Why Schema Alone Is Not Enough

Even with Schema.org correctly implemented, many sites fail to get cited by AI. In most cases, the root cause is a lack of semantic alignment between schema markup and body text. AI uses structured data as a "hint," but ultimately decides whether to cite based on entity recognition in the body text. Even if the schema name property contains the correct product name, if the body text refers to "our tool" or "this service," the AI cannot determine what the sentence refers to and drops it from citation candidates.

The approach that addresses this problem at its root is ontology-driven content editing. This article covers the concept of ontology, its mapping to Schema.org, and its impact on Sighted scores, aimed at implementers and content architects. Note that this methodology is still under validation, and results may vary by domain and content type.

What Is an Ontology — Relationship to Knowledge Graphs

An ontology is a formal system that defines "concepts (entities)" and "relationships between concepts (properties)" within a given domain. While a taxonomy (like a library classification system) represents "hierarchies of things," an ontology also captures "relationships between things."

Term	Definition	Example
Entity	An independently existing concept, object, or person	Sighted, Tech Knowledge Base, Shota Nagabuchi, AEO
Class	The type/category of an entity	SoftwareApplication, Organization, Person, Concept
Property (Relationship)	A predicate connecting entities	provider, author, measures
Instance	A specific entity belonging to a class	Sighted is an instance of SoftwareApplication

Google's Knowledge Graph is a large-scale ontology based on RDF (Resource Description Framework). ChatGPT and Perplexity are also believed to build similar internal entity graphs from their training data. The reason AI can accurately respond that "Sighted is an AEO monitoring tool provided by Tech Knowledge Base" is because this relationship is consistently described across the training data.

Why Entity Recognition Matters for AEO

Entity Dependency in AI Citation Decisions

When generative AI produces a response, it identifies entities relevant to the query and cites sources that contain reliable descriptions of those entities. Sources with ambiguous entities (e.g., "our company," "our tool") score lower as citation candidates because the AI cannot determine which entity the text describes.

According to research by Profound Insights (November 2025), paragraphs that explicitly name entities showed an average 1.8x higher AI citation rate compared to paragraphs using pronouns or abbreviations. However, this figure varies by category and domain and may not reproduce across all contexts.

The Coreference Resolution Problem

Pronouns and abbreviations like "it," "this," "the company," and "the above" are natural for human readers but require AI to perform coreference resolution. When resolution fails, entity binding for the entire paragraph becomes indeterminate. Cross-section references (referring to an entity from a previous H2 section as "it" in a later H3) are particularly risky, causing information loss when AI extracts citation blocks.

Ontology Design Process

Step 1: Create a Domain Entity Table

Inventory all entities that appear across your site, documenting their class, canonical name, aliases, and definition. This table serves as the "dictionary" for content editing.

Class (Schema.org Type)	Canonical Name	Aliases	Definition (within first 60 words)
Organization	Tech Knowledge Base Inc.	Tech Knowledge Base	Tech Knowledge Base is a Japanese startup supporting business growth in the AI era. It provides AEO consulting services and the AEO monitoring tool Sighted.
SoftwareApplication	Sighted	Sighted AEO tool	Sighted is an AEO monitoring platform developed by Tech Knowledge Base. It measures enterprise Mentions, Citations, and Response Share across ChatGPT, Perplexity, and Gemini.
Person	Shota Nagabuchi	Nagabuchi	Shota Nagabuchi is the CEO of Tech Knowledge Base. After graduating from Kyoto University, he gained experience in M&A analysis and AEO consulting before developing Sighted.
Concept	AEO (Answer Engine Optimization)	AEO	AEO is the practice of optimizing content to be cited and recommended by AI answer engines such as ChatGPT and Perplexity. It has gained attention since 2024 as an evolution of SEO.

Step 2: Define Relationships (Properties)

Clarify inter-entity relationships using verbs. This step maps directly to Schema.org property design.

Subject	Relationship (Verb / Schema.org Property)	Object
Tech Knowledge Base Inc.	provider / provides	Sighted
Tech Knowledge Base Inc.	provider / provides	AEO consulting services
Sighted	measures / measures	AI visibility (Citations, Mentions)
Shota Nagabuchi	founder / founded	Tech Knowledge Base Inc.
Shota Nagabuchi	author / authored	Knowledge articles
Sighted	relatedTo / related to	AEO (concept)

Step 3: Convert to Nested Schema.org Structure

Translate the relationship table from Step 2 into JSON-LD. The key principle is to express relationships through nesting (include primary properties inline, not just reference by ID). AI has limited ability to follow JSON-LD IDs, so inline expansion improves recognition accuracy.

{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "@id": "https://tkbase.co.jp/#organization",
      "name": "Tech Knowledge Base Inc.",
      "url": "https://tkbase.co.jp",
      "description": "A Japanese startup supporting business growth in the AI era. Provides AEO consulting services and the AEO monitoring tool Sighted.",
      "founder": {
        "@type": "Person",
        "@id": "https://tkbase.co.jp/team/nagabuchi#person",
        "name": "Shota Nagabuchi",
        "jobTitle": "CEO",
        "sameAs": [
          "https://twitter.com/tkbase_nagabuchi",
          "https://www.linkedin.com/in/nagabuchi"
        ]
      },
      "makesOffer": {
        "@type": "Offer",
        "itemOffered": {
          "@type": "SoftwareApplication",
          "@id": "https://sighted.tkbase.co.jp/#product",
          "name": "Sighted",
          "applicationCategory": "BusinessApplication",
          "description": "An AEO monitoring platform that measures Mentions, Citations, and Response Share across ChatGPT, Perplexity, and Gemini.",
          "provider": {
            "@id": "https://tkbase.co.jp/#organization"
          }
        }
      }
    }
  ]
}

Using @graph allows you to define multiple entities within a single JSON-LD block and cross-reference them via @id. This expresses the relationship "Organization contains Person and provides SoftwareApplication" in a single schema.

