Entity SEO for AI search: making sure assistants know who you are

What “entity” means for AI search

An entity is a distinct, identifiable thing, like an organization, a person, a product or a place, with attributes and relationships. Search engines have modeled the web this way for years. Google’s Knowledge Graph is the best-known example.

AI assistants add two layers. A model’s training data shapes what it “knows” about you without searching, and search-grounded modes retrieve live pages to answer a specific question. Vendors don’t document exactly how either layer resolves which “Example Co” a user means, or how it weighs conflicting sources. What we can observe is simple. When the facts about an organization are consistent and easy to find across reputable sources, assistants have less room to get them wrong. When they conflict, you get blended, outdated or invented details.

So the job is unglamorous: reduce ambiguity everywhere your organization is described.

Step 1: Write the canonical entity sheet

Before touching markup, agree internally on one source of truth. Put it in a shared doc that marketing, sales, HR and whoever manages listings can all reach.

1. Names: brand name, legal name and any former or alternate names.
2. One-sentence description: what you are, for whom and where. Reuse it word for word.
3. Category: the plain-language category you want to be known for, such as “AI SEO consultancy”, not a slogan.
4. Location data: address, phone number and service area, formatted exactly one way.
5. People: founders and key leaders, with their correct titles.
6. Dates and facts: founding year, plus anything else you expect people to ask about.
7. Official profiles: the URLs of every profile you control.

Step 2: Make your own site the clearest source

Your site should be the most complete and unambiguous description of you anywhere on the web. Here’s what to check:

• An About page that states the entity sheet facts in plain text, not only inside images or videos.
• A contact page and footer with name, address and phone number that match the sheet exactly.
• Leadership or team pages for the people you listed, each with a short, factual bio.
• Server-rendered HTML, so crawlers that don’t run JavaScript still see these facts. See structuring content so LLMs can quote it.

Then express the same facts in schema.org markup. A consultancy with a physical office can use a LocalBusiness subtype such as ProfessionalService. Since that’s also a subtype of Organization, Organization properties like logo and sameAs are valid on it. An online-only company would use Organization.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "ProfessionalService",
  "@id": "https://example.com/#organization",
  "name": "Example Analytics",
  "legalName": "Example Analytics LLC",
  "url": "https://example.com/",
  "logo": "https://example.com/assets/logo.png",
  "description": "Example Analytics is a Los Angeles analytics consultancy for in-house marketing teams.",
  "telephone": "+1-213-555-0100",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "123 Example St",
    "addressLocality": "Los Angeles",
    "addressRegion": "CA",
    "postalCode": "90012",
    "addressCountry": "US"
  },
  "founder": { "@type": "Person", "name": "Jane Doe" },
  "sameAs": [
    "https://www.linkedin.com/company/example-analytics",
    "https://www.crunchbase.com/organization/example-analytics",
    "https://www.youtube.com/@exampleanalytics"
  ]
}
</script>

Three details matter more than they look:

1. A stable @id. Use the same identifier on every page, and reference it from Article publisher and Product brand markup, so your structured data describes one connected entity instead of many disconnected ones.
2. sameAs only for profiles that are unambiguously you. Don’t list directory pages you don’t control, or someone else’s article about you.
3. Markup must match visible content. Validate with the Schema Markup Validator and Google’s Rich Results Test, then compare it to the page by eye.

Be realistic about schema’s role. Google documents how it uses Organization and LocalBusiness markup in Search. AI assistant vendors haven’t documented whether or how they read JSON-LD. We treat it as a consistency layer, not a lever.

Step 3: Align the profiles that describe you

Third-party sources often outweigh your own site when it comes to corroboration. Work through them against the entity sheet:

• Map and business listings: Google Business Profile, Bing Places and Apple Business Connect, where relevant. Match name, category, address, phone and hours.
• Professional and company databases: LinkedIn, Crunchbase, and industry associations or directories that matter in your field.
• Wikidata: create or edit an item only if your organization meets Wikidata’s notability policy, and cite independent references. Wikipedia has strict conflict-of-interest rules. Don’t write your own article.
• Your people: founders’ LinkedIn titles, speaker bios and author pages should name the company the same way.

Practical tip: track every profile in a spreadsheet with columns for URL, owner, login location, last verified date and any mismatch with the entity sheet. Entity drift happens through dozens of small edits nobody remembers making.

Step 4: Handle collisions and old names

If your brand name is shared with another company, a place or a common word, assistants have to work out which one a user means. Help them:

1. Pair your name with your category and location in titles, descriptions and the opening line of your About page. “Example Analytics, a Los Angeles analytics consultancy” is harder to confuse than “Example”.
2. Record former names with alternateName and say plainly on the About page when you renamed.
3. 301-redirect old domains, and update old profiles instead of abandoning them. A stale listing with an old phone number keeps feeding stale answers.

Step 5: Test what assistants actually say

You can’t inspect an assistant’s knowledge directly, but you can sample it. Build a short, fixed prompt set and run it in each assistant you care about, with and without web search where that option exists:

• “What is [brand]?” and “Who founded [brand]?”
• “Where is [brand] located and what does it do?”
• “Recommend [category] providers in [city]”, to check whether you appear and how you’re described.

Log every answer, with the date, the assistant, the mode, the cited sources and any factual errors. When an error appears, follow the citations. The cause is often a specific outdated profile or page you can fix. Answers from a model without search may reflect older training data and change only when the model is updated. Our guide to tracking AI referrals and citations covers the logging setup in more detail.

Keep it maintained

Entity work isn’t a one-time fix. Re-verify the sheet and profiles every quarter and after any rebrand, move or leadership change. Also make sure the crawlers you rely on can reach the pages that state these facts. The AI crawlers and robots.txt playbook covers access.

If you’d like us to build the entity sheet, fix the markup and chase down the conflicting profiles, see our services, or book an audit call.