Topical authority

Search systems infer topical authority through a combination of signals: the breadth and depth of a domain's coverage of a subject, the internal link structure connecting that coverage, external citations pointing at pages within the topic, and — increasingly — how consistently a domain is cited as a source across related queries in AI Overviews and other generative surfaces.

There is no single public topical-authority score, but practical proxies exist: the percentage of a defined topic's query universe for which the domain ranks in the top 10, the average position across that query set, and — for generative surfaces — the citation rate across the same set. Tracking these together over time is a more reliable signal than any individual page's rank.

Generative retrieval systems appear to favour domains that demonstrably know a subject in depth, since a site with comprehensive, mutually-consistent coverage of a topic is a lower-risk citation source than a single isolated page making an unsupported claim. A domain's broader topical footprint effectively becomes a trust signal the model can lean on when deciding whether to cite a specific passage.

This means the ROI calculus on content investment has shifted: a business publishing five thin, unconnected pages across five different subjects is now less likely to earn either strong classical rankings or AI citations than the same effort concentrated into one well-built cluster on a single, commercially important subject.

The most common failure is content sprawl: publishing broadly across many loosely-related subjects to hit a volume target, which spreads authority thin and signals no clear specialism to either search engines or actual readers.

A second is orphaned content — genuinely good pages published without internal links connecting them to a pillar or to each other, which prevents the cluster effect from compounding even when the individual pages are strong.

A third is stopping too early: publishing a pillar and three or four spokes, then moving on, when the competitive query set for the topic actually requires ten or twenty facets covered before authority becomes defensible.

Define the topic's full facet map before writing anything: list every subtopic, question, and comparison a genuinely comprehensive resource on the subject would need to cover, drawing on Search Console query data, GBP Q&A, and competitor content audits.

Build one clear pillar page covering the topic broadly, then commission supporting pages for each facet in depth, each one genuinely substantive rather than a thin placeholder built only to exist.

Interlink deliberately in both directions: the pillar links out to every spoke with descriptive anchor text, and every spoke links back to the pillar and sideways to closely related spokes.

Track ranking and citation performance across the full topic query set monthly, and prioritise filling the weakest-performing facets rather than adding volume to already-strong ones.

A local business's addressable topical universe is naturally narrower than a national publisher's, which is an advantage: a well-run local HVAC company can realistically build genuinely comprehensive coverage of 'commercial HVAC in Central Texas' — permitting quirks, climate-specific maintenance schedules, real local pricing — in a way no national aggregator can match with generic content, and that specificity is exactly what both classical ranking and generative citation reward.

Topical authority is the outcome; content clusters are the structural method used to build it; internal linking is the mechanical connective tissue that makes the cluster function as a unit rather than a pile of disconnected pages. None of the three works well without the other two.