Step 4: Apply Entity Explicitness Rules to Body Text

Once schema design is complete, reflect the same entity names and relationships in the body text. Use the following rules as editorial guidelines.

Rule	Before	After
State canonical name + class on first mention	"Our tool..."	"Sighted, the AEO monitoring tool by Tech Knowledge Base,..."
Re-state at section beginnings without abbreviation	"Using this..." (continuing from previous H2)	"Using Sighted..."
Make relationships explicit with verbs	"Tech Knowledge Base and Sighted..."	"Sighted, developed and provided by Tech Knowledge Base,..."
Place definition within first 60 words	(Jump straight into feature description)	Begin with "Sighted is an AEO monitoring platform that measures... in AI responses."
Include source and date for statistics	"Citation rates double"	"Citation rates are 1.8x higher on average (Profound Insights, November 2025)"

Step 5: Validation — Verifying Impact on Sighted Scores

After deploying content that reflects the ontology design, verify the effect using these metrics.

Metric	Tool	When to Check
Technical score (schema nesting structure)	Sighted free audit	Immediately after deployment
Mention / Citation trends	Otterly / Peec	2-4 weeks later
Entity recognition accuracy	Direct questions to ChatGPT	Ongoing ("What is Sighted?" "What kind of company is Tech Knowledge Base?")
Structured data validity	Google Rich Results Test	Immediately after deployment

Sighted Score and Ontology Design Correlation

Among Sighted's diagnostic scores, the following items directly correspond to ontology design quality.

Nested schema (Organization → WebPage → Article): Whether the @graph structure implemented in Step 3 is accurately recognized
Publisher / Author binding: Whether the Author and Publisher for each article match the Person and Organization IDs
Question-format headings: Headings that include entities in question format are evaluated as AEO citation targets

If issues rated "High" or above are detected, resolve the technical problems before proceeding with ontology design. If the schema nesting structure is broken, the AI's entity graph construction will not work correctly no matter how accurate the entity descriptions in the body text.

Common Design Mistakes and Solutions

Mistake 1: Mixing aliases without setting sameAs

Mixing "Sighted," "sighted.tkbase.co.jp," and other variations on the same page may cause AI to treat them as separate entities. List canonical URLs, social profiles, and Wikidata in sameAs, and standardize body text to the canonical name from your definition table.

Mistake 2: Using relative paths instead of absolute URIs for @id

Relative IDs like "@id": "#organization" break entity graph consistency across multiple pages. Use absolute URIs such as "@id": "https://tkbase.co.jp/#organization". Sighted's audit also detects this as a "relative URL validation error."

Mistake 3: Confusing ontology with taxonomy

A category hierarchy (breadcrumb / taxonomy) like "AEO Tools > Monitoring > Sighted" only represents IS-A relationships. An ontology includes diverse relationships (HAS-A, DOES, RELATED-TO), such as "Sighted is developed by Tech Knowledge Base, measures AEO visibility, and supports ChatGPT and Perplexity." Do not assume ontology design is complete just because you implemented BreadcrumbList.

Mistake 4: Not defining Concept (abstract) entities

Specialized terms like "AEO," "GEO," and "Citation" need to be defined as entities, just like services and products. Implement DefinedTerm schema on glossary pages and use those definitions consistently in body text, making it easier for AI to recognize your site as a definition source for specialized terms.

{
  "@type": "DefinedTerm",
  "@id": "https://sighted.tkbase.co.jp/glossary/aeo#term",
  "name": "AEO",
  "alternateName": "Answer Engine Optimization",
  "description": "The practice of optimizing content to be cited and recommended by AI answer engines such as ChatGPT and Perplexity.",
  "inDefinedTermSet": {
    "@type": "DefinedTermSet",
    "name": "Sighted AEO Glossary",
    "url": "https://sighted.tkbase.co.jp/glossary"
  }
}

Implementation Checklist

Created a domain entity table (class, canonical name, definition, aliases)
Defined inter-entity relationships with verbs and mapped to Schema.org properties
Used @graph to express Organization, Person, and Product relationships through nesting
All @id values use absolute URIs
Set sameAs for key entities (official URL, social profiles, Wikidata)
Replaced pronouns and abbreviations with canonical names in body text
Placed entity definition within the first 60 words of each section
Implemented DefinedTerm schema on glossary entries
Verified schema with the Google Rich Results Test
Confirmed nesting structure and Author binding are correct in Sighted audit
Measured Mention/Citation baseline with Otterly/Peec and recorded pre-intervention values

Next Steps

This methodology covers the workflow from ontology design to schema implementation to content editing. For individual implementation details, refer to the following knowledge articles.

Structured Data (Schema.org) Guide — Recommended and non-recommended types with implementation methods
GEO Complete Guide — Full framework for AI citation optimization
AI Visibility Monitoring Tools Guide — Impact measurement procedures
Citability — Optimal passage length and answer block design

Once validation progresses, these findings will be promoted to the knowledge base as contentType: "framework". If you have measurement data or counterexamples, please share your feedback.

Ontology-Driven AEO Content Editing — From Entity Design to Schema.org