Entity SEO: Getting Into Google's Knowledge Graph

Google stopped being a keyword-matching engine around 2012. The Hummingbird update—and every algorithm iteration since—moved the system toward understanding entities: real-world objects, organizations, people, and places that exist independently of any particular document. If Google doesn't have a confident model of your client's organization as an entity, no amount of keyword optimization fully compensates.

This matters more now than it did two years ago. AI Overviews, ChatGPT Search, Perplexity, and every other answer engine pull from entity understanding, not keyword frequency. If you're not in the Knowledge Graph, you're invisible to the AI layer of search—and that layer is growing fast.

Entities vs. Keywords

A keyword is a string: "best accounting software." An entity is a thing: QuickBooks, the product made by Intuit, headquartered in Mountain View, used by 7 million small businesses. Google knows these are the same thing from different angles because it has a Knowledge Graph node for QuickBooks with hundreds of attributes and relationships.

When someone searches "accounting software alternatives," Google doesn't scan pages for those words. It identifies the entities in the query—accounting software as a category, alternatives as a relationship type—and surfaces entities it understands as fitting those relationships.

For your clients: if Google has no Knowledge Graph node for their organization, it treats every query about them as keyword matching, which is slower, less confident, and less likely to generate AI Overview citations. Building entity presence is the prerequisite for GEO (generative engine optimization).

How the Knowledge Graph Works

The Knowledge Graph is a massive structured database of entities and their relationships. Nodes are entities (organizations, people, products, locations). Edges are relationships (works for, headquartered in, produces, founded by).

Google populates Knowledge Graph nodes from:

1. Wikipedia / Wikidata — highest-trust source. If your client has a Wikipedia article, they almost certainly have a Knowledge Graph node.
2. Google Business Profile — the primary source for local businesses. Verified GBP = Knowledge Graph presence for that entity.
3. Structured data on the site — Organization, Person, LocalBusiness schema with sameAs links pointing to authoritative external sources.
4. Mentions across the web — consistent name, address, phone across directories, press mentions, industry databases.
5. Google's own crawling and extraction — Google infers entity attributes from crawled pages even without explicit schema.

The practical implication: structured data alone doesn't create a Knowledge Graph node. It confirms one that other signals already suggest exists. If there are no external signals, schema just annotates an unknown entity.

Organization Schema with sameAs

The sameAs property is the linchpin of entity SEO. It tells Google "this entity on our website is the same entity as this external node." When those external nodes are authoritative (Wikipedia, Wikidata, Crunchbase, LinkedIn, a government registry), Google can merge its knowledge of the external entity with your on-site assertions.

A complete Organization schema block:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Meridian Accounting Group",
  "url": "https://meridianaccounting.com",
  "logo": "https://meridianaccounting.com/images/logo.png",
  "description": "Full-service CPA firm serving small and mid-size businesses in the Pacific Northwest since 2008.",
  "foundingDate": "2008",
  "numberOfEmployees": {
    "@type": "QuantitativeValue",
    "value": 24
  },
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "850 SW Broadway, Suite 1200",
    "addressLocality": "Portland",
    "addressRegion": "OR",
    "postalCode": "97205",
    "addressCountry": "US"
  },
  "contactPoint": {
    "@type": "ContactPoint",
    "telephone": "+1-503-555-0199",
    "contactType": "customer service"
  },
  "sameAs": [
    "https://www.linkedin.com/company/meridian-accounting-group",
    "https://www.facebook.com/MeridianAccounting",
    "https://www.wikidata.org/wiki/Q12345678",
    "https://en.wikipedia.org/wiki/Meridian_Accounting_Group",
    "https://www.bbb.org/us/or/portland/profile/accounting/meridian-accounting-group-1296-XXXXXXX"
  ]
}

Use Recon's Schema Markup Validator to check that existing schema is valid and includes sameAs links. Most client sites have Organization schema with zero sameAs references, which dramatically reduces its value as an entity signal.

Wikidata: The Underused Entity Registry

Wikipedia gets all the attention, but Wikidata is the machine-readable knowledge base that feeds the Knowledge Graph more directly. You don't need a notable Wikipedia article to get a Wikidata entry—you just need to be a real, verifiable entity.

