Topical authority is an outcome you infer, not a score you install.

“Topical authority” is used as though Google exposes an authority meter for every subject. It does not. The phrase can still describe a useful observation: some sites repeatedly publish strong material in an area, connect it coherently, and become recognised sources. The mistake is turning that observation into a mechanical quota of pages.

Google's people-first guidance asks whether a site has a primary purpose or focus. Its link guidance says every important page should be linked from another page and recommends contextual internal links that help readers and Google understand the site. Those are sensible inputs. Neither document says that a pillar plus a fixed number of spokes creates authority.

Begin with a decision system, not a keyword tree

Choose a subject the site has a reason and ability to cover. Then map the decisions people make inside it. Search demand can reveal the language and frequency of those needs; customer questions, product constraints, support records, sales calls, and professional practice reveal what the answer must contain.

Each proposed page needs a distinct job. If two queries require the same answer and page type, one strong page usually owns both. If they require different evidence or actions, a split may help. This is why “best,” “cost,” “how,” and “versus” sometimes deserve separate pages and sometimes collapse into one complete guide.

Four observable inputs are better than an imaginary score

Coverage

Can the intended audience complete the important tasks within the subject, or do obvious decisions send them back to search? Coverage is not total word count. A small set of decisive pages can cover a subject better than a large set of definitions.

Evidence

Do the pages contribute direct experience, original data, careful synthesis, examples, calculations, or implementation detail appropriate to their claims? Summarising the same sources across fifty pages expands inventory without expanding what the site knows.

Coherence

Can a reader and crawler discover the relationships that matter? Navigation should expose major destinations. Contextual links should answer the next useful question where it arises. An automated “related posts” strip cannot carry every conceptual relationship.

Recognition

Do relevant people cite, share, mention, revisit, or search for the work? External recognition is earned at the page and source level. It cannot be replaced by internally declaring a publication authoritative.

Audit before you expand

Most established sites already have fragments of the subject: old articles, product documentation, category pages, FAQs, and overlapping campaign pages. Use the content audit template to decide what to keep, improve, merge, or remove before commissioning another cluster. Expansion built on unresolved overlap often creates more pages competing for the same job.

When a real information gap remains, choose the page type and evidence that solve it. Then link from the pages where that answer becomes useful. Google explicitly recommends crawlable anchors and contextual internal links; the link guidance is pleasantly free of anchor quotas or cluster diagrams.

Authority grows as a side effect of being worth relying on

The causal chain is slower and less theatrical than a score: the site publishes useful work, people can find and understand it, some choose to rely on or cite it, and repeated recognition strengthens the source's standing. A defensible distribution strategy can increase the chance that the right people encounter the work. It cannot rescue material with no reason to be cited.