Guides AI search and AI visibility
Why do entities matter for search and AI?
What an entity is, how Google's Knowledge Graph and AI assistants decide who your brand is, and why being clearly understood now matters as much as ranking.
Most SEO still starts with keywords: what people type, and which page should rank for it. But search engines and AI assistants have a question to answer first. Before they decide what to say about you, they need to work out who you are.
That’s what “entities” are about. A brand that Google and AI assistants clearly understand is easier to recommend, cite and describe correctly. A brand they’re unsure about can be left out, confused with someone else or described from out-of-date sources.
What is an entity?
An entity is a distinct thing that can be identified: a person, a company, a place, a product, a publication. “Apple” the company and “apple” the fruit are different entities, even though the word is the same.
Google made the shift from keywords to entities public in 2012. Introducing its Knowledge Graph, Google said it was building a model that “understands real-world entities and their relationships to one another: things, not strings.” At launch it held more than 500 million objects. By 2020, Google said it had “over 500 billion facts about five billion entities”. That’s the most recent figure I’ve found from Google.
Where does Google get its view of you?
From many places, not just your own website.
Google says its Knowledge Graph draws on “materials shared across the web, as well as from open source and licensed databases.” It names Wikipedia as “a commonly-cited source, but it’s not the only one”, and says it draws on “hundreds of sources from across the web.”
The visible result is the knowledge panel, the box that can appear beside search results for a person, company or publication. Google says knowledge panels “are automatically generated”, and it doesn’t offer a way to apply for one. If one exists, its subject can often claim it and suggest changes, but Google says “not all knowledge panels are claimable.”
Your own site still plays a part. Google says structured data about your organisation can help it “disambiguate your organization in search results”, in other words tell you apart from others with a similar name. But it’s one source among many.
Why entities matter for AI search
AI assistants face the same problem as Google. To recommend or describe a brand, they have to be confident which brand they’re talking about.
The evidence here is still thin. A controlled study published in June 2026 on friction AI’s newsletter tested 12 sportswear brands across five AI models and 14,140 responses. Within the brands’ home category, a stronger Google Knowledge Graph presence went with being recommended more often. But the link was weak, rested on nine brands, and disappeared entirely once questions crossed into other categories. What mattered more, the authors found, was whether a brand was clearly associated with that category in the first place. They call this “category coding”.
In my experience, AI assistants repeat what the web consistently says about a brand: what it is, what it does and what it’s known for. I cover the wider research on this in Why does ChatGPT recommend your competitors and not you?
What goes wrong
In my experience, entity problems rarely look like SEO problems. They look like this:
- an AI assistant describes your company using details that are years out of date
- your brand is confused with another business, place or person with a similar name
- your knowledge panel is missing, shows the wrong logo or links to the wrong site
- a site with strong content gets few or no AI citations, while a closely related site with similar content is cited regularly
I’ve seen that last one with a client. The content was fine. What was missing was a clear, consistent picture across the web of who the site was.
For publishers, the same applies to journalists. Authors are entities too, which is one reason author information matters. I cover that in Do author pages help with Google’s E-E-A-T?
Why it matters
Ranking still matters. But in my view, more and more discovery happens in places where there’s no list of ten links, only an answer. In that answer, an assistant either names you or it doesn’t, and either describes you correctly or it doesn’t.
I wrote in Publishing Strategies that your AI reputation is being built off-site. Entities are a large part of why. You can’t fully control what the web says about you, but you can understand it, and that’s where the work starts.
What the right answer depends on
What will make a difference for you depends on how distinctive your name is, how consistently you’re described across the web, whether you have sister brands, products or people that also need to be recognised, and which search engines and AI assistants your audience uses.
How I help
I check how Google and the main AI assistants understand your brand today: what they think you are, where they get it from, and where they get it wrong. You come away with a clear picture of the gaps and a plan in order of impact, covering your site and the sources that shape what others say about you.
That’s part of my AI visibility consulting.
Not sure whether Google and AI assistants know who you are? I'll look at it with you on a free 30-minute call.
Got a question first? See the questions clients ask me.
Explore services