Entity SEO: make an answer engine say who you are, correctly
Entity SEO comes down to one structured data block, a short list of profiles you control, and writing your own name identically in seven places. Here is the block, the checklist, and how to see what an engine currently believes.
Entity SEO is the work of making a machine confident about who you are before it decides whether to quote you. It has almost nothing to do with keywords. It comes down to one block of structured data on one page, a list of links to profiles you already control, and writing your own name the same way everywhere. That last part is the one that fails.
Read this first
This chapter covers what you declare on your own site. What other people say about you off-site is a different job with different levers, and the two are easy to confuse. Keep the terms apart: being mentioned somewhere is not the same as being cited as a source.
You should already have structured data working in principle. Which types still earn anything is in schema markup for AI search. This chapter adds one type to that set and explains why it behaves differently from the rest.
Why entity SEO is a disambiguation job, not a keyword job
A search engine can rank a page without knowing what company published it. An answer engine has to attribute. To write "according to Acme" it needs to be sure which Acme, and its confidence comes from agreement between independent sources.
That is what the sameAs property is for. 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." You are not describing yourself. You are pointing at places that already describe you, so a machine can cross-check.
Structured data is not a claim about who you are. It is a set of addresses where that claim can be verified.
The property is widely deployed. Schema.org's own page reports usage on more than 10 million domains, based on Google's web index in July 2026. Deployment is not the same as being useful, and a sameAs list pointing at abandoned profiles is worse than none — it invites a check that fails.
Most sites stop one step short
We measured this on the 30 homepages we survey. Sixteen of 27 reachable homepages carried any JSON-LD at all, and twelve of those marked up an organization. Two of the twelve — Wired and Ars Technica — declared an Organization with no sameAs at all, which is the pattern this chapter is about: the block exists, and the part that lets a machine verify it is missing. The full numbers are in the organization schema survey.
Two publishers is not a trend, and we are not presenting it as one. What makes it worth naming is that both are large newsrooms with functioning engineering teams, which rules out the easy explanation. The declaration is the part a template generates; the profile list is the part somebody has to sit down and assemble, and it is the half that carries the verification.
Where the markup goes, and where it does not
One page, not every page. Google's guidance is explicit: "We recommend placing this information on your home page, or a single page that describes your organization, for example the about us page. You don't need to include it on every page of your site." Read 2026-08-18 on the Organization structured data documentation.
Google also says to "use the most specific schema.org subtype of Organization that matches your organization" — OnlineStore for a shop, SoftwareApplication alongside it for a product, and plain Organization when nothing more specific fits. A vague type is not wrong; it is just less information for the same effort.
The deliverable: a block you can paste today
Put this in the <head> of your homepage. Replace every value, and delete any sameAs line whose profile you cannot log into.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Acme Widgets",
"legalName": "Acme Widgets Limited",
"url": "https://www.acmewidgets.com",
"logo": "https://www.acmewidgets.com/logo.png",
"description": "Acme Widgets makes fastening hardware for furniture manufacturers.",
"foundingDate": "2019-04-01",
"sameAs": [
"https://www.linkedin.com/company/acme-widgets",
"https://github.com/acmewidgets",
"https://www.crunchbase.com/organization/acme-widgets",
"https://en.wikipedia.org/wiki/Acme_Widgets"
]
}
</script>Four sameAs entries beat twelve. Each one is a claim a machine can test, and each dead link is a failed test you volunteered for.
The deliverable: seven places your name must match
Run down this list once. Every row is visible from outside your company, which is exactly why a mismatch costs you.
| Where | What to check |
|---|---|
| Homepage title | The brand string, spelled and cased once, consistently |
| Visible h1 or wordmark | Same string as the title, not a stylised variant |
Organization.name | The name customers use, not the registered entity |
Organization.legalName | The registered entity, stated once, only if it differs |
og:site_name | Matches Organization.name exactly |
| Each sameAs profile | The profile's own display name matches too |
| Body copy and bylines | No informal short form used as if it were the name |
The seventh row is the one that quietly breaks the other six. A company called Acme Widgets that writes "at Acme" in every blog post has given a machine two candidate names, and a second entity called Acme somewhere in the world will happily absorb the ambiguity.
How to check what an engine currently believes
Three steps, none of which needs a paid tool.
- Fetch your homepage with curl and confirm the JSON-LD is in the delivered HTML, not injected by a script after load. A block a crawler never sees does nothing.
- Paste the page into Google's Rich Results Test and confirm the Organization block parses with no errors.
- Ask two answer engines to describe your company in one sentence, and write down what comes back. Repeat monthly and keep the answers. A wrong answer that changes is progress; a wrong answer that repeats is an entity problem.
Step three is deliberately soft. There is no verified measurement here — you are recording an impression over time, and that is worth something only because you wrote down the first one.
Do it with the same wording each month, and keep the answers in one file with dates. Two sentences a month is enough. The value is in having a record when someone asks whether the rebrand landed, because the alternative is arguing from memory about what an engine said in the spring.
Two ways this goes wrong
Both look like extra diligence and both make the situation worse, which is why they survive review.
- Putting the block on every page. It is not penalised, but repeating an organization declaration on a thousand pages does not strengthen it, and it makes a wrong value a thousand times harder to correct.
- Listing profiles you do not control. A directory listing someone else created, with a stale address, is not evidence of identity. It is a second entity with your name on it.
Where this chapter stops
None of this makes an engine cite you. It removes one specific failure — being confused with something else — and that is all it does. We cannot tell you how much any engine weights these signals, because none of them publish that, and anyone who gives you a number for it is guessing.
It also does nothing about what gets said off-site, which is the harder half of the problem and does not respond to markup at all. Checking that these seven places still agree, after a rebrand or a redesign quietly changes one of them, is what QueryWin is being built to do.
Common questions
What is entity SEO, in one sentence?
Making a machine resolve your brand name to one unambiguous organization, so that anything it reads about you accumulates in one place instead of splitting across two.
How many sameAs links should I have?
As many as you actively control, which for most companies is three to six. There is no documented threshold, and a longer list of dead profiles is worse than a short live one.
Does a Wikipedia page matter?
It is a strong reference because it is independent and machine-readable. You cannot create one for yourself, and treating that as a goal is how companies end up with a deleted article and a bad reputation among editors.
Will this get my brand into AI answers?
No. It makes correct attribution possible when an engine already has something to say about you. Being described accurately and being described often are separate problems.
Part of the QueryWin handbook · Level 2
