What topical authority really is, and how to build it for Google and AI search
Topical authority is the SEO name for being the site search engines and AI assistants trust on one subject because you cover it in depth. Google has no general score by that name, but it asks whether a site has a clear focus and runs a news system called topic authority. You build it with a topic map, clusters, internal links and updates.
Few SEO terms are sold harder or defined more loosely. Some guides present topical authority as a hidden Google score you can pump up by publishing two hundred articles in a month; others wave it away as a buzzword. The useful version sits in between, and it is mostly about how you plan content, not about a number.
This guide covers what Google has and has not said, with dates, then how to build a topic map, using a fictional lawn care company as the worked example. It ends with how to measure progress in Search Console and what 12 ChatGPT answers showed about which sites get cited again and again on one subject.
What topical authority means
Topical authority describes a site that search engines treat as a dependable source on a whole subject rather than on a single keyword. You see it when pages from the same site rank, and get cited, across many of the different questions people ask about that subject.
It is an outcome, not a lever. You do not switch it on; you earn it by covering a subject thoroughly, organizing that coverage so the pages support each other, and showing that real expertise sits behind it. Most of the vocabulary around it comes from content planning:
- Topic map
- The full list of questions and subtopics people have about your subject, grouped by intent. It is the plan everything else follows.
- Pillar page
- A broad page that answers the main question of a subject and links out to the detail.
- Cluster page
- A focused page that answers one specific question in depth and links back to its pillar.
- Topic clusters
- A pillar plus its cluster pages, tied together with internal links. Some teams call it a hub-and-spoke model; the shape is the same.
- Content gap
- A question on your topic map with no page that answers it well, on your site or anywhere else.
What Google has actually said about topical authority
Google does not name “topical authority” as a ranking factor for websites in general. It does run a system called topic authority, but the public description is about news. Here is the record, oldest first.
| Date | Source | What it says |
|---|---|---|
| May 23, 2023 | Google Search Central blog, “Understanding news topic authority” | A system that helps surface expert sources for newsy queries in areas such as health, politics or finance. Signals: how notable a source is for a topic or place, original reporting that others cite, and reputation, including awards and recommendations from professional societies. |
| Nov 13, 2023 | Search Engine Journal, on a Reddit thread | John Mueller endorsed a comment calling topical authority a fancy SEO term and told the poster not to worry about it. |
| May 28 and 29, 2024 | Leaked API documentation analyzed by iPullRank; Google’s reply to The Verge | The documents name a siteFocusScore (how focused a site is on one topic) and a siteRadius (how far its pages stray from that focus). Google cautioned against assumptions drawn from out-of-context, outdated or incomplete information. |
| Dec 10, 2025 (last update) | Google’s guide to its ranking systems | Lists systems for reliable information, reviews, original content and site diversity, among others. None is a general topical authority system. |
| Jul 10, 2026 (last update) | Google’s guide to its generative AI features | A large number of pages does not make a site higher quality or more relevant. Pages written for each variation of a query, mainly to manipulate results, breach its scaled content policy. |
| Oct 5, 2026 (last update) | Google’s guide to helpful, people-first content | Asks whether your site has a primary purpose or focus, and warns against producing lots of content on many topics in the hope that some of it performs. |
Sources linked at the end of the article. Google confirmed the leaked documents were real, but not whether those attributes are current or used in ranking.
Read together, the record says less than most guides claim and more than the skeptics allow. Google’s news topic authority post shows it can judge expertise by subject and by place. Its helpful content guidance rewards a clear focus and real experience. And if the leaked attributes are still in use, something measures how tightly a site sticks to its subject.
What Google rejects is the volume version of the idea. Mueller’s “don’t worry about it”, reported by Search Engine Journal, was aimed at the theory that linking a batch of articles together earns a badge. It was not a claim that expertise on a subject is irrelevant.
Why subject depth matters for AI search citations
AI assistants break a question into several searches and then choose sources they trust for that subject. A site that covers the subject in depth has more chances to be retrieved for each of those searches. An engine that associates a source with a topic will also go looking for it by name, as our check below shows.
