I reference the Knowledge Graph constantly when I talk about entity SEO, and I realized I'd never actually written the plain explanation of what it is on its own. So here it is, properly.
"Things, Not Strings," the Phrase That Still Explains It Best
When Google launched the Knowledge Graph on May 16, 2012, engineer Amit Singhal described the shift with a phrase that's still the cleanest summary I know: search moving from "strings" to "things." Before this, a search engine mostly matched the text you typed against text on pages. After this, Google started building an actual model of the real-world objects and concepts those words referred to, and the relationships between them.
The scale of this database has grown enormously since launch. At launch, it held roughly 500 million entities and 3.5 billion facts, built initially from sources like Freebase, Wikipedia, and the CIA World Factbook. Within seven months it had tripled. Different sources report meaningfully different current figures, some citing around 5 billion entities and 500 billion facts, others citing figures as high as 54 billion entities and 1.6 trillion facts, and Google hasn't published an official, current, authoritative count. I'd treat the exact number as directionally "enormous and still growing" rather than pin to one precise figure nobody can fully verify, the honest answer is that it's large enough that the exact count matters less than whether your entity is clearly represented inside it at all.
What Actually Makes Up an "Entity" in This System
Every entity in the graph carries an internal machine identifier, often called a kgmid, along with typed properties (facts specific to that entity) and relationship edges connecting it to other entities. A person entity connects to their employer, their body of work, their professional credentials. An organization entity connects to its people, its services, its location. These connections are what let Google answer a question about one entity by pulling in verified, related facts about connected entities, the mechanism behind a knowledge panel showing a company's founder, headquarters, and related services all in one place.
Where the Data Actually Comes From
The graph draws from multiple input streams, not one single source. Public open-web data, Wikipedia and Wikidata specifically (Wikidata matters more than people realize, since it's structured and machine-readable in a way raw web pages aren't), licensed data feeds for things like sports scores and stock prices, and direct input from entities that have claimed and verified their own knowledge panel. This last point matters practically: if you've never claimed your own entity's panel or actively fed it consistent information, you're leaving a meaningful input source unmanaged.
What Is a Knowledge Graph? Google's Entity Database, Explained Properly — self-check
Reviewed: 0/5 (0%)
Why This Matters More Now Than It Did Five Years Ago
For most of the Knowledge Graph's history, its visible impact was mostly cosmetic from a marketer's perspective, a nice knowledge panel, a richer-looking search result. That's changed. Google's AI systems, including AI Overviews and Gemini-powered answers, draw on the same underlying entity infrastructure when deciding what to trust and cite. A brand with a clear, well-connected entity presence has a real structural advantage in AI-era search that a brand relying purely on keyword-optimized pages doesn't have access to, no matter how well-written those pages are.
This is the exact infrastructure layer I'm working with when I talk about entity SEO, and it's why I framed the difference between keyword-first and entity-first strategy the way I did in entity SEO vs. keyword SEO. The Knowledge Graph isn't a side feature to optimize for occasionally. It's increasingly the substrate everything else sits on.
The Practical Starting Point If You're Building Entity Presence From Zero
If none of this currently applies to your brand, here's where I'd actually start, in order. Claim and verify your Knowledge Panel if one exists or is eligible. Make sure your Organization and Person schema (for named individuals, founders, key experts) is complete and specifically includes a sameAs array linking to your verified profiles elsewhere, LinkedIn, Wikidata if eligible, industry directories. And make sure those external profiles say the same thing, consistently, since inconsistency across sources is exactly what erodes the confidence Google's systems place in an entity, the same principle I covered in the NAP-consistency comparison in the entity SEO piece linked above.
Frequently Asked Questions
Is having a Knowledge Panel the same as having strong entity SEO? Direct answer: A Knowledge Panel is one visible outcome of strong entity signals, not the whole picture. Plenty of entity-optimization work, schema depth, sameAs consistency, topical authority, happens without ever producing a public-facing panel, but still meaningfully affects how confidently AI systems cite the entity. How is the Knowledge Graph different from a website's own schema markup? Direct answer: Schema markup is what a site provides to help Google understand it. The Knowledge Graph is Google's own internal database that ingests that schema, along with many other sources, into a unified model. Good schema is an input to the graph, not the graph itself. Can a small business or individual realistically appear in the Knowledge Graph? Direct answer: Yes, entity recognition isn't reserved for large, famous brands. Clear, consistent, well-structured entity signals can establish a smaller entity's presence, though it typically takes longer to accumulate the same depth of connected facts that a larger, longer-established entity already has. Does the exact size of the Knowledge Graph matter for SEO strategy? Direct answer: Not really, in practical terms. What matters is whether your specific entity is clearly and accurately represented within it, the graph's total size is interesting context, not something you can act on directly. How often does the Knowledge Graph update? Direct answer: Continuously and dynamically rather than on a fixed schedule, entities and facts get added, corrected, and reconnected on an ongoing basis as new source data becomes available and verified.Understanding this structure properly is the foundation of every entity SEO engagement I run, it's not an abstract concept, it's the actual system I'm building signals for. See how your brand's current entity presence looks.