same properties, and they are exactly the ones the guidelines describe as evidence rather than assertion.

• A claim that names who measured it, over what period, is checkable. A claim that floats is not.
• A primary source linked in the passage lets the answer inherit the source's standing. A link to an article about the source does not.
• A named author with an independently verifiable profile answers the "who" question Google asks — is it self-evident to visitors who authored this? — inside the passage itself, where it travels.

Failure modes

• Treating authority as site-wide. Publishing outside your demonstrated competence on the theory that domain strength carries. It does not carry to a rater reading one page.
• Bylines without bios. A name with nothing behind it satisfies a checklist and no reader.
• Citing summaries. Linking a blog post about a specification instead of the specification is a link that adds nothing and can be wrong.
• Undisclosed conflict. Reviewing your own product in the voice of an independent assessment is the case the guidelines call out by name.
• Statistics with no origin. A percentage with no source is the fastest way to make a page unquotable by any system that has to stand behind what it repeats.
• Freshness theatre. Bumping a date without changing anything trains crawlers to ignore your dates.

How to build it, concretely

1. Publish a real author identity — a name, a linked profile, and a statement of why this person is the one writing on this subject.
2. Cite primary sources inline, with the date you checked them, and prefer the operator's or standards body's own document.
3. State methodology whenever you publish your own measurement: what was measured, over what window, with what sample.
4. Disclose commercial interest wherever you write about your own category. See source citation for what a citation actually asserts.
5. Mark up the author and organisation with schema so the identity is machine-readable as well as visible.
6. Scope your topics. Fewer subjects covered with demonstrable experience beats broad coverage with none — a point GEO and AI search both turn on.

Frequently asked questions

Is it a published score?

No. There is no authority number from a search engine; vendor scores are models.

Which E-E-A-T component dominates?

Trust. Untrustworthy pages have low E-E-A-T regardless of the other three.

Where does the assessment come from?

Self-description, independent third-party evidence, and the page's own content.

Does a byline help by itself?

Only if it can be checked. A name with no verifiable profile adds nothing.