Building Topical Authority Schema
Topical Authority is mathematically computed by Google. Learn the step-by-step process to build a complete entity schema that serves as the foundation for all content and structure decisions.
What you'll learn
- Understand what siteEmbedding, siteFocusScore, and siteRadius are and how they relate to authority
- Learn the 6-step process for building a topical authority schema
- Know how to identify the central topic and collect all subtopics using data, not intuition
- Understand cosine similarity scoring and how to use it to validate subtopics
- Build a SILO graph from the collected entity data
What Is a Topical Authority Schema?
A Topical Authority Schema is a structured map of all topics and their relationships on a site. It shows which central topics anchor the site, what subtopics extend from them, how they interconnect, and what coverage gaps exist compared to competitors.
Google evaluates site topical authority through three calculated parameters:
| Google Parameter | Meaning | Schema Impact |
|---|---|---|
| siteEmbedding | The average semantic vector of the entire site — "what is this site about" | More homogeneous topics = cleaner, stronger signal |
| siteFocusScore | How focused the site is on a single topical area | No off-topic content = higher focus score |
| siteRadius | The spread of pages around the central topic in semantic space | Smaller radius = narrower, deeper expert site |
Topical Authority = Topical Coverage × Historical Data. The schema defines Topical Coverage — how completely you cover the topic. Without a complete schema, content is created randomly and coverage never becomes systematic.
Step 1: Identify the Central Topic
The schema is built outward from the center. The central topic is the single concept Google should most strongly associate with your site. For a large site, there may be several central topics — each forming a separate SILO branch.
How to identify it correctly:
- GSC: highest impression queries — Google already associates your site with these topics. The highest-impression query cluster reveals the current central topic in Google's model.
- Ahrefs Organic Keywords: top 10 by traffic — confirms what Google is ranking you for today.
- The client's primary product/service — what they primarily sell and monetize, not everything they offer.
The central topic should be as specific as possible. Not "finance" but "CFD trading UK." Not "health" but "weight loss for women over 40." The more specific the central topic, the cleaner the schema and the lower the siteRadius.
Step 2: Collect All Subtopics
Subtopics are not invented — they are discovered from data. Use five sources:
- Competitor Top Pages in Ahrefs — open Top Pages for 3-5 competitors with similar business models, sort by keyword count (not traffic). The URL patterns (
/cfd-trading/brokers/,/cfd-trading/strategies/) reveal their SILO structure. - People Also Ask + Google Autocomplete — each PAA question and autocomplete suggestion represents a subtopic or attribute users associate with the central topic.
- Semantic mapping tools — tools like agents.frac.tl/Semantic_Graph automatically build entity relationship maps from a seed term.
- Wikipedia categories and See Also — Wikipedia structures knowledge similarly to how Google models topics. The category structure and related articles section reveals the canonical organization of a knowledge domain.
- LLM query fan-out — asking an LLM to expand a topic list can surface subtopics that weren't in the initial research, but every AI suggestion must be validated for cosine relevance in Step 3.
Step 3: Score Cosine Relevance
Not every collected subtopic belongs in the schema. Subtopics that are semantically distant from the central topic increase siteRadius and reduce siteFocusScore — they hurt authority more than help it.
Each candidate subtopic should be scored for relevance to the central topic. Tools like the Moonlit Platform (app.moonlitplatform.com) compute cosine similarity scores. Any subtopic scoring below 0.6 similarity should be excluded or placed in a separate, clearly bounded SILO.
Steps 4–6: Group into SILOs, Prioritize, Map Links
Once validated subtopics are collected, group them into SILO clusters — logical topic blocks that share a common theme. Each SILO gets a hub page (the "pillar") and multiple supporting pages. The hub page links down to supporting pages; supporting pages link back up to the hub and across to relevant sibling pages.
Prioritize SILOs by: competitive gap (what competitors rank for that you don't), search volume, and implementation difficulty. The schema then directly feeds the content plan, site structure, and internal linking strategy.
Key Takeaways
- siteEmbedding, siteFocusScore, and siteRadius are Google's computed topical authority metrics — the schema directly influences all three
- The central topic must be as specific as possible — broad central topics dilute the schema and reduce focus score
- Subtopics must be collected from data (competitor pages, PAA, Wikipedia) — not from intuition
- Cosine similarity scoring validates whether a subtopic belongs in the schema or hurts siteRadius
- The completed schema feeds content plan, site structure, and internal linking — it is the strategic foundation of the project