Topical authority: what Google has said, what is inferred, and how to measure it

What is topical authority?

Topical authority is the degree to which search engines treat a site as a reliable source on a particular subject, so that its pages are more likely to rank across the queries in that subject, including new pages that have not built signals of their own yet. It is an outcome you can observe and measure, not a setting you can read in any tool.

Topical authority is the outcome; topic clusters are one mechanism for building it. How to structure a cluster is covered in the guide to pillar pages and topic clusters. This guide is about what the outcome is, what the evidence says produces it, and how to tell whether you have it.

What Google has actually said about topical authority

Most writing on topical authority treats it as a known ranking factor. The documented record is shorter, and worth knowing exactly, because it shapes what to build.

The news topic authority system

In May 2023 Google published an explanation of a topic authority system used in Search and Google News. As Search Engine Land reported, it helps surface expert sources for newsworthy queries in specialised areas such as health, politics and finance, and Google described three signals: how notable a source is for its topic or location, influential and original reporting, which Google gauges from links and citations among other things, and the reputation of the source, based on its history of high-quality reporting and recommendations from expert sources. Google’s advice to publishers was to keep providing great coverage of the areas they know well.

That system is documented for news. Nothing Google has published says the same system ranks B2B guides or product pages. What it does show is the kind of evidence Google treats as authority on a topic: being known for it, being cited for original work on it, and a track record of getting it right.

Site-level signals, confirmed by a spam policy

The clearest confirmation that sites carry topic-relevant signals comes from a spam policy. Google’s site reputation abuse policy applies where third-party content is published on a host site mainly because of the host’s established ranking signals, which the policy says were earned primarily from its first-party content. In other words, a site’s standing is earned by its own content on its own subjects, and Google treats lending it to unrelated content as abuse.

What the helpful content guidance asks

Google’s helpful content guidance asks whether your site has a primary purpose or focus, and whether you are producing lots of content on many different topics in the hope some of it performs. Those are questions for self-assessment, not a description of a signal, but they point the same way: focus is expected, and sprawl is a warning sign.

Claim	Evidence	Strength
Google has a system called topic authority	Documented by Google for news queries in 2023	Documented, for news
Notability, original reporting and reputation matter for it	The three signals Google described for that system	Documented, for news
Sites earn ranking signals from their own content	Stated in Google’s site reputation abuse policy	Documented
A site has a measurable focus on a topic	Attributes named siteFocusScore and siteRadius in the 2024 leak	Inferred; use unknown
Topical authority makes new pages rank faster	Vendor studies, such as the one Graphite published	Correlational
Publishing more pages on a topic raises authority	Common advice, no direct evidence	Unsupported as stated

What the 2024 documentation leak does and does not show

In 2024, internal documentation for Google’s Content Warehouse API was published, listing thousands of attributes. Several are often cited as proof of topical authority. Search Engine Land’s analysis describes siteFocusScore as denoting how much a site is focused on a specific topic, siteRadius as measuring how far page embeddings deviate from the site embedding, and siteEmbeddings as compressed site and page embeddings.

The same analysis is careful about the limits: its author writes that we do not know to what extent Google uses these scoring factors. An attribute in an API schema shows that something is stored, not how much it weighs in ranking, or whether it is used for ranking at all.

Topical authority, domain authority and E-E-A-T

Three terms get used interchangeably, and keeping them apart avoids building the wrong thing.

Term	What it means	Whose term it is
Topical authority	How reliable a site is treated as on one subject, observable in how widely it ranks across that subject	Industry term; Google uses topic authority for its news system
Domain authority and similar scores	A third-party estimate of a site’s overall link strength, across all subjects	SEO tool vendors; not a Google metric
E-E-A-T	Experience, expertise, authoritativeness and trustworthiness, used by Google’s quality raters to assess pages and creators	Google’s quality rater guidelines

The difference that matters in practice: a site can have a strong link profile overall and still be a weak source on a topic it has only touched, and a small site can win a narrow topic against much larger ones because it covers that topic better and is cited for it. Topical authority is subject by subject.

How to choose the topics you want authority in

Topical authority is earned one subject at a time, so the first decision is which subjects. Most sites try to be an authority on too many, and end up covering each thinly. Choose few, and choose them on four tests.

Test	The question	Why it matters
Business fit	Do people searching this topic become customers, or influence those who do?	Authority in a topic your buyers never search is traffic without a pipeline
Right to speak	Can the team say something first-hand: data, experience, a tool, a strong view?	Original evidence is what makes a source notable, in Google’s news system and in practice
Demand	Is the topic’s total search volume worth a year of work, once realistic click-through is applied?	A topic small enough to win completely may still be too small to matter
Winnable	What share do the leading sites hold, and are they specialists or generalists?	Generalist leaders with thin coverage are easier to overtake than specialists

Score each candidate topic against the four, and pick the two or three that pass all of them. Business fit and right to speak are the tests teams skip, and they are the ones that decide whether the authority is worth having and whether it can be earned. A topic that passes on demand alone produces the familiar pattern of a blog with traffic and no customers.

