Entities, schema & sameAs

ExpertDuration ~50 min hands-onTools Google Rich Results Test, Schema Markup Validator (validator.schema.org), Wikidata, A text editor

Why this lesson

An AI answer engine doesn’t just read your words — it tries to work out what thing your brand is and whether it can trust it. That “thing” is an entity: a person, company, product or place the model can recognize, tell apart from everything with a similar name, and connect to what it already knows. Two businesses called “Apex Roofing” exist in every country; a page of prose can’t separate them, but a machine-readable identity can. This lesson is how you hand the engine that identity on a plate. It’s built almost entirely from structured data, and one property inside it — sameAs — is the highest-leverage single addition you can make for GEO. Get this right and you stop being an anonymous string of text and become a known entity the model is willing to cite by name.

The explainer

Entities and the knowledge graph

An entity is a distinct, identifiable thing — your company, your founder, a product — that a search or answer engine can reason about as a unit, not as a bag of keywords. Entities live in a knowledge graph: a giant network of entities and the relationships between them (Google’s Knowledge Graph, Microsoft’s Satori, and the open Wikidata graph that feeds them all). When the graph “knows” your entity, an AI model inherits that knowledge for free. When it doesn’t, you’re a guess.

The work of getting into the graph is called entity linking — explicitly connecting the entity on your site to its established node elsewhere so the engine can match the two and merge what it knows. You don’t do that with clever copywriting. You do it with schema.

Organization and Person schema

Structured data — JSON-LD, the machine-readable script you met in Level 2 for rich results — is the tool here, but we’re using it for a different job: to build an identity. Two types do the heavy lifting:

• Organization schema describes the business itself — legal name, logo, URL, contact point, and its profiles across the web.
• Person schema describes the humans behind it — an author or founder — with jobTitle, knowsAbout (the topics they’re an authority on), and their own profile links. This is your author identity signal, and it’s what powers E-E-A-T for AI: a real, verifiable expert attached to the content, not an anonymous byline.

Keep one rule from the 2025–26 currency notes in mind: schema must be server-rendered. JavaScript-injected JSON-LD faces delayed processing at Google and is effectively invisible to AI crawlers that don’t run JS (we prove this in 3.6). Also skip the tempting types that have been cut back — HowTo rich results were removed in 2023, and FAQPage rich results are now restricted to government and health-authority sites. They won’t hurt as plain data, but don’t build your strategy on them.

sameAs — the single highest-impact GEO addition

Inside Organization and Person schema sits one array: sameAs. It’s a list of URLs pointing at the same entity on authoritative third-party platforms:

"sameAs": [
  "https://en.wikipedia.org/wiki/Your_Company",
  "https://www.wikidata.org/wiki/Q12345678",
  "https://www.linkedin.com/company/your-company",
  "https://www.youtube.com/@yourcompany",
  "https://www.crunchbase.com/organization/your-company"
]

This is the disambiguation move. It tells the engine “the Apex Roofing on this page is this Apex Roofing — the one on Wikipedia, the one that is Wikidata entity Q-whatever, the one with this LinkedIn.” A Wikipedia or Wikidata link is the strongest signal in the list because it plugs you straight into the knowledge graph. Aim for five or more links, and lead with the ones the engines already trust: Wikipedia, Wikidata, LinkedIn, YouTube, Crunchbase. This is why sameAs completeness is weighted so heavily in the Schema Score — it’s the cheapest, highest-return line of code in GEO.

speakable

One more property worth adding: speakable. It marks the specific sentences on a page — via CSS selectors or an xpath — that are safe to read aloud as a spoken answer. Voice assistants and some AI surfaces use it to pick a clean, self-contained snippet instead of guessing. It’s a small win, but it pairs perfectly with the citable answer blocks you wrote in 3.2: mark your best 40–60 word answer as speakable and you’ve told the engine exactly which line to quote.

Put it together and you’ve done something structural: converted a website into a recognized entity with a verifiable human behind it and explicit links to the knowledge graph. That’s the foundation every per-engine tactic in the next lesson builds on.

Do it

1. Pick your real (or practice) site. Open its homepage source and check whether Organization schema already exists — search the raw HTML for "@type":"Organization".
2. Write an Organization JSON-LD block: name, url, logo, contactPoint, and a sameAs array with every authoritative profile you own — LinkedIn, YouTube, Crunchbase, Facebook, and (if they exist) your Wikipedia page and Wikidata Q-ID. Get to five or more links.
3. If you have a named author or founder, add a Person schema block with jobTitle, knowsAbout, and their own sameAs (LinkedIn, personal site, author profiles).
4. Add a speakable property pointing at your single best answer sentence on the page.
5. Paste it into validator.schema.org and the Rich Results Test. Fix every error and warning.
6. Confirm the schema is in the raw HTML, not injected by JS. If it’s JS-injected, note it as a fix for lesson 3.6.
7. Don’t want to start from a blank file? The course ships six fill-in JSON-LD starter templates (Organization, LocalBusiness, Article-author, Product, Software/SaaS, WebSite+SearchAction) — grab them from the Level 3 workbook and edit rather than write from scratch.

Level 3 workbook — entity checklist, sameAs link list & the 6 JSON-LD schema starter templateslevel-3-workbook.pdf113 KBOriginal course material — free to use

Terms introduced

Terms introduced in this lesson

• Entity
• Knowledge graph
• sameAs
• Wikidata
• Entity linking
• speakable

Each term links to the glossary, where it is defined by the way you just used it.

Check

Check yourself

1. What is the single most impactful schema addition for getting recognized by AI answer engines?

   A long meta descriptionThe sameAs property, linking your entity to authoritative profiles like Wikipedia, Wikidata and LinkedInA keyword-stuffed alt attributeFAQPage schema on every page

2. Why is a Wikidata or Wikipedia link in sameAs so powerful?

   It is required for HTTPSIt automatically earns you backlinksIt boosts your page speedIt connects your brand to a node in the knowledge graph, letting engines disambiguate you from similarly-named things and trust the connection

3. Your Organization schema is injected into the page by JavaScript after load. What is the risk for AI visibility?

   It makes the schema more trustworthyAI crawlers generally do not execute JavaScript, so JS-injected JSON-LD is invisible to them (and faces delayed Google processing) — schema must be server-renderedNone — all crawlers run JavaScriptIt only affects mobile devices

Check answers

You can move on when you can… describe what an entity and a knowledge graph are, write valid server-rendered Organization + Person schema for a site, and populate a sameAs array with five or more authoritative links including Wikipedia/Wikidata.

Mark this lesson complete

Go deeper

• Google — “Introduction to structured data markup” & the sameAs reference on schema.org: the canonical definitions for the properties you just wrote.
• Wikidata — “Help:Items”: how entity nodes and Q-IDs work, so you understand exactly what you’re linking to.
• Next: 3.5 · Optimizing per engine — take this entity and tune it for each AI platform’s specific preferences.