TL;DR: A topical authority map plans every page your site needs to cover a subject properly and how those pages connect. It earns rankings and AI citations because search systems reward concentrated coverage and split single questions into several related searches. Most guides treat the map as a publishing list, but pages that stray off your core subject actively weaken the site, so it should tell you what to cut as clearly as what to write.
Ask most tools for a topical authority map and what comes back is a list of pages to write. Sixty articles, eighty, sometimes more, sorted into tidy clusters and handed over as a content plan for the year.
Which would be fine if volume were the thing that mattered. Google measures how tightly a site sticks to its core subject and how far each individual page drifts from it, and both of those measurements were named in the 2024 leak of its internal search documentation. A page that sits outside your subject can leave the whole site looking less focused than it did before you published it.
That makes the map a two-sided document rather than a production quota, because deciding what to remove moves the same signal as deciding what to add, and it usually costs a good deal less.
This piece covers what a topical authority map is, why it matters more now that AI systems answer a question by running several searches at once, how to build one, what to cut, which cluster to build first, and how to tell whether any of it is working.
What is a topical authority map?
A topical authority map is a planning document that sets out every page a website needs in order to cover a subject completely, grouped into clusters, with the internal links between those pages defined in advance. It exists to build topical authority, which is the credibility a site earns when search engines and AI systems treat it as a reliable source on a specific subject.
The credibility comes from coverage rather than from size. Topical authority is judged subject by subject, so a small specialist site can outrank a much larger one on the ground it genuinely owns, and a large site with scattered interests can lose to it.
That is worth separating from domain authority, which is a third-party estimate of a site’s overall backlink strength produced by tools such as Moz and Ahrefs. Google doesn’t use it as a ranking factor. It tells you roughly how strong a site’s link profile looks, and nothing at all about whether that site knows anything about your subject.
Topical authority map vs topic cluster
These get used as if they mean the same thing. They don’t, and the difference matters when you’re scoping the work.
| Term | What it is | Scope |
|---|---|---|
| Topical authority map | The plan covering every cluster, page and link needed to own a subject | The whole site, or a whole subject area |
| Topic cluster | A group of interlinked pages covering one subtopic from different angles | One subtopic |
| Pillar page | The central page of a cluster, covering its subtopic broadly and linking to every supporting page beneath it | One page |
A cluster is a component. The map is the architecture that decides how many clusters you need, which ones come first, and which pages you already have that belong in them.
Which raises the obvious question of why any of this deserves budget in 2026, when it has been standard content strategy advice for the better part of a decade.
Why topical authority matters more in AI search
Topical authority matters more in AI search because AI systems break a single question into several related searches before answering it, so a site covering a subject from many angles turns up in more of them. Google documents the behaviour, and the strongest independent analysis of AI citation ranks coverage of those searches near the top of what works.
The mechanism has a name. Query fan-out is a set of concurrent related searches that a model generates from one user question in order to gather enough material to answer it, and Google describes it in its own guidance with a worked example: ask how to fix a lawn full of weeds and the system may also search for the best herbicides, how to remove weeds without chemicals, and how to stop them coming back. Three searches you never typed, all resolved before you see an answer. We’ve written separately about how AI systems expand a single prompt into multiple searches if you want the detail.
A site with one page on the subject can satisfy one of those searches. A site with a properly built cluster can satisfy all four.
There is now decent evidence for how much that matters. Cyrus Shepard scored 23 factors associated with earning AI citations across 54 experiments, patents and case studies, weighting each one by how well the evidence actually holds up (Zyppy Signal, May 2026). The top five:
- URL accessibility, 9.5 out of 10
- Traditional search rank, 9.4
- Fan-out rank, 9.3
- Preview control, 9.2
- Query-answer match, 9.2
At the other end of the same list sits llms.txt, scoring 2.0, with no credible evidence that it does anything. Worth remembering next time someone tries to sell you one.
Two things follow from that ranking. The sources chosen for AI Overviews are not always drawn from the top ten organic results, so ranking well is no guarantee of being cited. But traditional search rank sits second on the list, which means most of the work that earns strong organic rankings is the same work that earns AI Overview visibility. This is part of why the question of whether traditional search still deserves the budget keeps getting answered badly.
And the citation is worth having. Across 53 brands and 5.47 million queries tracked between January 2025 and February 2026, brands cited inside an AI Overview earned roughly 120% more organic clicks per impression than brands that weren’t cited on the same query (Seer Interactive, April 2026). Same search, same panel, twice the traffic.