Define each chosen topic tightly enough to measure. ‘Marketing’ is not a topic; ‘B2B content briefs and editorial process’ is. The boundary should be clear enough that two people would put the same keyword inside or outside it, which is what makes topic share comparable from month to month.

Why publishing more pages does not build topical authority

The most common advice on topical authority is to publish more on the topic. Volume is the easiest input to control, so it gets mistaken for the mechanism. The documented signals point elsewhere.

Google’s helpful content guidance asks whether content provides original information, reporting, research or analysis, and whether it provides substantial value compared with other pages in search results. A page that restates what already ranks fails both questions however many siblings it has. Its spam policies go further, defining scaled content abuse as many pages generated for the primary purpose of manipulating search rankings and not helping users, with generating many pages using generative AI without adding value as the first example.

Volume also has a mechanical cost. Every new page on a topic is another page that can compete with an existing one for the same intent, another page for the internal links to support, and another page to keep current. Twenty pages that each answer a distinct question well build more than sixty that overlap.

How topical authority is built

Taken together, the documented signals point to four things: cover the topic fully, publish evidence others cite, keep the site focused, and make the coverage easy to understand as a whole. None of them is volume for its own sake.

1

Coverage

A page for each distinct question in the topic, one page per intent, so the site answers the subject rather than a few keywords in it.

2

Original evidence

Data, tools, worked examples and first-hand experience others reference. This is the closest equivalent to the original reporting signal in Google’s news system.

3

Focus

Subjects close to what the site is known for. Pages far outside them are a separate bet, not a contribution to the topic.

4

Structure

Clusters, internal links and clear entities, so that what the site covers is easy to see as a whole.

A synthesis of the documented signals above, not a published Google list

Coverage: map the topic before you write

Coverage starts with a topical map: every question in the subject, grouped by intent, with one URL assigned to each. The Topical Map Generator builds one from your own keyword list and maps it to your existing pages, so the gaps and the duplicates are visible before anything is written. Duplicates matter as much as gaps: two pages answering one intent split the signals both should have earned, which is the problem covered in the keyword cannibalization guide.

Original evidence: give people something to cite

In Google’s news system, original reporting is gauged partly by links and citations. Outside news, the equivalent is work others reference: your own data, a calculator or tool, a benchmark, a detailed worked example from real projects. One genuinely useful asset on a topic does more than ten pages restating what already ranks, because it earns the links and mentions that make a source notable.

Focus: decide what the site is about

Write down the three to five subjects the site should be known for and judge every new page against them. A page that sits outside them may still be worth publishing, for a campaign or a partner, but plan it as a standalone piece, not as part of the topic’s authority. A content audit is the moment to decide what to do with old pages that sit far outside those subjects.

Structure: make the topic legible

Clusters and internal links make a topic’s coverage visible as a whole: the pillar connects the parts, and every part links back. Entities help too: naming the organisations, products, people and concepts that define a subject, consistently, is how a page shows what it is about. The Entity Extractor lists the entities a page names and compares them with competing pages. The link rules are in the internal linking strategy guide.

How to measure topical authority: topic share

Topical authority has no score in Google Search Console or anywhere else, so measure the outcome it should produce: winning a large share of a topic’s demand. Kevin Indig proposed this as topic share: the share of a topic’s search traffic a site captures, compared with competitors, because no site can earn more than the topic’s total demand.

The method is simple to describe and needs care in the definition. Define the topic as a fixed keyword set, measure what share of its demand each site wins, and repeat monthly on the same set.

1. 1DefineList the topic’s keywords: a head term’s matching terms from a keyword tool, filtered to the ones that really belong to the topic.
2. 2FreezeSave the list. Changing it each month makes the trend meaningless.
3. 3MeasureIn a keyword tool that reports traffic share by domain, load the list and record each site’s share.
4. 4CheckIn Search Console, filter to the same queries and record your own clicks and the number of queries you rank for in the top ten.
5. 5CompareTrack your share against the two or three sites that lead the topic, monthly.

Kevin Indig’s examples show what the numbers look like: on the ecommerce topic, using the top 3,000 keywords by volume, he reported Shopify at 11%, BigCommerce at 10% and NerdWallet at 3%. Even leaders in a large topic hold a minority share, so the useful question is not whether you lead, but whether your share is rising on the topics you chose.

