Most people who say "we need topical authority" mean "we need to publish a lot of articles about our subject". That is not what it is. Topical authority is a condition, not a tactic: it is the state where a search engine has enough evidence, across enough of your pages, to treat your site as a reliable source for a whole subject rather than for isolated queries. You can publish two hundred articles and never reach it. You can reach it with forty, if the forty are the right forty and they are connected properly.
This guide is about the mechanism: why coverage matters, why depth matters more than people think, why publishing order changes the outcome, and how to tell from Search Console whether you are building authority or just a backlog.
What topical authority is, and what it is not
Search engines do not rank sites. They rank documents, and they use site-level signals to decide how much to trust each document. When your site has many documents that all deal with the same subject, link to each other coherently, and consistently satisfy the people who land on them, the engine accumulates evidence that your site is a source for that subject. New pages on the same subject then start with a head start. That head start is what practitioners call topical authority.
It helps to be precise about what it is not:
- It is not word count. A 4,000-word article on a subtopic nobody searches for adds almost nothing.
- It is not a number of articles. Fifty pages that each answer the same three questions slightly differently are one page's worth of evidence, plus a cannibalisation problem.
- It is not domain authority. Third-party "DA" scores measure links, not subject coverage. A site with strong links can still have zero authority on a topic it has never written about.
What it is, operationally, is the product of three things: how much of the subject you cover, how deeply each part is handled, and how clearly your site signals that the parts belong together. I think of those as coverage, depth and structure, and the rest of this guide follows that order.
Coverage versus depth: the two axes
Coverage is the breadth of the subject you address. Depth is how completely you handle each part. Most sites are unbalanced on one axis.
A site with coverage but no depth has a page for every query in the keyword tool and none of them say anything a reader could not get from the first result already. These sites get impressions and poor click-through, and they rarely hold positions. A site with depth but no coverage has three superb long-form guides and nothing around them. Those guides can rank, but the site never earns the subject-level trust that would make the fourth guide easy.
The useful question for each candidate page is not "is there search volume?" but "does this page add a distinct part of the subject that a serious reader would expect a complete source to have?" If the answer is no, the page is cannibalising something you already have, or it is filler.
| Pattern | What it looks like | What happens in search | Fix |
|---|---|---|---|
| Coverage, no depth | Hundreds of 600-word posts, one per keyword | Wide impressions, low CTR, positions drift between 8 and 20 | Consolidate into fewer, complete pages; redirect the remainder |
| Depth, no coverage | A few excellent guides, nothing supporting them | Guides rank for head terms; the site cannot extend into adjacent terms | Build the supporting layer the guides implicitly rely on |
| Coverage and depth, no structure | Good pages that do not link to each other | Each page performs alone; no compounding | Hub pages, cluster linking, consistent navigation |
| All three | Mapped subject, complete pages, deliberate links | New pages index and rank quickly; query coverage widens on its own | Maintain; extend to adjacent subjects |
Building a topic map
A topic map is the document that defines coverage before you write anything. It is the single most important artefact in a content programme and the one most teams skip. Without it, every article is an isolated decision, and isolated decisions produce an isolated site.
My approach is to map the subject the way a knowledgeable person would explain it, and only then check it against search data. If you start from the keyword tool you will reproduce whatever your competitors already published, including their gaps.
Define the core subject and its boundary
Write one sentence describing what the site is a source for. "Everything about home loans in India" is a subject. "Finance" is not; it is a category too broad to own. The boundary matters as much as the centre, because it tells you what to say no to.
Break the subject into its natural parts
List the sub-subjects a complete treatment would need. For home loans: eligibility, interest rates and how they are set, documentation, the application process, tax treatment, prepayment and foreclosure, balance transfer, and the lender landscape. These become clusters. Use search data, forums, support tickets and sales calls to find parts you missed.
Identify the pages each cluster needs
Within each cluster, list the distinct questions and tasks. Group queries that one page should answer together. The output is a set of page briefs, each with a primary intent and the related queries it also covers. My guide to keyword mapping covers how to make these groupings reliably.
Assign each page a role
Hub, cluster page, or supporting page. Hubs explain a whole cluster and link out to its pages. Cluster pages handle one distinct part completely. Supporting pages answer narrow questions that would bloat a cluster page. The roles determine internal linking later.
Audit what already exists
Map existing URLs onto the plan. You will find pages that fit, pages that overlap each other, and pages that belong nowhere. Decide for each: keep, rewrite, merge, or remove. This step usually reduces the amount of new writing needed by a surprising margin.
Source context: why the same page performs differently on different sites
A page about "how to calculate home loan EMI" on a bank's site, a personal finance publisher, and a property listings portal are three different documents in the engine's eyes, even if the words are identical. The site's existing subject, its other pages, its entity associations, and the context of the links pointing in all shape how a new page is interpreted. I call this source context.
The practical consequence is that you should build authority outward from what your site already is. A payroll software company trying to become a source on income tax law will struggle, not because the writing is bad, but because the site has no contextual claim to it. A payroll software company becoming the source on payroll compliance, TDS on salary, and Form 16 is a natural extension. The second path is slower to describe and faster to achieve.
Internal linking patterns that build authority
Internal links are how you tell the engine which pages belong together and which page is the primary one for a sub-subject. I have a full guide to internal linking and a framework for doing it systematically; here I will cover only the patterns that matter for topical authority specifically.
