Organization schema for AI: WebSite and sameAs

Ask an AI engine about a startup called Octo and it has a problem to solve before it can say anything useful: the name could mean an octopus, the prefix for eight, or any of several real companies that share it. Any system answering about a named brand has to settle which real-world thing the name refers to before it can attach a fact, a price, or a citation. Entity schema is how your site answers that question in a form machines can read: an Organization node for the brand, a WebSite node for the site, and sameAs links that tie both to external profiles confirming the identity. Google documents using it, and it costs a few lines of JSON-LD.

Which company are you?

A human reading about Octo works out from context whether the page means the animal, the prefix, or the company. A machine needs that context made explicit. Knowledge graphs are one established way to resolve ambiguous names: maps of real-world entities and the relationships between them, with Wikidata and Google's Knowledge Graph as prominent examples, and Wikipedia as the reference source such records commonly point at. Answer-engine vendors do not document whether or how they use them. In principle, a brand that resolves cleanly to one entity record is easier to attach facts to than one that does not. That is a mechanism you can reason about, not a measured effect on citations.

The stakes sit at the entity level, not the page level. A September 2025 study running controlled experiments across several product verticals, languages and paraphrased prompts (arXiv:2509.08919) reports: "Our key findings reveal that AI Search exhibit a systematic and overwhelming bias towards Earned media (third-party, authoritative sources) over Brand-owned and Social content, a stark contrast to Google's more balanced mix." In one of its experiments, fifty unbranded "best cola" style prompts put to ChatGPT and Perplexity returned major brands in about 62 percent of brand mentions, and across its language tests the set of brands an engine named held up better than the set of domains it cited. None of those numbers is a schema effect, and the study does not claim they are. What they show is that brand-level patterns in AI answers are more stable than domain-level ones, which makes the brand, not the URL, the useful unit of analysis. The study does not describe how any engine arrives at that. Being a resolvable entity is the job; entity schema is the part of that job you control on your own site.

How AI crawlers see your schema

Entity schema ships as JSON-LD inside a <script type="application/ld+json"> block. That placement matters more than it looks: the block sits in the raw HTML of the page, so any crawler that fetches the document gets the schema in the same request, even if it never executes a line of your JavaScript.

The failure mode is client-side injection. A December 2024 analysis by Vercel and MERJ measured the major AI crawlers, among them GPTBot from OpenAI, ClaudeBot from Anthropic, and PerplexityBot, and found that none of them render JavaScript at crawl scale: they fetch your HTML but never run your scripts. Schema that is injected client-side, by a React app after hydration or by a tag manager, exists only after rendering, which means those crawlers never see it. The same JSON-LD, moved into the server response, is visible to all of them.

The action is short: emit your Organization, WebSite, and sameAs markup in the server-rendered HTML, and confirm it is present when you view the page source with JavaScript disabled. Which specific engines do render is a separate topic; for identity markup the safe assumption is that the crawler reading your page runs no scripts at all.

Organization schema: your brand's identity

The Organization node is the machine-readable statement of who the brand is. Google recommends starting with name, url, and logo, and says of url that it "helps Google uniquely identify your organization". There are no required properties.

This is not an inferred benefit. Google's Organization documentation states: "Adding organization structured data to your home page can help Google better understand your organization's administrative details and disambiguate your organization in search results. Some properties are used behind the scenes to disambiguate your organization from other organizations (like iso6523 and naics), while others can influence visual elements in Search results (such as which logo is shown in Search results and your knowledge panel)." That is a search engine on record consuming this markup for identity: disambiguation behind the scenes, and the knowledge panel and logo in front of it.

The Organization node is also the anchor the rest of your markup points at: an Article or WebPage names it as publisher, so one well-formed brand node serves the whole site. A minimal version looks like this:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Octo",
  "url": "https://example.com/",
  "logo": "https://example.com/logo.png",
  "sameAs": [
    "https://www.linkedin.com/company/octo",
    "https://github.com/octo",
    "https://www.crunchbase.com/organization/octo"
  ]
}
</script>

WebSite schema: your site's identity

The WebSite node names the site itself with a name and a url. Google uses it to set the site name shown above a result, which is the site-level identity a reader sees before they see the page. Most CMS SEO plugins emit it automatically, which is why the gap shows up mainly on hand-rolled sites: they ship an Organization node and per-page Article markup, and leave the site-identity surface between them empty. Declaring both nodes closes it.

