Topical authority

Topical authority is the state of being recognised — by readers, by other publishers, and by the systems that rank and retrieve — as the source people go to for one specific subject, rather than as a shallow source for many. It is a property of a site or an author on a topic, not a score, and not something a single page earns on its own.

What the term actually rests on

The phrase is industry vocabulary, not Google's. The nearest thing to an official definition sits in Google's Search Quality Rater Guidelines, the 182-page document dated 11 September 2025 that Google gives to human raters. Under E-E-A-T it defines the A directly: "Authoritativeness: Consider the extent to which the content creator or the website is known as a go-to source for the topic." The same section adds a caveat worth keeping: "most topics do not have one official, Authoritative website or content creator."

Two things follow that are routinely got wrong.

It is scoped to a topic, not to a domain. The guidelines instruct raters that reputation research "should be performed according to the topic of the page", and give the case explicitly: a website can be "a go-to source for one type of content (e.g., humorous videos), but an untrustworthy source for a different type of content (e.g., financial information)." Authority does not pool at the domain and spill onto whatever is published next.

It is not a ranking factor. Google's own helpful content guidance states that "E-E-A-T itself isn't a specific ranking factor", and that rater data "is not used directly in our ranking algorithms". Raters describe the target; the systems approximate it with other signals. Anyone selling a topical-authority score is selling a proxy for a proxy. See E-E-A-T and domain authority for the two most-confused neighbours.

Why it matters when the reader is a retrieval system

An AI assistant answering a question does not read your site. It retrieves passages — see RAG — and the selection is made against a query, not against your homepage. That changes what authority has to look like in practice.

Depth on one subject produces the thing retrieval needs: many distinct, non-overlapping passages that each answer one real question, tied together by consistent naming of the same entities. A site with forty pages spread across nine subjects has, per subject, four thin passages competing with each other. A site with forty pages on one subject has forty candidate passages and no internal competition.

Consistency of naming is the other half, and it is closer to entity SEO than to keywords: the same product, person and concept named identically everywhere, so that the same thing being described across pages is legible as the same thing.

Failure modes

• Publishing breadth to signal depth. Adding a ninth subject dilutes the eight you had. The guidelines' topic-scoped reputation rule means the new subject starts from zero regardless of what the domain has earned elsewhere.
• Treating a third-party authority metric as the thing itself. Those metrics are link-graph estimates published by tool vendors. They are not Google's, and Google has no equivalent public number.
• Hiding the creator. Google's guidance asks whether it is "self-evident to your visitors who authored your content" and whether pages "carry a byline, where one might be expected". An unattributed page forfeits the creator half of the assessment entirely.
• Assuming size is the barrier. The guidelines state that many smaller websites have little reputation information and that a page can still receive the Highest rating without it.
• Writing pages that only make sense in sequence. A passage lifted into an answer arrives without the pillar page above it. If the claim depends on context three pages back, it fails at exactly the moment it was selected.

How to build it

1. Pick a subject narrow enough to exhaust. The test is whether you can list every question a practitioner asks about it. If you cannot list them, the subject is too broad to own.
2. Cover the unglamorous questions. Competitors skip the ones with no search volume. Those are frequently the ones a retrieval system has nothing else to return for.
3. Attribute everything to a named person with a linked profile, and keep that name stable across the subject.
4. Cite primary sources inline, not summaries of them — operator documentation over an article about operator documentation. This is also what makes your own page quotable; see source citation.
5. Link within the subject, not across all of them. Internal links are how you tell a crawler which pages belong to one body of work.
6. Update rather than republish. Coverage that goes stale reads as abandonment of the topic, which is the opposite of the claim being made.

Frequently asked questions

Is it a ranking factor?

No. Google states E-E-A-T is not a specific ranking factor and that rater data is not used directly in ranking.

Where is it defined?

In Google's rater guidelines, as the content creator or website being "known as a go-to source for the topic".

Does it carry across subjects?

No. Reputation is researched per topic; a trusted source for one type of content can be an untrusted source for another.

Can a small site have it?

Yes. The guidelines allow a page to reach the Highest rating with little reputation information available.