Google’s own example of this, from its July 2026 guide, happens to be about lawns. A question like “how to fix a lawn that’s full of weeds” can fan out into searches on herbicides, chemical-free weeding and prevention. A site with a solid page for each of those angles is a candidate three times over; a site with one thin page on weeds is a candidate once, if at all. Our explainer on query fan-out shows how those background searches work in ChatGPT too.
There is a limit worth knowing. In classic results, Google’s site diversity system generally shows no more than two listings from the same site at the top. Depth does not let one site take over a results page. In AI answers the rule is different, and the next section shows what that looks like.
Does ChatGPT cite several pages from one site? What 12 answers showed
On October 27, 2026, we asked ChatGPT 12 questions from a US location: six about lawn care and six about Roth IRAs, two subjects with well-known authorities. Then we counted which sites it cited and how many different pages it took from each.
| Measure | Lawn care (6 questions) | Roth IRA (6 questions) |
|---|---|---|
| Answers that ran a web search | 4 of 6 | 6 of 6 |
| Citations in total | 15 | 21 |
| Who got cited | University extension services and one county government: all 15 citations on .edu or .gov sites | irs.gov took 16 of the 21 citations |
| Pages from the leading source | 11 institutions; only Iowa State appeared in two answers | 9 different IRS pages; Publication 590-A in 5 of 6 answers |
| Answers citing 2 or more pages from one source | 2 of 4 (Virginia Tech, Texas A&M) | 6 of 6, all from the IRS |
| Commercial sites cited | None | Fidelity in 2 answers; Vanguard, Kiplinger and one smaller finance site in 1 each |
| Background searches naming the authority | All 4 used the word “extension” or named an extension service | All 6 began with “IRS” |
Own check: 12 questions to ChatGPT from the US, October 27, 2026 (data/blog/pruebas/topical-authority.json).
Our own check, run on October 27, 2026. AI answers change over time, so the same question may get a different answer today.
The Roth IRA answers are topical authority in its purest form. ChatGPT wrote “IRS” into every background search, cited the IRS in every answer and usually pulled two to four different IRS pages into the same reply: rules, limits, forms and publications. One source owns the subject so clearly that the engine asks for it by name.
The lawn answers show a second form, where authority belongs to a network rather than a domain. ChatGPT went to land-grant university extension services and picked the one that matched the situation: Virginia Tech and Fairfax County for fall fertilizing (the request appeared to come from Northern Virginia), Texas A&M for aeration. No lawn care company, retailer or fertilizer brand was cited, although a Lowe’s page did turn up once among the pages it read.
Commercial sites can break in, but it is hard next to a primary source. Fidelity’s education pages were retrieved in five of the six Roth searches and cited in two, with a different article each time. Vanguard was retrieved in five and cited once.
Two lawn questions, on watering new seed and dethatching, got full answers with no search at all, straight from what the model already knew. For basic how-to topics, being part of the long-standing written record matters as much as being searchable today. Our guide to where ChatGPT gets its information explains that split between training data and live search.
The lesson for a business is not to out-write the IRS on tax law or a university on turfgrass science. Find the slice of the subject where you are the primary source (your prices, your local conditions, your cases, your process) and cover it better than anyone. That is what a topic map is for.
How to build topical authority, step by step
Building topical authority is planning first and publishing second. These eight steps work for a local service company as well as for a software brand; only the size of the map changes.
- Pick a subject you can defend. Narrow enough to cover completely, close enough to what you sell that the traffic is worth having. “Lawn care in central Ohio” beats “gardening”.
- Map the questions. Pull them from Search Console, “People also ask” boxes, sales and support conversations, forums and the follow-up searches AI tools run. Write each one in the customer’s words.
- Group them into clusters. One pillar per major theme, one cluster page per distinct intent. Two questions with the same answer belong on the same page.
- Audit what you already have. Match existing URLs to the map. Merge near-duplicates, redirect the losers, and mark thin pages for a rewrite before you write anything new.
- Write where you have experience. First-hand detail, your own data, named authors with real credentials. Google’s helpful content guidance asks specifically for expertise that comes from having done the thing.
- Link the cluster. Pillar to every cluster page and back; each cluster page to two or three siblings, in context, with descriptive anchor text.
