Skip to main content
friendly4AI LogoMaking websites AI-friendly - Your website optimization platform for AI systemsfriendly4AI
  • Product
  • Leaderboard
  • Pricing
  • Blog
Sign in
  1. Methodology
  2. ›
  3. Parameters
  4. ›
  5. Entity Grounding

Entity Grounding

stable

Category: entity · Methodology v4.5

Entity grounding measures how clearly your site declares who it is to AI knowledge systems.

Signal Source

Source
https://{domain}
Kind
html_dom

Score Bands

VerdictCondition
PassOrganization or Person JSON-LD schema is present with sameAs containing two or more authoritative targets (Wikipedia, Wikidata, or LinkedIn)
PartialOrganization or Person JSON-LD schema is present with exactly one authoritative sameAs target
Failno Organization or Person schema found, sameAs property absent, or all sameAs values point to non-authoritative targets

Description

Entity grounding checks whether your homepage links its real-world identity to authoritative knowledge graphs through JSON-LD sameAs. friendly4AI passes a domain (100) when an Organization or Person schema lists two or more sameAs targets on Wikipedia, Wikidata, or LinkedIn. It scores partial (50) for exactly one such target, and fail (0) when no Organization or Person schema exists, when sameAs is absent, or when every sameAs value points to a non-authoritative target.

What does entity grounding measure?

Entity grounding measures how clearly your site declares who it is to AI knowledge systems. friendly4AI scans the homepage HTML for JSON-LD blocks whose @type is Organization or Person, then reads the sameAs property — an array or string of external URLs that point to the same entity elsewhere on the web. Each sameAs value is checked against three authoritative entity-graph targets:

  • Wikipedia — wikipedia.org, any language edition
  • Wikidata — wikidata.org Q-number URLs
  • LinkedIn — company pages at linkedin.com/company/ or profile pages at linkedin.com/in/

How many authoritative targets you have determines the tier. If no Organization or Person schema is present at all, the parameter scores zero no matter what other structured data sits on the page — see structured-data and structured-data-coverage.

Why does entity grounding matter for AI-readiness?

AI language models build entity knowledge by linking mentions to knowledge-graph entries — mainly Wikidata and Wikipedia — and to social-graph anchors like LinkedIn. Add sameAs links to those sources in your Organization schema, and any AI system reading your homepage can resolve your entity, confirm your identity, and trust what it cites about your organisation. Drop the grounding, and the model has to guess your identity from plain text. That gets ambiguous fast, especially when your name is a common one. Solid entity grounding sharpens citation accuracy across ChatGPT, Gemini, Claude, Perplexity, and Bing Copilot, since all of them query entity graphs when they assemble factual answers. For Person entities, the same signal feeds author-authority.

How is entity grounding scored?

Under the v4.5 methodology, this Entity and Schema parameter scores on a three-tier gradient:

  • Pass (100) — an Organization or Person JSON-LD block is present and sameAs holds at least two authoritative targets from Wikipedia, Wikidata, or LinkedIn.
  • Partial (50) — the schema is present and sameAs holds exactly one authoritative target. Grounding exists, but it is incomplete.
  • Fail (0) — no Organization or Person schema is found, sameAs is absent, or all sameAs values point to non-authoritative targets (for example, only your own social-media profiles or internal pages).

The count combines all Organization and Person blocks on the page.

How do I fix entity grounding?

  • Add a JSON-LD Organization block to your homepage <head> (or inline it as a <script type="application/ld+json">) with a sameAs array that lists your Wikipedia article URL, your Wikidata entity URL (format: https://www.wikidata.org/wiki/Q<number>), and your LinkedIn company page URL.
  • No Wikipedia article yet? Create a Wikidata entity first. Wikidata accepts entries for organisations that have no Wikipedia article and issues a Q-number on the spot.
  • Keep the sameAs values canonical: the exact Wikipedia page URL including its language subdomain, the wikidata.org/wiki/Q… form, and the linkedin.com/company/your-slug format.
  • The pass tier needs two or more authoritative links. To move from partial to pass, the smallest fix is a second authoritative target next to the one you already have.
  • Before you re-scan, validate the full JSON-LD block with Google's Rich Results Test and confirm it parses without errors.

Version History

Introduced
v4.5
Last changed
v4.5

Key takeaways

  • Signal: https://{domain}
  • Category: Entity & Schema
  • Passes when: Organization or Person JSON-LD schema is present with sameAs containing two o…

Related Parameters

  • Structured Data
  • Structured Data (schema.org/JSON-LD) Coverage
  • Author Authority Signals

View full methodology changelog · All parameters · GEO/AEO glossary

Suggest a change to this parameter


friendly4AI LogoMaking websites AI-friendly - Your website optimization platform for AI systemsfriendly4AI

The starting point for making your website AI-friendly.

ai@friendly4.ai
ProductsGEO ScannerAI VisibilityMethodologyDemoLeaderboardHow we comparePricing
friendly4AIAbout usBlogFor developersQuickstartAPI referenceContact usFAQsGlossary
LegalTerms and ConditionsPrivacy PolicyAI usage policy
friendly4AI © 2026