For established businesses (5+ years, multiple web mentions, GBP verified), creating a Wikidata item is often feasible:

1. Create a Wikidata account
2. Create a new item for the organization
3. Add properties: instance of (Q4830453 = business), name (P1448), official website (P856), founding date (P571), country (P17), located in (P131), LinkedIn ID (P4264)
4. Once the Wikidata Q-number exists, add it to sameAs in the site's Organization schema

This isn't appropriate for every client—Wikidata moderators remove entries for entities with insufficient verifiability. But for clients who have been covered in local press, have a GBP with 50+ reviews, and have consistent directory listings, a Wikidata item is achievable and valuable.

Consistent NAP: The Entity Fingerprint

Name, Address, Phone (NAP) consistency across the web is the entity SEO equivalent of link building. Google uses NAP patterns to confirm that mentions of your client across different sites refer to the same real-world entity.

Inconsistencies that break entity matching:

• "Main Street" vs. "Main St." vs. "Main St" (different conventions across directories)
• Phone number formatted as (503) 555-0199 in one place, 503.555.0199 in another
• Business name as "Meridian Accounting Group LLC" in state records but "Meridian Accounting" on the website

These aren't just citation consistency issues. They're entity disambiguation failures. Google isn't sure if these are the same business or two different businesses. When it's unsure, it doesn't merge the signals—it holds them as candidates and discounts both.

Audit NAP across: Google Business Profile, Bing Places, Apple Maps, Yelp, Facebook, LinkedIn, industry directories, state business registry (Secretary of State lookup), BBB. The name and phone on the website should match all of these exactly.

Topical Authority as Entity Signal

Beyond organizational entity, topical authority is the signal that tells Google "this entity is an expert source on [topic]." It's built through:

• Depth of coverage: 15 thorough articles on a topic outperform 150 thin ones
• Internal linking structure: a hub-and-spoke model that connects topic pieces signals intentional expertise
• Expert attribution: author pages with credentials, LinkedIn profiles, professional association memberships
• External validation: other authoritative sites linking to your content on the topic

For agencies selling GEO services, topical authority is why some clients appear in AI Overviews and others don't. It's not about keyword density. It's about Google's confidence that this entity is authoritative on this subject. A dental practice with 10 detailed articles about pediatric dentistry is more likely to appear in AI Overviews for pediatric dental questions than one with a single "Services" page.

How Entity Understanding Feeds AI Overviews

Google's AI Overviews pull from a combination of the Knowledge Graph and document-level understanding. For informational queries, entities with Knowledge Graph presence and strong topical authority get cited. For local queries, verified GBP entities with positive signals get featured.

The mechanism: AI Overviews aren't doing raw retrieval at query time. They're drawing on pre-computed entity knowledge plus retrieval augmentation. Entities Google already understands get the benefit of the doubt. Unknown entities get excluded even when their content is relevant.

This is why entity SEO is now a prerequisite for GEO, not a separate discipline. If your client isn't a known entity to Google, they can't get cited in AI Overviews for their target topics regardless of how good their content is.

Clients optimized for AI visibility need: Knowledge Graph presence (via GBP, Wikidata, consistent NAP), valid Organization schema with sameAs, topical authority signals, and structured data on all key pages. That's the stack.

What to Skip

Skip obsessing over getting a Google Knowledge Panel. Knowledge Panels are a display feature, not a ranking mechanism. You can have strong entity presence in the Knowledge Graph without a visible Knowledge Panel. Chasing the Panel is a vanity metric—build the underlying entity signals instead.

Skip low-quality directory submissions at scale. 200 directory submissions from a citation building service often do more harm than good—many of those directories have NAP inconsistencies and some are spam targets. 20 high-quality, accurate citations in relevant directories beats 200 inconsistent ones.

Skip treating Wikipedia as an SEO tactic. Wikipedia editors will delete promotional articles. You need genuine notability (multiple independent reliable sources covering the entity). For most SMBs, Wikidata is the right target, not Wikipedia.

The Opinionated Take

Entity SEO is where agencies can build a durable moat. Keyword optimization is commoditized—any decent tool can run a gap analysis. But building a client's entity presence in the Knowledge Graph requires manual work: verifying directories, creating Wikidata items, building consistent citation profiles, structuring schema correctly with sameAs links. Most agencies don't do it because it's not as legible as "we added 10 keywords." That's the opportunity.

The audit one-liner for entity SEO: If your client's Organization schema has no sameAs links and their NAP is inconsistent across directories, they're an unknown entity to the Knowledge Graph—and unknown entities don't get cited in AI Overviews, regardless of content quality.

Related reading

• Structured Data AI Search Loves
• Building an AI-Visible Website
• How AI Search Engines Rank Websites