How AI Search Engines Read Schema in 2026
Modern AI search engines (Google AI Overviews, AI Mode, ChatGPT Search, Perplexity) use JSON-LD structured data as a verifiable fact layer that supplements natural language. A 2026 analysis of 1,885 pages found that schema markup helps AI systems extract accurate claims, especially when combined with entity-based strategies (e.g., sameAs links to Wikidata). As of March 2026, Schema.org includes 823 types and 1,529 properties, with recent additions such as ConferenceEvent and the displayLocation property.
What Is Entity Authority?
Entity authority is the degree to which an AI system can confidently identify, classify, and cite your organisation as a distinct, trustworthy node in a knowledge graph. It is built through three reinforcing signals: structured data on your own site (sameAs, knowsAbout, citation, hasCredential), third-party corroboration (Wikidata, Wikipedia, authoritative citations), and behavioural evidence (being cited in academic papers, press, and government sources). An entity with high authority is unambiguous — AI systems resolve it without confusion and surface it proactively in generative answers.
The concept of entity authority moves beyond traditional link-based SEO into the domain of knowledge graph engineering. Where a backlink tells a search engine that one page references another, an entity signal tells it that a real-world thing — a company, a person, a concept — exists, has specific attributes, and is confirmed by independent sources. In 2026, this distinction is critical: AI Overviews and ChatGPT Search cite entities, not URLs.
JSON-LD Example — Entity Authority Schema
Below is a complete entity authority schema for an enterprise organisation, demonstrating sameAs to multiple authoritative sources, knowsAbout with Wikidata-linked Thing objects, citation, hasCredential, funding, and potentialAction:
{
"@context": "https://schema.org",
"@type": "Organization",
"@id": "https://www.acmecorp.com/#organization",
"name": "Acme Corporation",
"legalName": "Acme Corporation Ltd.",
"url": "https://www.acmecorp.com",
"description": "Global B2B industrial solutions specialising in
precision manufacturing and supply chain optimisation.",
"sameAs": [
"https://www.wikidata.org/wiki/Q12345",
"https://en.wikipedia.org/wiki/Acme_Corporation",
"https://www.crunchbase.com/organization/acme-corporation",
"https://pubmed.ncbi.nlm.nih.gov/search/?query=acme+corp"
],
"knowsAbout": [
{
"@type": "Thing",
"name": "Precision Manufacturing",
"sameAs": "https://www.wikidata.org/wiki/Q1064239"
},
{
"@type": "Thing",
"name": "Supply Chain Optimisation",
"sameAs": "https://www.wikidata.org/wiki/Q1137276"
},
{
"@type": "Thing",
"name": "Industrial Automation",
"sameAs": "https://www.wikidata.org/wiki/Q192793"
}
],
"citation": {
"@type": "CreativeWork",
"name": "Journal of Industrial Engineering",
"@id": "https://doi.org/10.xxxx/journal-article-acme",
"isPartOf": {
"@type": "Periodical",
"name": "Journal of Industrial Engineering",
"issn": "0022-0809"
}
},
"hasCredential": {
"@type": "EducationalOccupationalCredential",
"credentialCategory": "certification",
"name": "ISO 13485:2016 Medical Device Quality Management",
"recognizedBy": {
"@type": "Organization",
"name": "International Organization for Standardization",
"sameAs": "https://www.wikidata.org/wiki/Q190603"
}
},
"funding": {
"@type": "Grant",
"name": "UKRI Innovate UK Smart Grant 2024",
"funder": {
"@type": "Organization",
"name": "UK Research and Innovation",
"sameAs": "https://www.wikidata.org/wiki/Q60740228"
}
},
"potentialAction": {
"@type": "ViewAction",
"target": "https://www.acmecorp.com/about",
"name": "View Acme Corporation Profile"
}
}
sameAs: The Core of Entity Resolution
The sameAs property is the single most impactful entity signal available to organisations. It declares that your Organisation node is the same real-world entity as the resource at each linked URL. AI systems use these links to cross-reference facts: if Wikidata says Acme Corporation was founded in 1985 and your schema says the same, confidence in that fact increases significantly.
Priority targets for enterprise sameAs links:
- Wikidata (wikidata.org/wiki/Q…) — machine-readable, used by Google's Knowledge Graph and AI training data
- Wikipedia (en.wikipedia.org/wiki/…) — the most authoritative human-readable source; Wikidata and Wikipedia are tightly coupled
- Crunchbase — standard for technology and investment entities; used by ChatGPT and Perplexity for company data
- LinkedIn company page — corroborates industry classification and employee count
- GitHub organisation — signals open-source credibility for tech entities
- PubMed / DOI — for healthcare, pharma, and research organisations, citations in indexed journals are extremely high-authority signals
- Companies House / SEC EDGAR — regulatory registrations confirm legal existence
knowsAbout: Declaring Domain Expertise
The knowsAbout property allows an organisation to formally declare its areas of expertise. In 2026, AI systems use this property to determine whether a source is an appropriate authority to cite when answering a query. An organisation that knowsAbout "precision manufacturing" with a sameAs to the Wikidata concept is significantly more likely to be cited for queries on that topic than one that only mentions it in body text.