Two Search Console measures complement topic share and need no third-party tool: the number of distinct queries in the topic for which you rank in the top ten, and how quickly new pages in the topic start getting impressions. If authority is building, both should rise. The same article cites a Graphite study reporting that pages on sites with high topical authority gained traffic faster than those with low authority; that is one vendor’s correlational study, but it suggests the time it takes new pages to gain impressions is worth tracking.

Topic share calculator

Use the calculator to turn your numbers into the three measures above. Enter the size of the topic as you defined it, how much of it you rank for, your clicks from it, and the leader’s share from your keyword tool.

Hypothetical defaults. The click share in particular varies by topic; a topic with many AI Overviews or answer boxes will have a lower one.

Keywords in the topic, as you defined itKeywords you rank in the top ten forTotal monthly search volume of the topicShare of searches that end in a click on any result%Your monthly clicks from the topic’s queriesThe leading site’s traffic share%

Coverage: keywords in the top ten

15.0%

Monthly clicks available in the topic

30,000

Volume multiplied by the click share

Your topic share

4.0%

Gap to the leader, in percentage points

8.0

Negative means you lead

Clicks a month at the leader’s share

3,600

Your share here is measured against an estimate of the clicks available, so use it to track change over time on a fixed keyword set rather than as an exact figure. A keyword tool’s own traffic share is the better comparison with competitors.

Coverage and share move differently, and the difference is diagnostic. High coverage with low share means you rank widely but low, which usually means the pages exist but are not the best answers. Low coverage with a decent share means you win a few large queries and miss the rest of the topic, which is a gap the topical map should close.

Topical authority and AI search

AI search raises the value of being a recognised source on a topic. Google’s guide to AI features describes AI Overviews and AI Mode using query fan-out, running multiple related searches across subtopics to build a response. A site that ranks across a topic’s subtopics has more chances to be retrieved for those searches than one that ranks for its head term alone.

The same guide says there are no additional requirements to appear in AI Overviews or AI Mode. Topical authority helps in AI search for the same reasons it helps in classic search: wide, reliable coverage and evidence others cite. How AI systems choose which passages to use is covered in the guide to what AI search actually retrieves.

A worked example: choosing which topic to build

Take a hypothetical B2B software company whose blog covers six subjects loosely. The topic share numbers below are invented to show the method; the decision logic is the point.

Topic	Coverage	Topic share	Close to the product	Decision
Expense management	45%	6%	Yes	Build: ranks widely but low, so improve the existing pages
Spend analysis	12%	2%	Yes	Build next: little coverage, so map and fill the gaps
Procurement software	30%	4%	Yes	Maintain: competitive head terms, revisit after the first two
Remote work tips	8%	1%	No	Stop: far from the product, prune or leave as is
Accounting careers	5%	0.5%	No	Stop: no link to the product or its buyers
Company news	n/a	n/a	n/a	Keep, outside the topic model

Hypothetical figures to show the method

The company picks two topics close to its product, one where it already ranks widely but low, and one where it barely exists, and stops spending on the two that sit far from its buyers. Over the following year it tracks coverage and topic share on frozen keyword sets for the two chosen topics, rather than total blog traffic, which would hide whether either topic is improving.

Common topical authority mistakes

Most topical authority mistakes come from treating it as a volume target. Each item below has the check that catches it.

• Publishing many thin pages to cover a topic. Check: does each page answer a distinct intent better than what already ranks?
• Two pages for one intent. Check: does the topical map have one URL per intent?
• Topics chosen for volume alone. Check: is the topic close to what the business sells and what the team can speak to?
• No original evidence. Check: is there at least one asset in the topic that others would cite?
• Strategy built on leaked attribute names. Check: would the plan still make sense if those attributes are unused?
• Measuring total traffic instead of the topic. Check: is topic share tracked on a frozen keyword set?
• Expecting it quickly. Check: is the review horizon a year, not a quarter?

Choose

Three to five topics written down, each close to the product and its buyers Each topic defined as a frozen keyword set Pages far outside the topics identified and treated separately

Build

A topical map with one URL per intent for each topic Duplicates merged before new pages are written At least one original asset per topic: data, a tool or a detailed worked example Each topic structured as a cluster with descriptive internal links

Measure

Topic share recorded monthly against the leading sites Top-ten coverage tracked in Search Console for each topic Time to first impressions for new pages in the topic recorded

Ticks are saved in your browser.

Building topical authority across a site, with a topic map, page templates and measurement the team runs every month, is the core of an SEO systems engagement. It also fits the argument of the SEO audit guide: audits fix defects, but authority comes from the system that decides what gets published.