JSON-LD sameAs: How to Link Your Brand to Wikidata for AI Entity Disambiguation

JSON-LD sameAs is the JSON-LD property that tells AI systems which real-world entity your website represents. Without it, the string "Apple" on your homepage is ambiguous — AI models resolve it against the most likely referent, which for Apple happens to be correct (the technology company) but for a less-famous brand might collapse into a competitor, a historic predecessor, or an unrelated entity entirely.

Of 500 Fortune Global 500 companies, 122 publish JSON-LD. Of those, 55 include a sameAs array. And of those 55, only 3 link to Wikidata: Apple (Q312), Comcast (Q1113804), and Repsol (Q174747). The entity disambiguation gap is 99.4% wide.

This guide walks through the 10-minute fix, using Apple's production pattern as the reference.

Step 1 — Find your Wikidata QID

Visit https://www.wikidata.org/ and search for your company name. Every notable entity has a canonical identifier starting with Q, followed by digits. Apple is Q312. Anthropic is Q104041691. Volkswagen is Q246. The QID appears in the URL of the entity's Wikidata page.

If no Wikidata entry exists for your company, create one first. Wikidata is open; anyone can add a well-sourced entity. This is a one-time ops task that takes about 30 minutes and requires citing 2-3 independent sources (company website, SEC filing, mainstream news coverage).

Step 2 — Add sameAs to your JSON-LD

Your existing JSON-LD Organization or Corporation block probably already has a sameAs array pointing to LinkedIn, X, and Wikipedia. Add one more line — the Wikidata URL — and redeploy. That is the entire fix.

Apple's production JSON-LD at apple.com includes (paraphrased structure):

{
  "@context": "https://schema.org",
  "@type": "Corporation",
  "name": "Apple Inc.",
  "url": "https://www.apple.com",
  "logo": "https://www.apple.com/apple-touch-icon.png",
  "sameAs": [
    "https://www.wikidata.org/entity/Q312",
    "https://en.wikipedia.org/wiki/Apple_Inc.",
    "https://www.linkedin.com/company/apple/",
    "https://x.com/Apple"
  ]
}

The critical line is "https://www.wikidata.org/entity/Q312". Everything else is either already in your markup or is SEO hygiene.

Step 3 — Verify

Test your deployed JSON-LD with Google's Rich Results Test (search.google.com/test/rich-results). Paste your URL, wait for the parse, confirm the Organization or Corporation block is detected and the sameAs array contains the Wikidata URL.

Then verify from an AI model directly. Ask ChatGPT or Claude: "What is the Wikidata QID for [your company name]?" If the model can answer correctly, the disambiguation is working.

Why this matters more for AI than for SEO

Google's Knowledge Graph uses Wikidata sameAs as a signal but also has its own entity resolution pipeline — your content can rank in search without it. AI answer engines have less infrastructure. When ChatGPT decides whether to cite "Apple" in an answer about iPhones, the model's ability to unambiguously identify your brand determines whether your content shows up. A sameAs to Wikidata is the single highest-signal, lowest-effort disambiguation you can add.

The entity disambiguation gap data shows this opportunity is almost entirely unclaimed. If you're reading this and implement the fix this quarter, you join a club of 4 Fortune 500 companies instead of 3.

Extended pattern: the full identity layer

Beyond the basic Wikidata sameAs, Apple's JSON-LD also includes:

• legalName — "Apple Inc." (distinguishes the corporation from its brand)
• numberOfEmployees — the strongest size-disambiguation signal (only 8.7% of Fortune 500 include this)
• foundingDate — 1976 (disambiguates from earlier companies named "Apple")
• founders — Steve Jobs, Steve Wozniak, Ronald Wayne as Person entities
• address — Cupertino, CA (geographic anchoring)

Each field makes the entity more precisely resolvable. But sameAs to Wikidata does 80% of the work by itself. The other fields are the polishing.