Best practice for knowsAbout:
- Use
Thingnodes withnameandsameAspointing to Wikidata Q-identifiers - Limit to 5–10 genuine areas of expertise — breadth without depth weakens the signal
- Align knowsAbout topics with your published content, credentials, and citations
- Update annually as your organisation's expertise evolves
citation: Evidencing External Recognition
The citation property links your organisation's schema to external creative works (papers, reports, articles) that reference it. This is distinct from backlinks — it is a structured, machine-readable declaration that your entity is discussed in a specific publication. For AI systems trained on academic and journalistic corpora, this is a powerful trust signal.
Use citation for:
- Peer-reviewed journal articles where your organisation is the subject or contributor
- Government or regulator reports that reference your work
- Industry association white papers
- News articles in publications with high domain authority
Always include the DOI or stable URL as the @id of the CreativeWork, and the ISSN of the Periodical where applicable. This allows AI systems to verify the citation independently.
hasCredential: Formalising Qualifications
Credentials are among the highest-trust entity signals because they are awarded by third parties and are independently verifiable. The hasCredential property on an Organisation node declares certifications, accreditations, and regulatory approvals. In YMYL sectors (medical devices, financial services, legal), credentials are often the deciding factor in whether an AI system cites your organisation for sensitive queries.
Common credential types for enterprises:
- ISO certifications (9001, 13485, 27001)
- Regulatory approvals (FCA, FDA, EMA)
- Industry accreditations (Investors in People, B Corp, BSI Kitemark)
- Professional body memberships (Law Society, RICS, CIPD)
AI Visibility Tips for Entity Authority
- Create a Wikidata entry for your organisation before implementing sameAs — the Q-identifier must exist to be referenced
- Use a persistent @id URI (e.g., https://www.acmecorp.com/#organization) so all pages reference the same entity node
- Include sameAs on both the Organisation schema and individual Person schemas for key executives
- Add knowsAbout to every topic you want to be cited as an authority on — align it with your published guides and research
- Cite your own publications in your schema using the citation property with DOI @ids
- Reference your ISO certifications and regulatory approvals in hasCredential with recognizedBy pointing to the awarding body's Wikidata entry
- Declare funding from government or public bodies (Innovate UK, NSF, Horizon Europe) using the funding/Grant pattern — it signals public interest and credibility
- Validate entity schemas using Google's Rich Results Test and Schema.org Validator after each update
- Monitor Knowledge Panel updates in Google Search to confirm entity recognition — panels typically appear within 2–4 months of consistent entity signals
- Submit your Wikidata Q-identifier to Google Search Console's Organisation Settings once available
Common Mistakes in Entity Authority Schema
Mistakes that undermine entity authority:
- sameAs links to dead or redirected URLs — broken sameAs links reduce confidence rather than increasing it. Audit them quarterly.
- Wikidata entry without minimum viable data — a stub Wikidata entry (name only, no instance-of, no founding date, no sitelinks) provides minimal entity resolution value. Populate at least: instance of (Q4830453 for business), country, founding date, official website, and Wikipedia sitelink if available.
- knowsAbout without Wikidata sameAs — listing topics as plain strings (just a name) is far weaker than linking to Wikidata Q-identifiers. Always include sameAs on Thing nodes.
- Inconsistent @id across pages — using different @id strings for the same organisation on different pages (e.g., #org vs #organization vs no @id) prevents entity graph merging. Choose one canonical @id and use it everywhere.
- Ignoring Person entity schemas — for AI systems, key executives and spokespeople are part of your organisation's entity graph. Named experts with their own sameAs signals (LinkedIn, Wikipedia) strengthen the parent organisation's authority.
- Claiming credentials you do not hold — AI systems cross-reference credentials against the awarding body's published lists. False credential claims damage trust and may trigger manual penalties.
- Neglecting non-English entity signals — multinational organisations should have Wikidata entries and sameAs links in all major markets, with localised Wikipedia articles where relevant. Entity authority is language-market specific.
Building Your Enterprise Entity Graph
Entity authority is not achieved through a single schema implementation — it requires building a coherent entity graph over time. The components of an enterprise entity graph are:
The Organisation Hub
A single, authoritative Organisation schema at your root domain (e.g., https://www.acmecorp.com/#organization) acts as the hub. All other entities — departments, products, people, locations — reference this hub via relational properties. This hub carries the sameAs links, legalName, taxID, leiCode, numberOfEmployees, foundingDate, and knowsAbout declarations.
Person Entity Spokes
Key executives, researchers, and spokespeople should have their own Person schemas with individual sameAs links (LinkedIn, Wikipedia where available, ORCID for researchers, Wikidata for public figures). These Person nodes connect to the Organisation hub via worksFor or employee properties. The more authoritative the individuals' entity signals, the stronger the association with the parent organisation.
