Topical authority is the third pillar of Search Quality Assurance (SQA-200)™ and carries a budget of forty points across fourteen checkpoints. It asks whether the content covers its subject as a deliberate map with one owning page per intent, or as a backlog of articles that happen to share a category.
Hub and spoke: why a map beats a publishing backlog
A hub and spoke architecture designates one page as the destination for a topic and surrounds it with supporting pages that answer narrower questions and link inward. The alternative, which most sites arrive at without choosing it, is a chronological backlog where every article is peer to every other and relevance is inferred by an engine from whatever the internal links happen to say.
The difference shows up in how the two structures fail. A map fails visibly: a missing spoke is an intent with no page, and the gap is in the map. A backlog fails invisibly: coverage looks complete because many articles exist, while the intent that converts has three partial answers spread across three posts and no page that owns it.
Intent mapping across the decision ladder
Checkpoint P3.06 requires every URL to be categorized by the stage of the decision it serves, and the framework maps those stages onto the five rungs of Decision Ladder Navigation: Orientation, Comparison, Proof, Price and Action. Categorizing a URL means committing to one rung, not noting that a page touches several.
The commitment is what makes the category useful. A page assigned to Comparison is then held to what a comparison-stage reader needs, and its internal links are expected to hand the reader upward rather than sideways. A page that claims three rungs is a page whose next step cannot be designed, which is precisely the failure the fifth pillar has to clean up later.
Cannibalization: the failure that looks like productivity
Cannibalization is two or more pages competing for the same intent, and checkpoint P3.13 exists because it is the one content failure that looks like success from the inside. The team published more, the archive grew, and the reporting shows activity. The engine, meanwhile, has to choose between near-identical candidates and distributes the signal across them instead of concentrating it.
Auditing for cannibalization means grouping URLs by the intent they target rather than by the keyword they mention, then consolidating each group to one owning page and redirecting the rest. The consolidation is the uncomfortable half, because it deletes work that took effort to produce. Keeping both pages is the option that preserves the effort and the problem together.
The fourteen content hub checkpoints
Below are the fourteen itemized checkpoints of pillar three. Six architectural items carry three points each, seven page-level items carry two, and the quarterly refresh item carries three, reflecting that a map left unmaintained stops describing the site within a year.
| Item | Checkpoint | Weight | Verification and execution standard |
|---|---|---|---|
| P3.01 | Topical Map Blueprint | 3.0 pt | Architect complete topical map mapping all core entities, sub-topics, and user intents. |
| P3.02 | Pillar Money Page Role | 3.0 pt | Designate high-converting pillar pages as destination hubs for cluster traffic. |
| P3.03 | Supporting Cluster Content | 3.0 pt | Publish comprehensive supporting cluster posts covering specific MOFU/TOFU questions. |
| P3.04 | Internal Link Silo Rigor | 3.0 pt | Restrict internal linking within topical silos, preventing cross-silo link equity leakage. |
| P3.05 | Contextual Anchor Precision | 3.0 pt | Use exact-match entity anchor text for internal links, banning 'click here' or generic phrases. |
| P3.06 | Intent-Funnel Mapping | 3.0 pt | Categorize every URL into TOFU (problem), MOFU (compare), or BOFU (action) intent. |
| P3.07 | Heading Hierarchy Logic | 2.0 pt | Maintain strict single H1 per page followed by logical nested H2/H3 sub-sections. |
| P3.08 | Intro Answer Placement | 2.0 pt | Deliver direct answer in first 50 words under H1 before detailed topic breakdown. |
| P3.09 | Experience Proof Stacking | 2.0 pt | Embed original data, case study metrics, or video proof inside technical guides. |
| P3.10 | Fact-Checking Citation | 2.0 pt | Cite authoritative primary sources (studies, standards) for all factual claims. |
| P3.11 | Scannable Typography | 2.0 pt | Format long-form text with short paragraphs, bold key terms, and bulleted lists. |
| P3.12 | CTR Asset Optimization | 2.0 pt | Craft compelling Title tags (< 60 char) and Meta Descriptions (< 155 char) with clear value. |
| P3.13 | Cannibalization Audit | 2.0 pt | Audit site for duplicate target intents, consolidating competing URLs into primary hub. |
| P3.14 | Quarterly Content Refresh | 3.0 pt | Maintain quarterly schedule to update outdated stats, broken links, and search intents. |
How to score the topical authority pillar
Scoring the topical authority pillar means summing the weights of the checkpoints that pass, and the weighting is deliberately uneven: the six architectural items carry three points each while the page-level craft items carry two. An auditor who passes every craft item and fails the architecture scores badly on purpose, because that is the state the pillar exists to catch.
Two items resist a quick verdict and should be scored last. The topical map is judged against the intents the business actually serves rather than against its own internal tidiness, and cannibalization is judged by grouping URLs by target intent rather than by the keyword each one mentions. Both take longer than the other twelve combined, and both are where the pillar's real findings come from.
What topical authority hands to AI-ready extraction
Topical authority hands the next pillar a page per intent, which is what makes extraction formatting worth doing. Answer-first structure applied to a page that does not own its question improves a candidate that was never going to be selected, so the fourth pillar's thirty points are contingent on the third pillar's map being real.
Continue to pillar four, AI-Ready GEO and AEO Optimization, which formats those owning pages so a generative engine can lift the answer out of them.