Google also measures how concentrated your site is on its core subject. That measure is called the site focus score, and its companion, site radius, tracks how far each individual page drifts from that centre. Both were named in the 2024 leak of internal search documentation, and together they explain why a map has to be a two-sided document. More on that shortly.
How to build a topical authority map
Building a topical authority map takes six steps: define the one subject the site should own, expand it into the questions people actually ask about it, group those questions into clusters, score each cluster on commercial value, map your existing pages onto the clusters, then set a publication order that finishes one cluster before starting the next.
Most of this is straightforward. Two steps get skipped often enough to be worth flagging, and they’re steps five and six.
Step 1: pick one subject you can credibly own
Pick something narrow enough to finish and broad enough to have real subtopics beneath it. “Property marketing” works. “Marketing” doesn’t, and neither does “property marketing email subject lines”.
Be honest about whether you can win it. If your technical foundations are weak, a map won’t fix that, and the SEO work has to come first.
Step 2: expand the subject into real questions
Pull every question people ask around the subject, not just the head terms. Keyword tools, People Also Ask boxes, your own sales calls, support tickets, and the queries in Search Console that nobody has looked at in a year.
Sales calls tend to produce the best material here, because they surface the questions people ask a week before they buy rather than a month before.
Step 3: group questions into clusters
Group by what the searcher wants, not by wording. Twelve phrasings of the same question belong on one page.
Each cluster gets a pillar page and a set of supporting pages beneath it. If two subtopics are closely related and neither has enough substance to stand alone, merge them rather than publishing two thin pages.
Step 4: score each cluster on commercial value
This is where most maps go wrong, because search volume gets used as the score. Volume tells you how many people are asking. It says nothing about whether any of them will buy.
The scoring method matters enough that it gets its own section further down.
Step 5: map what you already have
Before commissioning a single new page, put every existing page onto the map. Most sites have more relevant content than they think, sitting in the wrong place, poorly linked, or duplicated across three near-identical posts.
Run a content gap analysis, which means comparing the subjects you cover against those your competitors cover to find what’s missing. Then run the same exercise inwards. You’ll usually find that a third of the plan is already written and just needs consolidating.
Step 6: sequence publication cluster by cluster
Finish one cluster before starting the next. A half-built cluster signals nothing.
Publishing across four clusters at once feels productive and delivers a site that covers four subjects shallowly. One completed cluster beats four partial ones, every time.
Before you commission anything from that list, though, there’s a question worth asking about the pages already on your site.
The half of the map most teams skip: what to cut
A topical authority map should identify pages to remove, merge or rewrite alongside the pages to create. Off-topic content carries a real cost, because search systems assess how concentrated a site is on its core subject and how far each page drifts from it, which means publishing more can weaken a site rather than strengthen it.
Three kinds of page tend to show up in an audit, and each has a different fix:
Pages genuinely outside your subject, usually company news, event write-ups, or a subject the business moved on from three years ago. Redirect them if they’ve earned links, remove them if they haven’t.
Thin pages written to fill a keyword gap, which cover a subject that belongs on the map but cover it badly. Rewrite these into the cluster properly rather than starting again.
Near-duplicates competing with each other, which is cannibalisation, where two or more pages on the same site chase the same query and split the signal so neither ranks as well as one page would. Merge them into the strongest version and redirect the rest.
The commercial argument for doing this first is simple enough: consolidating what you have is a fraction of the cost of commissioning what you don’t, and it moves the same measurement in the same direction.
There’s a risk in the opposite approach too. Google’s guidance on generative AI features states that creating separate content for every possible search variation, including fan-out queries, primarily to influence rankings or AI responses, falls under its scaled content abuse policy, and adds that you don’t need to worry about capturing every long-tail variation. Which is worth holding onto, because a tool that hands you eighty page titles is describing exactly the behaviour that policy names.
So the map gets shorter. The next question is what order to build what’s left.
How to decide which cluster to build first
Prioritise clusters on three factors rather than search volume alone: how much revenue the subject connects to, how far behind your competitors you currently are on it, and how many related sub-queries it generates. A cluster with modest volume that sits close to a buying decision is usually worth building before a high-volume cluster that doesn’t.
Take a marketing agency with two candidate clusters. The first covers “what is SEO” and its variations, worth a few thousand searches a month. The second covers how to brief an agency, what retainers cost, and how to compare proposals, worth a few hundred.