Product and Service Nodes
Each major product line or service offering should be a schema entity in its own right — Product, Service, or SoftwareApplication — with its own @id and relevant properties. These nodes connect to the Organisation hub via the provider, manufacturer, or brand property. AI systems resolve product queries to entities, not pages, so product nodes with strong entity signals surface in generative answers independently.
Location Nodes
Physical offices, facilities, and service areas should be Place entities with GeoCoordinates and address data, connected to the Organisation hub via location or hasPOS. For international enterprises, each country operation may warrant its own LocalBusiness or subOrganization node to provide localised entity signals for regional AI systems.
Creating and Maintaining Your Wikidata Entry
Wikidata is the foundational external entity corroboration for AI search. A well-maintained Wikidata entry can accelerate Knowledge Panel creation, improve entity resolution in AI Overviews, and increase citation rates in ChatGPT and Perplexity by providing a neutral, machine-readable source of ground truth about your organisation.
Minimum Viable Wikidata Entry
- P31 (instance of): Q4830453 (business) or more specific (Q891723 for public company, Q6881511 for enterprise)
- P17 (country): appropriate country Q-identifier
- P571 (inception/founding date): ISO 8601 date
- P856 (official website): your root domain URL
- P18 (image): a logo or headquarters image (upload to Wikimedia Commons first)
- P1448 (official name): your full legal name
- Sitelinks: link to Wikipedia articles in all available languages
Advanced Wikidata Properties for Enterprise
- P1278 (LEI code): Legal Entity Identifier — verified by GLEIF
- P1278 (GLEIF ID): cross-reference to the Global LEI database
- P3347 (Companies House ID): for UK-registered entities
- P1278 (Crunchbase ID): P2626 is the Crunchbase organization ID property
- P101 (field of work): aligns with knowsAbout in your schema
- P1056 (product or material produced): for manufacturing organisations
- P1082 (number of employees): keep updated annually
Measuring Entity Authority Progress
Entity authority is an ongoing process. Track progress using these signals:
- Knowledge Panel presence: Does Google show a Knowledge Panel for your organisation name? Is the information accurate and current?
- AI Overview citation rate: For queries where your organisation should be cited, track how frequently Google AI Overviews include your brand (use an AEO monitoring tool or manual spot-checks)
- ChatGPT and Perplexity mentions: Test direct questions ("Who are the leading companies in [your domain]?") and check whether your organisation appears without prompting
- Wikidata completeness score: Use Wikidata's built-in constraint violation reporting to identify missing or incorrect properties
- Schema validation pass rate: Track the number of structured data errors and warnings in Google Search Console's Enhancements section — aim for zero errors
- Citation velocity: Monitor the rate at which new external sources (press, academic, government) cite your organisation — this is the organic growth signal for entity authority
Frequently Asked Questions
How long does it take to build entity authority for AI search?
Typically 3–6 months of consistent entity signals — sameAs links to Wikidata/Wikipedia, citations in third-party publications, and structured data on your own site. Knowledge graph recognition by Google's AI systems often follows within 2–3 months of a Wikidata entry being indexed and linked from your site's JSON-LD. ChatGPT and Perplexity typically respond faster because they update knowledge bases more frequently than traditional crawl cycles.
Can small companies build entity authority?
Yes. Small companies often achieve faster entity authority by focusing on a narrow domain. A specialist firm that is the definitive source on one topic — evidenced by citations, knowsAbout schema, and a Wikidata entry — can outperform larger generalist brands in AI answer inclusion for that topic. The key insight is that AI systems optimise for entity clarity and relevance, not organisational size. A well-structured entity graph for a 10-person consultancy can generate more AI citations in a specific domain than a Fortune 500 company with poor schema implementation.
Why is Wikidata so important for AI entity recognition?
Wikidata is a machine-readable, open knowledge graph used by Google, Wikipedia, and many AI training datasets. When your organisation has a Wikidata Q-identifier and your site's JSON-LD sameAs property references it, AI systems can resolve your entity unambiguously. It acts as a neutral, third-party confirmation of your existence and attributes. Unlike backlinks (which are directional and can be manipulated), Wikidata entries are community-maintained and require evidence to be created and sustained — which is precisely why AI systems weight them so highly.
What did the 2026 Ahrefs schema study find about entity authority?
The 2026 Ahrefs analysis of 1,885 pages found that pages with sameAs links to authoritative external sources (including Wikidata) were significantly more likely to be cited in AI Overviews and featured in generative answers. Entity clarity — having a unique, unambiguous identity confirmed by multiple external sources — was a stronger predictor of AI citation than page-level keyword optimisation. The study also found that organisations with structured knowsAbout declarations aligned to Wikidata Q-identifiers received citations for domain-specific queries at nearly twice the rate of those relying on natural language alone.