# Topical SEO: Build authority with focus, not volume - Episode: P1E05 (Part 1: Modern SEO) - YouTube: https://www.youtube.com/watch?v=eb9Ro3F_250 - Page: https://irongoo.com/videos/topical-authority - Duration: 14:29 - Published: 2026-10-26T15:15:00+05:00 Build a topical map that strengthens your site's focus and supports the pages that earn. Learn what to cover, what to refuse, and how to spot content that drifts beyond your scope. ## What this episode covers - Defining your site's purpose, audience and business model before planning pages. - Giving each page one main purpose, with a core that earns and supporting topics that send readers inward. - Writing a refusal list so unrelated ideas don't quietly become pages. - Testing internal links for reader relevance, with clear anchors and useful surrounding text. - Using page embeddings to flag potential outliers for human review, with the limits of comparisons to Google's leaked fields. - Finding coverage gaps and understanding why topical authority also takes time, clicks, links and reputation. ## Key moments - [00:00 Intro](https://irongoo.com/videos/topical-authority?t=0) - [00:40 Source context](https://irongoo.com/videos/topical-authority?t=40) - [02:32 One purpose per page](https://irongoo.com/videos/topical-authority?t=152) - [03:24 Core, outer and the refusals](https://irongoo.com/videos/topical-authority?t=204) - [05:42 Links need a reason](https://irongoo.com/videos/topical-authority?t=342) - [07:39 Drift, measured](https://irongoo.com/videos/topical-authority?t=459) - [10:27 Topical authority](https://irongoo.com/videos/topical-authority?t=627) - [13:57 Wrap up](https://irongoo.com/videos/topical-authority?t=837) ## Transcript ### Intro Here's a field from Google's leaked documentation: site focus score, a number for how much a site sticks to one topic. Right next to it sits site radius, for how far each page drifts from the site as a whole. Nobody outside Google knows what they weigh. But Google bothered to store them, which means it can look at your site as one thing, with a centre. So where's yours? Today we build a topical map that holds up for our example site ReelRadar, from what it covers to what it refuses, and then we plot every page by meaning to catch the ones that wandered off. ### Source context Every map starts before the first topic, with three questions about the business. What is this site, who is it for, and how does it earn? We call the answer source context. It's our working model rather than a Google term, and it decides the angle of every page you write. Take one topic, the new Dune film, on two different sites. A trailer and where to watch site covers the trailer, the release dates by country and where to stream the earlier films. A film business newsletter covers the budget, the box office and what the studio expects. Same film, different pages, because the two sites earn in different ways. We see the same thing in our client work, far from film. For a dental clinic whose patients fly in from abroad, every treatment page needs an angle a local clinic never writes: how many visits it takes, how long to stay in the city, and what aftercare looks like back home. The neighbourhood clinic down the road writes about parking. Same treatment, different source context. Google asks site owners the same thing in its own guidance. Does your site have a primary purpose or focus? Do you have an existing or intended audience that would find the content useful if they came to you directly? Those are source context questions, in Google's own words. So write it down in one sentence before you plan a single page. ReelRadar helps English-speaking film fans watch trailers and find where to watch films, and it earns from display ads and streaming referrals. Every topic on the map gets tested against that sentence, including the boss's favourite idea. Your source context also leaks out through the parts that repeat on every page: the menu, the footer, the sidebar. Read them like a stranger would. If ReelRadar's footer links to celebrity news and a gadget shop, every single page is telling readers, and machines, that this site is about more than films. ### One purpose per page Same discipline, one level down. One page, one main purpose. Google defines main content as any part of the page that directly helps it achieve its purpose, which only works if the page has one. A ReelRadar film page exists to show the trailer and say where to watch. Cast and runtime support that purpose without competing with it. The usual failure is a page with two purposes. Half where to watch guide, half box office analysis. It serves two different searchers and satisfies neither of them. When you catch yourself writing a second purpose into a page, that's a second page, or a refusal. Here's a quick test we use. Write the page's purpose as one search a person would actually type, like where can I watch Dune Part Two in the UK. Any section that doesn't help answer that, or the obvious next question, belongs somewhere else. ### Core, outer and the refusals Now zoom out to the whole site. Every map has a central entity, the thing the site is about. For ReelRadar that's films, and more precisely, watching them. Around it sit two rings. The core carries