Entity SEO, measured on 248 sites

Entity-based SEO, usually shortened to entity SEO, is the work of making a search engine understand what a site is about and who is behind it, not just which words appear on it. Nine Google searches for the term return the same few guides, and most of them teach one half of it: covering the topics around a subject. The other half, a site declaring its own identity in a form a machine can resolve, is the half Google documents, and nobody on those results pages counts who does it. We read 248 homepages and counted, in September 2026.

What is an entity in SEO?

An entity is a single, identifiable thing: a company, a person, a product, a place. Google introduced the idea to Search in May 2012, in a post titled Introducing the Knowledge Graph: things, not strings. It described “an intelligent model … that understands real-world entities and their relationships to one another: things, not strings”, holding “more than 500 million objects” at launch. Google’s Knowledge Panel Help now calls the Knowledge Graph “our database of billions of facts about people, places, and things”.

Google’s own example was the Taj Mahal. The word can mean a monument, a Grammy Award-winning musician or a casino in Atlantic City, and the Knowledge Graph keeps those apart. For a business, the question entity SEO asks is the same one: when a search engine or an AI assistant reads your name, can it tell which thing you are?

Entity SEO is two jobs

The term covers two different pieces of work, and the guides ranking for it mostly mix them.

1. Saying who you are. One Organization or Person on the site, with a stable identifier (@id), the site’s WebSite node pointing at it as publisher, and sameAs links to the profiles elsewhere that are also you. This is what Google’s Organization documentation describes.
2. Covering your topic’s entities. Writing about a subject so the related things, and how they relate, are stated on the page. This is topical coverage, often called semantic SEO.

Of the 25 ranking pages we could read across the nine searches, 15 talk in “topic cluster” or “topical” terms, and 7 mention sameAs at all. When we asked ChatGPT, Claude and Perplexity “entity based seo” on 25 September 2026, all three led with the topic job and none mentioned sameAs or @id. This page is mostly about the first job, because it is the one Google documents and the one that can be counted.

What Google documents about your own entity

Google’s page Organization structured data, last updated 8 September 2026, says adding it to a homepage “can help Google better understand your organization’s administrative details and disambiguate your organization in search results”. Four details from it are worth having exactly:

• “There are no required properties; instead, we recommend adding as many properties that are relevant to your organization.”
• “We recommend using the most specific schema.org subtype of Organization that matches your organization”, for example OnlineStore rather than OnlineBusiness for an ecommerce site.
• It goes “on your home page, or a single page that describes your organization … You don’t need to include it on every page of your site.” Claude’s answer told us to put it “on every key page”; that does no harm, and this site does it, but Google does not ask for it.
• sameAs is “a URL to your organization’s profile page on a social media or review site”. Google’s example is a profile, not Wikipedia.

On where Knowledge Graph facts come from, Google’s Knowledge Panel Help lists “public sources”, licensed data, and “content owners … who suggest changes to knowledge panels they’ve claimed”, and says panels “are created automatically … when there is enough information available on the open web”.

And for AI features, Google’s July 2026 guide is direct: “Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add.”

How many sites actually do it

We took every domain that ranked in a top ten or was cited by an AI Overview across the 75 US search results pages we captured for SEO, GEO, AI-search and ecommerce queries in September 2026. 248 homepages answered. Each was read the way our audit tool reads a site: which entity does it declare, and does that entity hold together?

Three quarters of these sites say who they are. A third say it in a form that holds together: one entity with an identifier, the site tied to it, and its profiles claimed. The gap between 76% and 32% is the gap between having an Organization block and having an entity.

Does the complete version help a site get cited? Not visibly in this data. Split by what happened on the search results pages, the share with all four is 30% for domains the AI Overview cited that never ranked, 31% for domains that ranked and were never cited, and 35% for domains that did both. The structured-data study found the same at the shallower level: the markup does not separate the two groups.

The sites teaching entity SEO

We read the homepages of the sites ranking for entity SEO the same way. 23 of 25 answered.

35% of the sites explaining entity SEO do all four, against 32% of the whole corpus. Ranking for entity SEO does not require doing it. That is a statement about homepages on one day, and some of these sites declare their entity on an About page instead, which Google’s documentation allows and this count does not see.

One oddity on these results pages: a support.google.com result ranks third or fourth on three of the nine searches. It is a question in Google’s Search Central Community forum, answered by two volunteers, not a Google answer.

The Wikipedia and Wikidata question

The AI Overviews for “entity based seo” and “google entity seo” both recommend linking your entity to Wikipedia or Wikidata with sameAs. Claude and ChatGPT listed them too, ChatGPT with “where applicable”.

