My objective with this little experiment was to develop a content architecture for any website and providing it as an artifact inside Claude.
This experiment leverages exceptional research done by Koray Tuğberk GÜBÜR on topical authority, how to build topical authority, and how to determine if a query deserves a page or not. This framework also utilizes research published on peec.ai and the importance of listicles in being cited, position of mention and the count of mentions.
The question I was trying to answer:
- Can it act as a gatekeeper validating if a new page was required for a query?
- Should an existing page be rewritten instead?
- Is quality important or quantity ?
- Can the framework work for a marketplace as well as a B2B site?
- Is the core – outer framework akin to the hub and spoke model where you have informational content as the outer layer linking to the commercial / money inner pages?
This is the output of that exercise:
Content architecture brief: Topical Authority Blueprint
Seven build rules distilled from current research on query templates, semantic SEO, and AI-engine (GEO) visibility — used as the standing instruction for how this site’s content is structured, linked, and expanded.
01 Query Deserves a Page (QDP)
The gate that decides whether a new query gets its own URL or folds into an existing page. This is the first filter applied before any page is built — it prevents both content gaps and self-cannibalization.
- Create a new pageonly when the query shows real search demand, contains a distinct entity, is dissimilar to existing pages, and fits a recognizable pattern shared with other queries.
- Consolidate, don’t proliferate.Near-identical queries (e.g. dozens of location–service combinations with no independent demand) get merged into one page, not spread thin.
- Programmatic pages are opt-in, not default.Generate a page for an entity variant only where the template + entity pairing has verified demand.
02 Core → Outer Architecture
Every topic cluster is built in two layers. The outer layer earns attention; the core layer converts it. Authority is designed to flow one direction: inward.
- Core sectionholds the commercial / money pages — the entities and concepts the business needs to rank for.
- Outer sectionis informational: it answers the surrounding query templates, captures early-stage search demand, and exists to funnel authority inward via internal links.
- Every outer page links to a core pagewith a contextual, non-generic anchor — not the other way around.
03 Entity × Query-Template Coverage
Topical authority is a two-axis completeness problem: which entities you cover, and which query formats you cover them in. The strongest sites do both, not one.
- Axis 1 — Entities:cover every entity in the class and its shared attributes (every product in a category, every symptom of a disease, every plan on offer).
- Axis 2 — Query templates:cover the recurring query shapes people use regardless of topic (“how to…”, “[X] vs [Y]”, “best [X] for [Y]”, “does [X] cause [Y]”).
- Map templates before writing.List the core templates in the niche, identify which entity variants each needs, and only spin up a new page where the predicate or entity type meaningfully changes.
04 Semantic Precision
Once the page architecture is right, relevance is won or lost at the sentence level. Structure content as explicit relationships, not just topic coverage.
- Write in entity–attribute–context triples.(Facility → located in → Thailand), (Therapist → specializes in → Treatment Method) — structure paragraphs so these relationships are stated directly, not implied.
- Subject position matters.“Advisors help families achieve independence” and “Families achieve independence with advisor help” match different query augmentations — lead sentences with the entity you want the page to rank for.
- Design the page’s components around query augmentation.One search intent (e.g. “rehab Thailand”) expands into finding, comparing, verifying, and pricing — give each its own visual/content component (hero, evidence strip, review carousel, tabbed attributes, process accordion) rather than one long paragraph.
05 Internal Linking & Signal Flow
Links are how outer-page authority actually reaches the pages that need to rank. Treat internal linking as a deliberate signal-routing system, not a navigation afterthought.
- Link to “quality nodes.”Concentrate authority into pages that represent core topical themes rather than spreading it evenly across every marginal page.
- Diversify anchor text.Don’t reuse the same anchor more than ~3 times — varied, contextual anchors reinforce topical relevance more than repetition does.
- Every outer page earns its keepthrough the clicks, impressions, and engagement it generates for the core page it supports — that’s the actual mechanism behind the link, not just the link itself.
06 AI Visibility (GEO)
Ranking on Google is necessary but no longer sufficient — AI engines cite a small, repeatable set of third-party sources. Win those, deliberately, rather than chasing every mention.
- Diagnose the market first.Fragmented / low-awareness categories: rank position in listicles is critical, AI has no prior knowledge to fall back on. Established categories: listicle presence complements existing brand strength. Regulated sectors: focus on the handful of trusted sources AI actually pulls from.
- Win the few listicles AI engines actually retrieve— not every listicle that exists. Five strong, frequently-cited placements outperform fifty scattered ones.
- Track three metrics separately:visibility (do you appear at all), answer position (how early you’re mentioned), and mention count — they move independently and need separate tactics.
- Match tactics to engine type.Tight-retrieval engines (ChatGPT, Copilot) weight individual listicles heavily; broad-retrieval engines (Google AI Overview) spread impact across more sources.
07 Technical Health & Measurement
None of the above works if the site isn’t crawlable and the results aren’t measured. This is the maintenance layer that keeps the architecture honest.
- Crawl hygiene targets:99%+ HTML crawl rate, 99%+ pages returning 200/304, discovery rate above 20%. Route all crawl budget to indexable, 200-status URLs.
- Avoid parameter-based URLsfor filtering/location logic — they fragment indexing. Prefer clean, server-rendered paths.
- Report the KPIs that matter:click growth after consolidation, average position on primary queries, non-indexed page reduction, and AI citation counts for the domain.
How to use this content framework in Claude to build your website's content architecture.
Save sections 1 to 7 above as “content framework.md” and import into your Claude project. Or save it is an artifact so that it is persistent.
In the project instructions, ask it to adhere to the content framework and design the content architecture of your website. You can create another markdown file to describe what is your business, a brief about the services you offer, who you serve etc so that it has some context.