How 30 Focused Articles Beat 300 for Topical Authority

What is topical authority, and why do Google and AI engines reward it?

Topical authority is the trust a search engine or an AI model places in your site for a specific subject, earned by covering that subject thoroughly and coherently rather than by publishing lots of unrelated pages. Google stopped matching keywords to documents in isolation years ago; it now weighs whether your domain shows familiarity with the whole web of related concepts, questions and entities around a topic. A site with twenty connected articles about dental implants usually outranks a site with one brilliant 5,000-word guide, because twenty articles prove sustained expertise instead of a single lucky swing.

AI answer engines apply similar logic with one extra filter: they have to be able to lift a usable answer straight off your page. Being retrieved and being cited are not the same thing. A model may pull a dozen candidate pages, then quote the two or three that define terms plainly, give steps in order, and put numbers and dates where they can be found. Depth helps here too, because a model that keeps meeting your domain across a cluster of related questions starts treating you as the reliable source on that subject.

This is also where organic marketing separates itself from paid. A PPC campaign buys attention for exactly as long as the card keeps working; pause it and the traffic stops with nothing left behind. Cluster content behaves like an asset instead. Each article keeps earning impressions, keeps feeding internal links to its neighbours, and stays eligible for citation inside ChatGPT, Gemini, Claude and Perplexity — places where ads cannot appear at all. It's worth checking your current GEO score before you write anything, so you have a baseline to measure against.

How do you pick the right three to five clusters for your business?

Pick three to five clusters that map directly to the services you actually get paid for, and ignore everything else for the first year. The test is blunt: if a topic can't plausibly end in an enquiry, a booking or a sale, it doesn't earn a cluster. Most small businesses only have three or four real money-makers anyway — an accountancy firm sells company formation, VAT and payroll; a physiotherapy practice sells back pain, sports injuries and post-op rehab. Those are your clusters. Write the shortlist on a single page and be ruthless about what doesn't make it.

Then size each one. A cluster is viable if you can list eight to fifteen genuinely distinct questions under it without repeating yourself — enough for a pillar page plus a real body of supporting articles. If you run dry after four, it's a subtopic and belongs inside something bigger. If you're still going strong at forty, it's two or three clusters wearing a trench coat, and you should split it before writing a word. This sizing step takes an afternoon and saves months, because it's what stops two of your own pages from fighting over the same query later on.

The hardest part is deciding what not to write. Industry news, award announcements, staff birthdays and generic "five tips" posts feel like output but contribute nothing to authority on any subject a customer searches for. I'd go further: if your blog already holds 200 scattered posts, the fastest win isn't publishing more — it's consolidating or retiring the two-thirds nobody has ever read and folding the useful parts into the clusters you just chose. A tight site about three subjects reads as an expert. A sprawling one about thirty reads as a content farm, to Google and to the models alike.

How do you map the questions inside a cluster so no article is wasted?

Map each cluster by collecting the questions your customers already ask, then sorting them by the stage of the decision they belong to. Start offline rather than in a keyword tool: your inbox, your consultation notes, the three things every caller wants to know before they'll book. Those questions come pre-validated by real demand. Widen the net afterwards with search autocomplete, the "people also ask" box and the phrasing competitors use in their own headings. Within an hour you'll have sixty or seventy raw questions per cluster, most of them variations on the same dozen underlying worries.

Now group them by intent, not by wording. "How much do implants cost", "implant price per tooth" and "are implants covered by insurance" are one article, not three — same underlying need, different keyboards. Each surviving group becomes exactly one page with one job, which is how you avoid cannibalisation, where two of your own articles split the signals and neither one ranks. Give every page a clear definitional answer near the top, then the detail underneath: steps in order, comparisons in a table, real figures where you have them. That structure is precisely what crawlers and models find easy to quote.

Depth also means saying things only you can say. Prices you actually charge, timelines from your own caseload, the complication you see twice a month, the question you wish clients asked sooner — that first-hand experience is the part no competitor can copy and no general-purpose model already holds. Keep every page answering its own question and nothing else; a 3,000-word article that wanders is less useful than a 900-word one that resolves a single query cleanly. Date your pages, keep the numbers current, and revisit anything containing a price or a legal threshold at least once a year.

How should you interlink and sequence publishing over the first year?

Finish one cluster completely before starting the next, and link every article up to its pillar and sideways to two to four siblings as you publish. A half-built cluster is the worst of both worlds: the money is spent and you're still invisible, because the pattern that earns authority is coverage, not cadence. So sequence by cluster, not by calendar. Eight to fifteen articles on your most commercial subject over three or four months will genuinely move rankings; the same number spread thinly across five subjects in the same period moves almost nothing. Pick the cluster closest to revenue and go all in.

On linking, the practical rules are dull and effective. Three to five contextual links per thousand words is plenty, placed inside the body text where the surrounding sentences explain what sits on the other end. Write anchors that describe the destination — "how long implant healing takes", not "click here" — and vary the phrasing so forty pages don't all point at the pillar with an identical string. Keep every important page within three clicks of the homepage, and never leave an orphan: if nothing links to a page, crawlers struggle to find it and readers never will. Audit quarterly.

Give the whole thing two to three quarters before you judge it. Clusters compound rather than spike: early articles get indexed, internal links push authority toward the pillar, the pillar starts ranking for the head term, and the long-tail pages you'd half forgotten begin pulling traffic on their own. The same pages are being retrieved and quoted by AI assistants in the meantime, which often happens well before page-one rankings do. If you want a sense of what that traffic is worth against your current ad spend, run the numbers first — patience is easier to justify with a figure attached.

What does a topical authority plan look like for a real business?

Take a four-chair general dental clinic. Three clusters cover almost everything worth ranking for: implants, clear aligners for adults, and anxious or first-time patients. Each gets one pillar page written for the head term, plus a first wave of supporting articles answering one question apiece, growing toward ten or twelve pieces as the cluster matures. That's roughly thirty focused articles across a year — set against the 300 assorted posts many clinics accumulate over a decade, which rank for nothing because no single subject was ever covered properly. Thirty wins, and it wins on a fraction of the budget.

The third cluster is the one most clinics skip and the one that converts best. Nobody searches "anxious patient protocol", but plenty of people search "does a root canal hurt" and "what happens at your first dental appointment" — and the person asking is far closer to booking than someone casually reading about implant materials. Clusters built around fear, cost and process tend to out-earn clusters built around clinical detail. The same holds across trades: "how much does a divorce cost" beats "family law explained" every time, because it matches what a real person types at eleven at night.

Running all of this by hand is a genuine job: research, drafting, linking, updating, tracking. That's the part Artiql automates. It builds the topical map for your business, sequences the articles cluster by cluster, writes them natively in English or Hebrew rather than translating, publishes to your own domain through a headless CMS with a review queue so you keep final say, wires the internal links as each piece goes live, turns articles into short video for YouTube, Instagram and TikTok, and tracks both rankings and your share of AI citations. To see the map for your own business, book a demo.

Frequently asked questions