the topics closest to how the site earns, and the outer ring covers the context that builds trust and sends readers inward. ReelRadar's core is a page per film, trailer pages, release dates by country, and where to watch pages for films with real demand. Most internal links point here, and this is where the money gets made. Every core page answers the source context sentence directly. The outer ring holds director pages, actor pages, genre hubs and franchise watch order guides. Few of them earn much on their own. Their job is to cover the context around films, catch people earlier in their search, and pass them inward, along with any links they collect. And they do collect links. In our experience, a franchise watch order guide or a genre hub earns links and mentions far more easily than a where to watch page. Nobody shares a commercial page out of love, but people share a good guide. The outer ring gathers that authority, and internal links carry it to the pages that earn. Now look at what's missing from both rings. No celebrity gossip, no box office tracker, no reviews of home cinema gear. Each of those is a real topic with real searchers. None of them fits the source context, so none of them gets a page. That's the part most maps leave out. A topical map is a plan of what you'll cover, and just as much a written list of what you refuse. The list protects you from the slow drift that happens when every good idea in a meeting quietly becomes a page. When the site grows, grow it in order of safety. First, more entities of the same type, which for ReelRadar means more films. Then more attributes for the entities you already have, like release dates for more countries. Then deeper context, like a watch order guide for a franchise. Every step stays inside the source context. Put the refusal list where the whole team can see it, and give one person the job of saying no. Most drift doesn't come from bad ideas. It comes from good ideas for a different site, pitched by people who never read the source context sentence. ### Links need a reason Links between sections need rules, or the map turns into a plate of spaghetti where everything links to everything. Our rule is simple. A link needs a real overlap between two pages, a reason a reader on page A wants page B right now. We call that a contextual bridge. Google's link guidance points the same way. It says Google uses links to judge the relevance of pages, that good anchor text is relevant to both the page it sits on and the page it links to, and that the words around a link matter. A link with no shared context gives those words nothing true to say. Here's a bridge we accept. An actor's page on ReelRadar links to the Dune Part Three trailer page, inside a sentence about the actor's role in the new film. A reader on the actor page wants to know what they're in next, so the overlap is real and the link means something in both directions. And here's one we refuse. A trailer page linking to an article about the same actor's school years. Both pages mention the actor, but nobody watching a trailer wants a childhood biography, and that biography shouldn't exist on ReelRadar in the first place. Shared words are not shared context. When a bridge passes, write the anchor about what the target page gives the reader, not about the keyword you want it to rank for. The first full trailer tells the reader exactly what they get by clicking. Click here tells them nothing, and dune part three trailer release date cast reads like a ransom note. So every link on the map has to pass four questions. Would a reader on this page want that page now? Is the anchor true to both pages? Does the sentence around it explain why the link is there? And do both pages belong on the map? Direction matters too. Most links should flow from the outer ring into the core, because the core is where the site earns and where you want authority to pool. Core pages link outward sparingly, when a reader genuinely needs the context to make a choice. ### Drift, measured So far this has been judgement. Now we measure. The leaked fields from the start of the episode describe a site embedding, a focus score and a radius. We can build a rough version of that idea ourselves with free tools: embed every page, find the centre, and see which pages are standing far away from it. The steps are short. Crawl the site and take each page's title and opening text. Turn each one into an embedding, a list of numbers that stands for its meaning, with any modern embedding model. Average them all to get the site's centre, then measure each page's distance from it. One detail decides whether this works. Embed the main content, not the whole HTML. If the menu, footer and sidebar go into every embedding, every page looks like every other page, and the plot shows you your template instead of your topics. Strip the repeated parts first. Here's ReelRadar's result, using example data. A hundred and forty pages, one dot each, squashed onto a flat plot so we can see them. Film and trailer pages cluster tightly around the centre. Director and franchise pages sit a little further out, which is fine, because that's the outer ring doing its job. Then the red dots. Nine pages sit well outside the rest: a piece on a director's