Home » AI Search (AEO) » Entity Relationship Graph: Mapping Who and What Your Website Connects

Entity Relationship Graph: Mapping Who and What Your Website Connects

Published 2026-10-04 | AI Search (AEO)

A website isn't only a pile of pages. Read carefully, it contains statements: this company was founded by that person, this product belongs to that category, this service is offered in these cities. Each statement connects two things. Gather enough of them and you have a graph, a map of who and what your site talks about and how they relate.

The reason to build one is practical. Search engines and AI assistants build their own version of this map from your content. If your pages state the relationships clearly, they'll draw it correctly. If you leave them to infer, they might guess wrong or skip you.

What a relationship looks like

Subject Relationship Object
Acme Consulting founded by Priya Shah
Acme Consulting headquartered in Leeds
Priya Shah works for Acme Consulting
SiteCheck is a product of Acme Consulting
SiteCheck belongs to category Website audit software
SiteCheck compared with Competitor tools
Blog post written by Priya Shah

Step 1: list your core entities

Start with the obvious ones: the company, its main people, products or services, locations and the main topics you cover. Keep the list to what matters. Twenty entities mapped well beats two hundred mapped poorly.

Step 2: extract relationships from your content

Read your key pages and write down every explicit relationship you find. The about page, product pages, team page and author bios are rich sources. You can use an entity extraction tool to suggest candidates, then confirm them by hand, because automated extraction makes mistakes with names and roles.

Step 3: find the gaps

Look at each entity and ask which obvious connections are missing from your pages. Common ones:

  • A product page that never names the company behind it
  • Articles without an author link
  • An author without a bio that states their role and employer
  • A service page that doesn't mention where you operate
  • No page that says what category your product belongs in

Step 4: say it in plain text first

State each relationship in a normal sentence on a page that people will read. "SiteCheck is a website audit tool made by Acme Consulting, a company based in Leeds." Systems learn a lot from sentences like that, and it helps visitors too.

Step 5: reinforce with structured data

Mark up the same relationships with schema.org. Use the founder property on Organization, brand and category on Product, and author and publisher on Article. Give entities stable identifiers with @id so they can be referenced from several pages. Make sure the markup matches the visible text.

Link people to their bio pages, products to their category pages and articles to their authors. Use descriptive anchor text, as the guide on internal anchor text explains.

Common mistakes

  • Marking up relationships in schema that aren't stated anywhere on the page.
  • Using inconsistent names, which breaks the connection between pages.
  • Describing categories in marketing language no one searches for.
  • Treating the graph as a one-off project rather than updating it when staff or products change.

Once the map exists on paper, the holes in it are usually obvious.

Frequently Asked Questions

Is an entity relationship graph the same as a knowledge graph?

A knowledge graph is the large database that search engines maintain. Your own graph is a smaller map of the entities on your site and how they connect, and it helps you describe yourself in a way that can feed theirs.

Do I need special software?

No. A spreadsheet with three columns works for a small site: subject, relationship, object. Tools can help extract candidates, but a person should review them.

Which schema types express relationships?

Common ones include Organization with founder, Product with brand and category, Person with worksFor, and Article with author and publisher.