15 of the 248 homepages have that link. They are ahrefs.com, apps.apple.com, coursera.org, hillwebcreations.com, hubspot.com, imd.org, insightland.org, sana-commerce.com, searchable.com, seranking.com, shopify.com, siteimprove.com, sproutsocial.com, victorious.com and wix.com: mostly brands large enough to have a Wikipedia article or public coverage.

The reason is in Wikidata’s own rules. Its notability policy accepts an item that links to a Wikipedia or other Wikimedia page, or “refers to an instance of a clearly identifiable conceptual or material entity that can be described using serious and publicly available references”. Otherwise, “your data unfortunately won’t be accepted”. For a small business, a Wikidata sameAs is an outcome of being covered in public, not a step on a checklist. What it can claim today is what Google’s own example names: profiles on social and review sites that it controls.

sameAs on you, and sameAs on a topic

schema.org defines sameAs as the “URL of a reference Web page that unambiguously indicates the item’s identity”. It is an identity claim, and where it sits decides what it claims.

On your Organization, it says “these profiles are also us”:

{
  "@type": "Organization",
  "@id": "https://example.com/#organization",
  "name": "Example Ltd",
  "sameAs": [
    "https://www.linkedin.com/company/example",
    "https://clutch.co/profile/example"
  ]
}

Inside an article’s about, it says “this topic is that Wikipedia entry”:

"about": {
  "@type": "Thing",
  "name": "Knowledge Graph",
  "sameAs": "https://en.wikipedia.org/wiki/Knowledge_Graph"
}

Both are correct. The guides that explain sameAs get it right; Inlinks and Search Engine Journal use the second form for topics. The AI Overview for “entity based seo” says “link your pages to authoritative external sources, like Wikipedia, using sameAs tags”, which is loose enough to suggest putting a topic’s Wikipedia page on your Organization. That would claim your company is the topic.

Our own entity graph, including when it broke

Elivate SEO’s homepage has all four: an Organization with the @id https://elivateseo.com/#organization, a WebSite naming it as publisher, and sameAs to LinkedIn, X and Clutch. No Wikipedia or Wikidata link, and none is claimed. Four things from running it:

• One list, two outputs. The profile list that feeds sameAs also renders the footer’s profile links, so the claim a machine reads and the one a person sees cannot disagree.
• Three sites, one person, and it drifted. elivateseo.com, ridamagrawal.com and easinvy.com each publish a Person node for the founder, and their sameAs lists have to match. X was added on one site on 10 September 2026, and the other two went six days without it before anyone noticed, by hand. A script now reads all three published sites and fails when they disagree.
• A profile removed. The YouTube channel came out of sameAs on 24 September 2026 when the firm stopped using it. A sameAs to a channel you do not run is a false claim, not a missing one.
• Our own audit read entities wrongly until 20 September 2026. It asked every site for an Organization by exact type, so a personal portfolio was told it had “No Organization node”. It now reads who the site says it is before judging it.

Whether Google has resolved our entity is a separate fact, and we have not measured it.

What to do, in order

Elivate SEO’s reading of Google’s documents and the census, for a business site:

1. Declare one entity on the homepage, using the most specific schema.org subtype, with an @id that never changes.
2. Point the site at it. The WebSite node’s publisher, and every article’s publisher, reference that @id, so there is one entity rather than several with the same name.
3. Claim only profiles that are yours, list the same ones everywhere the entity appears, and remove any that stop being true.
4. Earn third-party records — directories and review sites with your name and description — before thinking about Wikidata.
5. Then the topic job: cover the entities your subject involves, and state how they relate.

Our free audit reports which entity a site declares and whether its sameAs is populated. The GEO service is this list done for a business rather than described.

What this does not show

• That a complete entity graph improves ranking or citation. It does not separate cited from ranked in our data. Google says it helps disambiguation; we did not measure disambiguation.
• Knowledge Graph membership. Whether any of these 248 domains, ours included, has a Knowledge Graph entity was not checked.
• Anything off the homepage. An entity declared only on an About page is counted as absent, so these figures are a floor.
• Every guide. Search Engine Land’s, which ranks on seven of the nine searches, refused our fetcher and was not read.
• Other industries. The 248 are sites that rank or are cited for SEO, GEO, AI-search and ecommerce queries in the US. A population heavy on marketing sites may differ from others in either direction.
• The topic job. We have not measured topical coverage, and nothing here says how much it matters.

Why we wrote it

Because the pages ranking for entity SEO explain the concept well and count nothing, and the part of it Google actually documents gets a paragraph. We counted 248 sites and the sites teaching the subject, quoted Google’s documents where the advice and the documentation disagree, and showed our own entity graph, including the week it was wrong.