- Update on a schedule. Prices, seasons, regulations and product details go stale. Put a review date on every page and keep it.
- Earn outside mentions on the subject. Reviews, trade press, associations and lists that tie your name to the topic. Our 90-day GEO strategy sets out that off-site work week by week.
Depth or breadth?
Depth first, almost always. A site that answers 40 questions about one subject well is in a stronger position than one with a single page on each of 40 subjects. Breadth comes later, by adding neighboring topics your customers also ask about: a lawn company moving into irrigation makes sense, moving into home insurance does not. Google’s helpful content guidance warns against entering a niche without real expertise just because it looks like traffic.
Internal links that do real work
Google’s link documentation says every page you care about should have a link from at least one other page on your site, and that there is no magic number of links per page. In a cluster, that means no orphan pages, a pillar that links to everything below it, and anchors that tell the reader what they will get (“when to overseed in central Ohio”, not “click here”). Keep each cluster in its own folder, such as /lawn-care/, so you can measure it later.
Worked example: a topic map for a lawn care company
Imagine a lawn care company in Columbus, Ohio, that we will call Fescue & Co. It is fictional, built only to show the structure: cool-season lawns, heavy clay soil and customers who mostly hire for fall work and weed control.
| Cluster (pillar) | Cluster pages | Proof only this company has | Leads to |
|---|---|---|---|
| Fall lawn care in central Ohio | When to fertilize in fall; overseeding dates; mulching or bagging leaves; is winterizer worth it | Seeding dates and germination notes from its own job log | Fall aeration and overseeding package |
| Weeds | Preventing crabgrass; dandelions; clover, kill or keep; weed control without chemicals | Before and after photos from client lawns, with dates | Weed control plan |
| Clay soil | Core aeration in clay; topdressing; why water pools; how to read a soil test | Aggregated soil test results from its customers | Aeration service |
| New seed and watering | How often to water new seed; summer watering; signs of overwatering | What worked on last year’s seeding jobs | Seeding service |
| Prices and hiring | What lawn care costs in Columbus; aeration cost; doing it yourself or hiring; how to choose a company; contracts explained | Real price ranges from its invoices | Quote request |
Fictional example by Vértigo Marketing. The structure is the point; the company does not exist.
Three choices make this map work. The weeds cluster mirrors the follow-up searches Google itself uses as an example, so each page has a real query behind it. The science pages link out to Ohio State’s extension service instead of trying to replace it, which costs nothing and borrows credibility. And the prices and hiring cluster is where the company is the primary source, the one slice no university or retailer can cover with first-hand numbers.
Notice what is missing: forty suburb pages with the town name swapped. If service genuinely differs in one suburb (different HOA rules, a different soil), that suburb earns a page. If not, a single service-area page does the job, and the copies would only invite a scaled content problem.
How to measure topical authority
You cannot read a topical authority score anywhere, because Google does not publish one and the scores in SEO tools are their own estimates. Measure coverage and performance by cluster instead.
- Search Console by folder. In the Performance report, add a Pages filter with “URLs containing” /lawn-care/, or use “Custom (regex)” for several folders at once. Compare the last three months with the three before, and with the same period last year.
- Query coverage. With the folder filter on, count the distinct queries that bring impressions and how many of them reach the first page. A cluster that is gaining authority answers more queries over time, not only more clicks on the same ones.
- Share of your question list. Take the questions on your topic map and mark which ones have a page, which rank and which get cited in AI answers. It is the most honest progress bar you will find.
- AI visibility. Filter the generative AI performance report in Search Console to the cluster’s pages to see impressions in AI Overviews and AI Mode. For ChatGPT and Perplexity, ask the cluster’s questions and count how many answers cite any of your pages, and how many different pages.
- Internal links. Crawl the site and check that every cluster page receives links from its pillar and from its siblings, with no orphans.
Pages filter > Custom (regex) > Matches regex
^https://www\.example\.com/(lawn-care|weeds|clay-soil)/Search Console uses RE2 syntax. Replace the domain and folders with your own.