Do not bother with the optional SearchAction property on WebSite. It was the hook for Google's sitelinks search box, and Google removed that element from search results globally in November 2024 and retired the documentation. Existing markup is harmless and Google says there is no need to strip it out, but it buys nothing. Keep the WebSite node at name and url.

sameAs: anchoring your brand to the knowledge graph

sameAs is the property that connects your self-declared identity to the identities other systems already trust. schema.org defines it as the "URL of a reference Web page that unambiguously indicates the item's identity. E.g. the URL of the item's Wikipedia page, Wikidata entry, or official website." In Google's Organization docs the field reads: "The URL of a page on another website with additional information about your organization, if applicable. For example, a URL to your organization's profile page on a social media or review site. You can provide multiple sameAs URLs."

There is an adjacent measured result worth knowing, though it tested neither sameAs nor any commercial answer engine. A paper presented at ISWC 2024 (arXiv:2505.02737) tested knowledge graphs as an aid to language models resolving ambiguous names: "To mitigate these issues, Knowledge Graphs (KGs) have been proposed as a structured external source of information to enrich LLMs. With this idea, in this work we use KGs to enhance LLMs for zero-shot Entity Disambiguation (ED)." The paper reports that anchoring entities to a knowledge graph improved disambiguation accuracy. Read that result for what it is: a disambiguation mechanism, one step upstream of citation. It shows models identify which entity a mention refers to more reliably when a knowledge graph grounds them, not that a linked brand gets cited more often.

Which targets are worth listing:

• Wikidata and Wikipedia are the references schema.org itself gives as examples of pages that unambiguously identify an item, so list them if you genuinely have an entry.
• LinkedIn, Crunchbase, GitHub, and official social profiles give a matching reference on another domain. Most are yours to publish, so treat them as consistency, not as third-party confirmation.
• Most small companies will not have a Wikidata node or a Wikipedia article, and that is fine. List the profiles you genuinely have; never point sameAs at an entry that is not yours or does not exist.

Consistency is what makes the set work. Every profile you list should carry the same brand name and the same homepage URL, because matching details across those profiles reduce ambiguity. A dead sameAs target, a profile that now returns a 404 or 410, is a broken identity claim, and arguably worse than no claim at all: it points anything reconciling your identity at a dead end. Checking that the listed profiles are still live belongs on the same maintenance schedule as checking your outbound links.

One boundary worth keeping in view: in the passage quoted above, Google names iso6523 and naics, not sameAs, as the fields used behind the scenes to disambiguate. The disambiguation value of sameAs is an industry inference from how knowledge graphs reconcile entities, not something Google has confirmed. It remains a sound, schema.org-defined identity signal; it is not a Google-documented disambiguation lever.

What's documented, and what isn't

Two engines are on record. Google documents consuming Organization structured data to disambiguate your organization and to drive the knowledge panel and logo. Microsoft, in Bing's guidance on inclusion in AI search answers, recommends schema markup so machines can "interpret with confidence." ChatGPT, Claude, and Perplexity document nothing about reading your Organization, WebSite, or sameAs markup; any claim that they use it for citation is inference, however confidently it is stated elsewhere.

No controlled study isolates a citation lift from entity schema. The payoff you can actually count is machine clarity: Google's documented disambiguation and knowledge panel, Bing's documented preference for schema, and one consistent brand identity presented to every crawler that fetches your pages.

FAQ

Does Organization schema get me cited by AI engines?

There is no controlled study showing a citation lift, so treat that claim as unproven. What is real: Google uses the markup to disambiguate your brand and drive the knowledge panel, Bing is on record recommending schema, and it gives every engine a clean, machine-readable identity.

What is the difference between Organization and WebSite schema?

Organization describes the brand or company: its name, logo, and url. WebSite describes the site itself: its name and url, the identity an engine uses for a branded reference. Most sites should declare both.

What is sameAs and which links should it point to?

sameAs links your entity to external profiles that identify it; profiles you control add consistency, not independent confirmation. Wikidata and Wikipedia are the highest value; LinkedIn, Crunchbase, GitHub, and official social profiles add matching detail. Only link profiles that are real, live, and yours.

Do ChatGPT, Claude, and Perplexity read my Organization schema?

None of them documents doing so, so treat any benefit there as unproven. Google documents using Organization data in Search. Microsoft recommends schema generally for AI search, without naming these entity types. In every case the markup is at least present in the HTML any crawler fetches.

Where should the JSON-LD live?

In the raw server HTML. Most AI crawlers do not run JavaScript, so schema injected client-side can be invisible to them. Whether a given engine parses it is engine-specific.