Hub to cluster, cluster to hub
Every cluster page links up to its hub and the hub links down to every cluster page. This is the minimum. It sounds obvious; most sites do it on the hub and forget it on the cluster pages, which leaves the hub with no inbound links from the content it is supposed to represent.
Sibling links with intent
Cluster pages should link to their siblings where the reader would naturally need them, in the body, with descriptive anchor text. A page on prepayment should link to the page on foreclosure charges at the sentence where the two concepts meet. A "related posts" widget at the bottom does not do the same job; it is templated, often rotated, and carries little meaning.
- Supporting page
- Cluster page
- Hub
- Adjacent hub
Publishing sequence
The order you publish in affects how quickly the engine recognises the cluster. This is the part most content calendars ignore, because they are organised by month and by author availability rather than by structure.
The sequence I use:
- Hub first, even if it is incomplete. The hub establishes the subject and gives every subsequent page a parent to link to. Publish it with links to the cluster pages that exist and update it as each new one goes live.
- Then the cluster pages for the core, highest-intent parts of the subject. These are the pages that would make a reader say "this site clearly knows this". They are also the ones most likely to earn links.
- Then the cluster pages that fill the obvious gaps. If your map shows eight parts and you have covered four, the next four matter more than any supporting page.
- Supporting pages last, and only the ones the data justifies. Once the cluster is live, Search Console will show you which narrow queries your cluster pages are already appearing for without ranking well. Those are the supporting pages worth writing.
How to measure topical authority
You cannot measure topical authority directly, but you can measure the behaviour it produces. The signals I rely on are all available in Search Console, and I would rather track three of them honestly than report a "topical authority score" from a tool that has invented one.
Query coverage
Count the number of distinct queries for which the cluster's pages receive impressions. Filter Search Console to the cluster's URL path (or a regex of its URLs) and compare the query count month over month. A growing cluster with real authority widens its query footprint faster than it adds pages. If you publish three pages and gain thirty queries, something is working. If you publish thirty pages and gain thirty queries, you are only ranking for your exact targets and the engine has not generalised.
Impressions breadth and its shape
Look at the distribution of impressions across queries, not the total. Authority shows up as a long tail of queries you never targeted appearing on cluster pages: the engine testing your pages against adjacent intents. When the test succeeds, positions on those queries improve without you touching the page.
Time to first impressions for new pages
Record, for each new page in a cluster, how many days pass between publication and first impressions in Search Console, and between first impressions and first page-one appearance. As authority builds, both intervals shrink. This is the clearest leading indicator I know, and it is the one I use in my cluster test in the SEO Lab.
When topical authority fails
I want to be specific about failure, because the generic advice makes it sound like the only way to fail is to not publish enough.
The volume trap
A team decides on "fifty articles in three months", writes briefs from a keyword export, and publishes. Half the articles overlap. The hub is written last or not at all. The site ends up with more pages competing for the same queries, which is the opposite of authority. I wrote about why this happens in a note on publishing volume.
Cannibalisation disguised as coverage
"Home loan eligibility", "home loan eligibility criteria", "who is eligible for a home loan" and "home loan eligibility calculator" look like four pages in a keyword tool. They are one page, perhaps two. Four pages means the engine has to pick one, and it will pick a different one each week.
Technical blockers
Authority cannot build on pages that are not crawled, not indexed, or rendered inconsistently. On more than one audit the "content problem" turned out to be a crawl and indexing problem: cluster pages were orphaned, blocked by a parameter rule, or sitting behind a JavaScript router the crawler did not follow. Check the technical layer before blaming the writing.
Authority is not what you publish. It is what the engine has concluded about you from what you published, how it connects, and whether it held up.
Frequently asked questions
How many pages do I need before topical authority starts to show?
There is no fixed number. What matters is whether the pages cover the subject's natural parts completely and are linked coherently. I have seen the signals appear with a hub and under a dozen cluster pages on a narrow subject, and fail to appear with hundreds on a broad one. Map the subject first; the page count falls out of the map.
How long does it take to build topical authority?
Content programmes take months, not weeks. The first leading indicators, such as new pages getting impressions faster and untargeted queries appearing, typically show within a few months of a well-structured cluster going live. Holding positions through core updates takes longer. If someone is promising a timeline in weeks, they are describing something other than authority.
Should I delete old content that does not fit the topic map?
Not reflexively. Check impressions, links and conversions first. Pages that fit nowhere, get nothing and overlap others are candidates for merging and redirecting. Pages that get traffic but sit outside the map may mean the boundary is wrong.
Can a new site build topical authority against established competitors?
Yes, on a narrower subject than the incumbents. A new site that tries to own "insurance" will lose; one that becomes the most complete source on cyber insurance for Indian SMEs can win that and extend from it. Narrow, complete, structured, then expand.
Is topical authority relevant for AI search and answer engines?
More so, if anything. Systems that generate answers retrieve passages from sources they treat as reliable on a subject, and a coherent, well-linked cluster is easier for those systems to retrieve and cite than scattered posts. I test this directly in the AI search citation experiment. The fundamentals have not changed; the number of places they matter has grown.
Where this connects
This guide is the thinking; the Topical Authority Framework is the step-by-step version I use on engagements, with the topic-map template and the sequencing rules. The cluster test in the SEO Lab is where I measure time-to-impression and query growth on a controlled cluster. The technical SEO and content architecture case study shows what happens when structure and coverage are fixed together. If you want this built for your site rather than explained, that is what my content SEO service does.