One caveat from Google’s help pages: filtering by query or URL can change the report totals, because anonymized queries drop out. Compare filtered numbers with filtered numbers, never with the unfiltered total. For the AI side, a sampler such as Slibur.com’s free AI visibility checker saves typing the same questions every month.
Mistakes that only look like authority building
Most failed topical authority projects fail in one of these ways. Each one feels productive while it is happening.
- Publishing for volume
- Google says outright that a large number of pages does not make a site better or more relevant. A hundred quick articles generated in a month look like a content farm, and its spam policies treat them like one.
- One page per keyword variation
- “Lawn aeration cost”, “cost of lawn aeration” and “how much is aeration” are one question. Three pages split the signals and compete with each other.
- Drifting off-topic
- A lawn company writing about crypto taxes dilutes its focus. Renting space on a strong site outside your field to rank there is worse: Google calls it site reputation abuse.
- Linking everything to everything
- Google’s view on internal links is blunt: if you think it’s too much, it probably is. Link where a reader would want to go next.
- Chasing a tool’s score
- Third-party authority scores can help you compare sites, but they are not what Google or ChatGPT read. A rising score with flat Search Console data is a vanity metric.
- Never revisiting old pages
- An outdated price or a rule that changed last year undercuts a whole cluster. Freshness is one of the ranking systems Google does document.
If the cluster work feels like a lot to run alongside everything else, our SEO service for US companies builds and maintains topic maps like the one above.
Frequently asked questions
Is topical authority a Google ranking factor?
Not by that name. Google runs a topic authority system for news and asks whether a site has a clear focus and real expertise, but it has never listed a general topical authority factor. In 2023 John Mueller told a Reddit user asking about it not to worry about the term.
What are topic clusters?
Topic clusters are groups of pages on one subject: a broad pillar page that answers the main question and several cluster pages that each answer one specific question, all linked together. They are a way to organize coverage so readers and search engines can follow it.
How many articles do I need for topical authority?
There is no magic number. Write one page for each distinct question on your topic map that deserves its own answer, and stop when the questions run out. For a local service business that may be 20 pages; for a software category it may be hundreds.
How long does it take to build topical authority?
Google gives no timeline. Pages have to be written, crawled and indexed before they can perform, so judge a cluster by quarters, comparing its folder in Search Console with the previous period and with the same months last year.
What is the difference between topical authority and domain authority?
Domain authority scores, such as Moz’s Domain Authority or Ahrefs’ Domain Rating, are link-based metrics created by SEO tool companies, not by Google. Topical authority describes how fully and credibly a site covers one subject. A small site can be strong on one topic with modest link metrics.
Does topical authority help in ChatGPT and other AI search?
Our check suggests it does. For six Roth IRA questions, ChatGPT searched for the IRS by name and cited it in every answer, often with several different IRS pages. Engines that split questions into many searches also give a site with deep coverage more chances to be retrieved.
Can a small business build topical authority?
Yes, by choosing a narrow subject it knows first-hand, usually its service in its area. A lawn company will not outrank a university on soil science, but it can be the best source on what lawn care costs and works in its own town.
Keep reading
- How to build a GEO strategy: a 90-day plan with owners and deliverables
- Query fan-out: the hidden searches behind AI answers, and how to match them
- Where does ChatGPT get its information? Training data, live search and the sites it cites
- Entity SEO for LLMs: how to make language models know, find and cite your brand
- GEO agency services in the United States
- AEO and GEO tools, compared
Sources
Figures and claims in this guide come from these sources:
- Google Search Central blog: Understanding news topic authority (May 23, 2023)
- Google: A guide to Google Search ranking systems (updated Dec 10, 2025)
- Google: Creating helpful, reliable, people-first content (updated Oct 5, 2026)
- Google: Optimizing your website for generative AI features (updated Jul 10, 2026)
- Google: Link best practices (updated Dec 10, 2025)
- Search Console Help: Advanced filtering and comparison
- Google Search Central blog: Search generative AI performance reports (Jun 3, 2026)
- Search Engine Journal: Google on topical authority (Nov 13, 2023)
- iPullRank: Google’s internal documentation leak (May 28, 2024)
- The Verge: Google confirms the leaked Search documents (May 29, 2024)
By the Vértigo Marketing Published Reviewed