When an AI assistant describes your company, our reading is that it assembles the description from what it can find and reconcile. If your name, address, founder and profiles disagree across the web, the machine has to guess. Entity SEO is the work of removing the guesswork.
This article covers the basics for a single company. It uses Google’s structured data documentation and the schema.org vocabulary, and nothing else.
What is an entity in SEO?
An entity is a thing that can be identified without ambiguity: a specific company, person or product. “Acme” is a string. “Acme Ltd, registered in England, founded by Jane Doe, at acme.example” is an entity. Search engines and AI systems work with the second kind. Your job is to make the second kind easy to recover from the first.
Where should the Organization markup go?
On one page, usually the home page. Google’s Organization documentation says “we recommend placing this information on your home page” or on a single page about the organisation. It adds that “you don’t need to include it on every page of your site”.
Pick that page deliberately. For most companies it is the home page, with the About page as a close second. Every fact in the markup must also be visible on the page. Google’s general policy on structured data requires that, and so do we.
The properties worth filling in
Google says “there are no required properties; instead, we recommend adding as many properties that are relevant to your organization”. Here is what each does, in the order we would prioritise it.
name and url. Use the same name everywhere. Google says the url “helps Google uniquely identify your organization”. You may trade under a short name and be registered under a long one. If so, put the trading name in name and the registered one in legalName. The schema.org definition of legalName is “the official name of the organization, e.g. the registered company name”.
logo. Google says the image must be at least 112x112px and must be crawlable and indexable. It helps Google choose which logo to show in results and knowledge panels.
sameAs. This is the property that does most of the entity work. schema.org defines it as the “URL of a reference Web page that unambiguously indicates the item’s identity”. Google says “you can provide multiple sameAs URLs”. List your official profiles: LinkedIn, Crunchbase, Wikidata if you have an item, and any registry page. Do not list pages that are not about you.
address, email, telephone, contactPoint. Google asks you to include every address property that applies to your country. It also asks for country and area codes in telephone numbers. Match the format you use on the Contact page exactly.
founder and foundingDate. schema.org defines founder as “a person or organization who founded this organization”. Google wants foundingDate in ISO 8601 format. If the founder has a profile page on your site, link to it with a stable @id.
Identifiers. Google lists vatID, taxID, iso6523Code, leiCode, duns, globalLocationNumber and naics as identifiers that disambiguate an organisation. It calls vatID “an important trust signal for users”. These are the hardest facts for a machine to confuse with another company, so add the ones you legitimately hold.
Give your people a profile page
Companies are often identified through their founders. Google’s ProfilePage documentation describes markup “designed for any site where creators (either people or organizations) share first-hand perspectives”. The mainEntity is the Person or Organization the page is about.
For a founder page, Google recommends alternateName for a public handle and description for a byline or credential. It also recommends image for a photo and sameAs for external profiles. The pattern is the same as for the organisation. One page, one entity, consistent facts, links out to the profiles that confirm them.
Then connect the two. The organisation’s founder should reference the person, and the person’s worksFor should reference the organisation. Use the same @id values across the site so every page contributes to the same graph rather than creating a fresh copy.
Organization
name, url, legalName, logo, description, with one @id such as https://example.com/#org
Home or About page
States every fact once, in visible text
Founder (Person)
ProfilePage, founder ↔ worksFor, its own @id
sameAs profiles
Only official pages about this organisation
Contact facts
address, email, telephone, matching the Contact page
Identifiers
vatID, leiCode, duns: the hardest facts to confuse
Parent company
parentOrganization, with its own profiles on its own node
Why do the profiles outside your site matter?
Because they are how a machine confirms what your markup says. Markup tells machines what you say about yourself. Confirmation comes from elsewhere. Go through every profile listed in sameAs and check the name, URL, description, address and founder match your page. Take a LinkedIn company page with a different founding year, or a registry entry under a former name. In our experience, that is the kind of conflict that produces a wrong sentence in an AI answer.
If you have a Wikidata item, check it with the same care. If you do not have one, do not create one for yourself. Self-created items may be deleted under Wikidata’s notability policy.
A checklist
Work through this once, then repeat whenever a fact changes.
- One page carries the
Organizationmarkup, and every property is also visible on that page. namematches the name used in the site title, footer and profiles.legalNameis set if the registered name differs.urlis the canonical home page URL with the same trailing slash you use everywhere.logois at least 112x112px, crawlable and indexable.sameAslists only official profiles about this organisation, each checked for matching facts.address,telephoneandemailmatch the Contact page character for character.founderreferences aPersonwith a stable@id, and that person’s page carriesProfilePagemarkup.- At least one legal identifier is present if you hold one.
- Nothing in the markup claims an award, rating, client or figure that the page does not show.
What we saw on our own properties
On four of the eight sites in our case studies, the page Google showed most often in Search Console’s Generative AI performance report was a page that says who the company is:
| Site | Top page in AI features | Its impressions | All AI-feature impressions |
|---|---|---|---|
| bagassepackaging.co.uk | Home page | 2,220 | 6,008 |
| bamboopaper.in | About page | 727 | 2,233 |
| castingforgingsource.com | Home page | 662 | 1,486 |
| singlecrystaldiamondring.com | Home page, then the story page (202) | 259 | 541 |
On bamboopaper.in the About page led by a wide margin, ahead of the shop (276) and the home page (257). Search Console does not say why a page was shown, so this is a pattern, not a cause. It is consistent with the advice above: the page that states the facts is a page these features use.
What we don’t claim
Clean entity data makes it easier for machines to describe you correctly. It does not control what they say, and it does not make a knowledge panel or an AI mention appear. Google’s documentation describes these properties as recommended, not as requirements for any feature. We make no promise beyond this: when your facts are unambiguous and confirmed, the machine has less room to get them wrong.