CiteFlow
Sign Up
← / Blog

Schema

schemaorganization-schemasameasentity-seo

Organization schema + sameAs: the entity backbone for AI citations

A practical guide to wiring Organization schema and 8-12 sameAs URLs so ChatGPT, Perplexity, and Gemini attribute citations to the right entity.

5 min read

TL;DR

  • Organization schema with a well-populated sameAs array is the cheapest, highest-leverage entity signal you can ship this quarter.
  • Aim for 8-12 sameAs URLs: Wikipedia, a Wikidata Q-ID, LinkedIn company page, Crunchbase, GitHub org, Bloomberg/OpenCorporates for legal entities, and platform profiles (YouTube, X, Facebook).
  • Publish one canonical Organization node at the root domain (/#organization), then reference it from every page via @id. Do not re-declare it per page.
  • This is entity disambiguation, not author E-E-A-T. Author byline schema (Person) is a separate object with its own sameAs.
  • Validate with Schema Markup Validator, then confirm the Wikidata round-trip: your site should be in the Wikidata item's "official website" (P856), and Wikidata should appear in your sameAs.

Organization schema is the JSON-LD block that tells crawlers "this website is operated by this entity." The sameAs property is the list of other authoritative URLs describing that same entity. Together they form the entity backbone that lets LLMs merge mentions of your brand across the open web, knowledge graphs, and their own training data into a single node they can confidently cite.

Why sameAs is disproportionately important for AI

Traditional search engines lean on links and click behavior to rank. Generative engines lean on entities. When ChatGPT, Claude, or Perplexity decide whether to cite "Acme Robotics" versus a competitor with a similar name, they need to resolve which Acme you are. sameAs gives them a directed graph of external corroboration — every URL you list is a claim of identity that the engine can cross-check against its own knowledge.

Google has documented sameAs as a knowledge-graph signal for years, and the Google Search Central documentation on Organization structured data still explicitly recommends it for entity disambiguation. The same field is consumed by generative engines that were trained on Common Crawl and post-trained on structured web data. If Wikidata already has an item for your company, linking to it in sameAs and getting your official site listed on the Wikidata item closes the loop — a pattern the Wikidata identifier documentation describes for reconciliation.

What belongs in the sameAs array

Quality beats length, but you should aim for 8-12 entries drawn from these tiers:

Tier 1 — knowledge graphs (must-have):

  • Your Wikidata Q-ID URL, e.g. https://www.wikidata.org/wiki/Q12345
  • Your Wikipedia article, if one exists (do not create a promotional stub — it will be deleted)

Tier 2 — business databases:

  • LinkedIn company page
  • Crunchbase profile
  • Bloomberg company page or OpenCorporates for the legal entity
  • Your ticker page on a financial data source, if public

Tier 3 — platform identity:

  • GitHub organization (critical for dev-tool and SaaS brands)
  • YouTube channel
  • X/Twitter profile
  • Facebook page

Tier 4 — vertical authorities:

  • G2, Capterra, or Product Hunt for SaaS
  • App Store / Play Store listings for mobile
  • IMDB, Discogs, or similar for media brands

Do not pad the array with UTM-tagged marketing URLs, blog author pages, or your own subdomains. Each entry should be a third-party page whose subject is unambiguously your organization.

The JSON-LD, with a stable @id

Publish this once on the homepage (or a /#organization fragment referenced from every page). Every other schema block on the site — Article, Product, WebPage — should reference the same @id in its publisher or provider field instead of re-declaring the org.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "@id": "https://example.com/#organization",
  "name": "Acme Robotics",
  "legalName": "Acme Robotics, Inc.",
  "url": "https://example.com",
  "logo": {
    "@type": "ImageObject",
    "url": "https://example.com/logo.png",
    "width": 512,
    "height": 512
  },
  "foundingDate": "2014-03-01",
  "founders": [
    { "@type": "Person", "name": "Jane Doe" }
  ],
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "500 Market St",
    "addressLocality": "San Francisco",
    "addressRegion": "CA",
    "postalCode": "94105",
    "addressCountry": "US"
  },
  "sameAs": [
    "https://en.wikipedia.org/wiki/Acme_Robotics",
    "https://www.wikidata.org/wiki/Q12345678",
    "https://www.linkedin.com/company/acme-robotics",
    "https://www.crunchbase.com/organization/acme-robotics",
    "https://github.com/acme-robotics",
    "https://www.youtube.com/@acmerobotics",
    "https://twitter.com/acmerobotics",
    "https://www.facebook.com/acmerobotics",
    "https://www.bloomberg.com/profile/company/ACME:US",
    "https://opencorporates.com/companies/us_de/1234567"
  ]
}

Then in an article on the same domain:

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "How six-axis arms outperform SCARA in cleanroom assembly",
  "publisher": { "@id": "https://example.com/#organization" },
  "author": { "@type": "Person", "@id": "https://example.com/team/jane-doe#person" }
}

The @id reference is what stitches your entity graph together. Without it, each page declares a fresh, disconnected Organization node.

Distinct from author byline schema

A common mistake is to conflate this with author E-E-A-T markup. Organization describes the publisher entity. Person schema on author pages — with its own sameAs pointing to the author's LinkedIn, ORCID, Google Scholar, Muck Rack, and personal site — describes the byline entity. Both matter, they serve different LLM disambiguation problems, and they should link to each other via worksFor and employee properties. If you have not shipped author schema, see the walk-through referenced in the CiteFlow schema archive.

Validation and the Wikidata round-trip

Three checks before you call it done:

  1. Paste the JSON-LD into the Schema Markup Validator. Zero errors, zero warnings on required fields.
  2. Run the page through Google's Rich Results Test to confirm Google parses the Organization node.
  3. Open your Wikidata item. Property P856 ("official website") should list your canonical URL. If it does not, add it. This closes the identity loop — Wikidata points at you, you point at Wikidata, and any engine querying either side gets a consistent answer.

Recrawl is not instantaneous. Expect two to eight weeks for the change to propagate through Google's Knowledge Graph and downstream LLM retrieval indexes.

FAQ

Does adding sameAs guarantee more AI citations?

No. It stabilizes attribution when you are already being mentioned. If nobody writes about your brand on the open web, sameAs cannot manufacture citations — it only ensures the mentions that do exist are correctly resolved to your entity.

How many sameAs entries is too many?

There is no hard cap, but past roughly 15 the marginal value drops fast and you risk diluting the signal with low-authority profiles. Prune anything that is not a third-party authority page about your organization.

Should subsidiaries have their own Organization schema?

Yes, if they have their own domain and public identity. Link them to the parent using parentOrganization and subOrganization properties, each with their own @id and sameAs array.

Sources

Related reading