childhood, a buying guide for popcorn makers, an explainer on how film cameras work. Each one was a reasonable idea in a meeting, and each one pulls the site's centre away from what ReelRadar is for. Why would a few strays matter? Because the centre is an average. Every off-topic page pulls it toward itself, and the more of them there are, the further the centre slides away from the films ReelRadar is actually about. The leak's radius field describes the same kind of spread. How far is too far? We don't trust a fixed number. We sort every page by its distance from the centre and read the twenty furthest by hand. The line usually draws itself, right where pages stop being slightly unusual film pages and start being about something else entirely. So what do you do with a red dot? Re-angle it, for example by turning the camera explainer into a page about the cameras behind films people want to watch. Move it to a site where it fits. Remove it and redirect to the closest relevant page. Or keep it on purpose, if it earns its place and you accept the cost. One thing to keep straight: this is our measurement, not Google's. The leak tells us Google stores something similar, not how it computes it or what it weighs. So use the plot to find pages that deserve a human look, and never to delete pages automatically. Run it every quarter. Drift happens slowly, one reasonable idea at a time, and a plot catches it long before traffic does. The notebook we use is free on the episode page, along with the map template. ### Topical authority Which brings us to the phrase everyone uses: topical authority. In our working model it has two halves. Coverage, meaning how completely you answer the entities and attributes inside your scope. And history, meaning how long you've performed well on that topic, in clicks, links and reputation. The history half has support in the court record. It describes Navboost learning from thirteen months of click data, and a quality score that's largely static and tied to the site rather than the query. A new site can't fake either one. It earns them over time, which is why coverage alone, published in a hurry, rarely wins the head terms. The coverage half you can measure today. Take the attribute plan from our previous episode, list every entity in scope against every attribute you committed to, and count the filled cells. The empty cells are your next pages and sections, ranked by demand and fit, and that's a much better to-do list than a pile of keyword ideas. For a new site, this flips the order of work. Build the core and the narrow, specific pages first, where competition is thin, and earn clicks and links there. Head terms come later, once there's a history to point to. Going after the biggest terms on day one is the classic way a new site stalls. Claim one. Publish hundreds of posts, fast, and topical authority is yours. Volume buys coverage at best, and history can't be bought with volume. Worse, Google's spam policy names scaled content abuse, many pages made mainly to manipulate rankings, no matter how they're created. Its guidance also flags lots of content on many different topics as a warning sign. Placebo, and at scale, a real risk. Claim two. Topical authority is a Google score, and you can raise it. Google has described one system with a similar name, and it's for news. A 2023 post on topic authority explains how Google finds expert publications for newsy queries in areas like health, politics and finance, using notability, original reporting and reputation. No sitewide topical authority score appears in Google's documentation or in the leak. As a score you can raise, it's folklore. Placebo. That news post still has a lesson for everyone. Google's advice to publishers was to provide great coverage about the areas and topics they know well, and reputation in that system includes a source's history of high-quality reporting. Narrow and deep beats wide and thin, even where Google built a system for it. Claim three. Google stores how focused a site is. Here the evidence is direct. The leaked documentation contains a site focus score and a site radius, described in plain words. Their weight is unknown, so the verdict is a narrow one: Google stores how focused a site is, and that alone is reason enough to keep the map tight. Works. Put the verdicts together and the picture is clear. Authority on a topic comes from covering a tight scope well and staying in it long enough to earn a history. It doesn't come from volume, and nobody can sell you the score. Want to start on Monday? Write the source context sentence. Draw the core and outer rings, then write the refusal list. Run the bridge test on your ten most linked pages. Then run the drift plot once, and put the next one in the calendar for three months from now. ### Wrap up Here's the episode on one screen. Source context decides the angle. One purpose per page. A core that earns, an outer ring that feeds it, and a written list of refusals. Links that pass the bridge test, and drift you measure every quarter. The topical map template and the drift detection notebook are free on the episode page at irongoo.com. Next time, we go from topics to queries, and decide when a query deserves a page, a section, or nothing at all.