The first cluster brings in students and junior marketers. The second brings in people six weeks from signing a contract. Score on volume and you build the wrong one, then spend nine months explaining to the board why traffic is up and the pipeline isn’t.
Here’s the weighting we’d apply:
| Factor | Weight | What it measures | Where the data comes from |
|---|---|---|---|
| Commercial proximity | 50% | How close the subject sits to a buying decision | Judgement, informed by your CRM or sales team |
| Coverage gap | 30% | How far behind competitors you are on the subject | Content gap analysis in Ahrefs or Semrush |
| Fan-out breadth | 20% | How many distinct sub-questions the subject generates | Keyword tools with a questions filter, or observing fan-out directly |
Commercial proximity takes half the weight on purpose, because the failure this is designed to prevent is a twelve-month programme that builds the biggest cluster first and shows no revenue by month nine.
Score each cluster out of five on all three, apply the weights, and build in descending order. The numbers are directional rather than precise, and their real value is that they force an argument about commercial proximity before anyone commissions anything, which is a conversation most content plans never have.
How to measure topical authority
Measure topical authority at cluster level rather than page level, using three signals: your share of total traffic across every keyword in a subject, the variety of branded searches that combine your brand with that subject, and how often AI systems cite you on it. Expect six to twelve months before any of it moves meaningfully.
Page-level reporting is what kills these programmes. A cluster of fifteen pages produces fifteen unremarkable rows in a rankings report at month four, and someone senior looks at it and asks whether the budget might be better spent elsewhere. Aggregate the same fifteen pages and the picture is usually a good deal healthier.
The three signals worth watching:
- Topic share, which is the proportion of total organic traffic across every keyword within a subject that your site captures.
- Branded topic queries, meaning the variety of searches combining your brand name with the subject. When people start searching for you and the subject together, the association has been made by the audience rather than the algorithm, which is the harder and more durable win.
- AI citation frequency on the subject, tracked over time rather than as a snapshot, because AI answers change week to week and a single reading tells you very little.
That third one is the awkward one, since Search Console has only just begun reporting AI citations. Our Search Radar tool gives an early directional read on whether the authority you’ve built is showing up in AI answers, which is more useful than waiting for a metric that may never arrive. It’s the same problem AI SEO work runs into constantly, and it’s worth building your reporting around the gap rather than pretending it isn’t there.
On timing, the honest range is six to twelve months before significant movement, with the first ranking improvements typically showing between months four and six.
The map is a filter, not a list
A topical authority map earns its keep by narrowing what you publish. The clusters you commit to get built properly and finished, the pages that pull the site away from its subject get merged or removed, and what’s left is a smaller plan that concentrates the signal instead of spreading it.
Sequencing is where the money sits. Build the cluster closest to a buying decision first, finish it, then move on, because that’s the difference between a programme that shows a return in month six and one that’s still explaining itself in month eleven.
If you’ve built the coverage and want to know whether AI systems are actually citing you for it, that’s the question our AI SEO services answer.
Frequently asked questions
What is the difference between topical authority and domain authority?
Domain authority is a third-party estimate of a site’s overall backlink strength, produced by tools such as Moz and Ahrefs, and Google doesn’t use it as a ranking factor. Topical authority is subject-specific and reflects how completely a site covers one area. A site with modest domain authority can outrank a much larger one on the subjects it covers more thoroughly.
How long does it take to build topical authority?
Expect six to twelve months before movement is meaningful, with the first ranking improvements typically appearing between months four and six (Ahrefs, June 2026). Coverage gets established first and the compounding gains come later. Reporting at cluster level rather than page level makes the early progress visible sooner, which matters when someone is deciding whether to keep funding it.
How many articles do you need to build topical authority?
There’s no fixed number, and treating it as a page count is the mistake the question invites. What matters is whether your site covers the subject more completely than the sites you’re competing with, and whether the pages connect to each other properly. A narrow subject might need fifteen well-linked pages where a broad one needs fifty.
Does a topical authority map help with AI search visibility?
Yes. AI systems use query fan-out, generating several related searches from one question before answering it, so a site covering a subject from many angles turns up across more of those searches. Independent analysis of 54 experiments, patents and case studies placed coverage of those expanded queries third out of 23 evidenced AI citation factors.
Should I remove old content when building a topical map?
Often, yes. Pages sitting outside your core subject dilute how concentrated that subject looks, and consolidating or removing them is usually cheaper than commissioning replacements. Merge near-duplicates into the strongest version, rewrite thin pages covering a subject worth keeping, and redirect anything genuinely off-topic that has earned links.
