Entity SEO: sameAs, about, mentions
The technical foundation that makes AI engines verify you exist. Schema gave you the structure; entity SEO tells AI engines who your brand is across the web.
Entity SEO vs page SEO
Traditional SEO optimizes pages. A URL ranks for a keyword. Page SEO asks: does Google think this URL is the best answer to that query?
Entity SEO optimizes entities. A brand, a person, a product, a place — verified as a real thing, connected to other real things across the web. Entity SEO asks: does the AI engine believe this entity exists, and does it know what it does?
The shift matters because AI engines synthesize answers from facts about entities, not pages. When ChatGPT says "the top GEO tools are X, Y, and Z," it's pulling from its entity model — its internal representation of which brands exist in the GEO category and what their attributes are. That model gets built from cross-source verification: same name, same description, same founders, same key facts across many independent sources.
If your brand is a coherent entity, AI engines cite you. If your brand is a fuzzy collection of inconsistent mentions, AI engines skip you in favor of clearer signals from competitors.
Three schema properties that define your entity
Schema.org gives you direct properties for entity definition. Three matter most for GEO:
sameAs — the entity verification array
The single highest-value entity property. sameAs is an array of URLs pointing to other places this entity exists on the web. When you ship sameAs: ["https://en.wikipedia.org/wiki/YourBrand", "https://www.linkedin.com/company/yourbrand", "https://www.crunchbase.com/organization/yourbrand"], you're telling every AI engine that the YourBrand on your site is the same YourBrand on Wikipedia, LinkedIn, and Crunchbase.
Why this works: AI engines cross-reference entities to verify they're real. If Wikipedia says YourBrand was founded in 2024 in Vancouver, and your homepage says the same, and your LinkedIn says the same, the entity is verified. The AI now trusts the entity enough to cite it. If those three sources disagree, the AI treats the entity as ambiguous and chooses a clearer competitor.
What to include in your sameAs array, in priority order:
- Wikipedia — the canonical entity backbone for AI engines, especially Claude and ChatGPT
- Wikidata — the structured-data counterpart to Wikipedia, directly machine-readable
- LinkedIn company page — strongest B2B verification signal
- Crunchbase — strongest startup/funding verification signal
- X/Twitter — strongest founder presence signal
- GitHub — relevant if you're a developer tool
- YouTube channel — strongest creator/educator signal
- Industry-specific authoritative directories — G2 for B2B SaaS, Capterra for software, Yelp for local, etc.
Three is the minimum. Five is good. Eight is excellent. More than ten starts hitting diminishing returns and risks linking to weak sources that dilute your entity signal.
about — what the page is fundamentally concerned with
The about property tells AI engines the primary topic of a page. It's distinct from keywords because it links to entity URIs, not text strings.
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "How GEO works for ecommerce brands",
"about": [
{
"@type": "Thing",
"name": "Generative Engine Optimization",
"sameAs": "https://en.wikipedia.org/wiki/Generative_engine_optimization"
},
{
"@type": "Thing",
"name": "E-commerce",
"sameAs": "https://en.wikipedia.org/wiki/E-commerce"
}
]
}The sameAs inside about is what makes this powerful. You're telling the AI engine "this page is about the specific concept of GEO as defined on Wikipedia," not "this page contains the string 'GEO' multiple times." The AI engine can then confidently include this page in its entity model for the GEO topic.
mentions — entities referenced but not the primary topic
The mentions property is the lighter-weight counterpart to about. Use it for entities the page references in passing — competitors named, products discussed, people quoted, places referenced.
Why bother: AI engines build co-occurrence graphs. If your "best GEO tools" article mentions Tool A, Tool B, and Tool C via mentions, the AI engine starts treating your domain as authoritative for that category. When users later ask "what tools were mentioned in the best-GEO-tools list," your entity gets cited because the structured mention is verifiable.
knowsAbout: the topic authority property
On your Organization schema (homepage, footer), add knowsAbout with the topics your brand has authority on:
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Your Brand",
"knowsAbout": [
"Generative Engine Optimization",
"Search Engine Optimization",
"AI search visibility",
"Schema markup",
"Entity SEO"
]
}This is the schema-level equivalent of E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness). AI engines reading knowsAbout get a structured declaration of your topic authority. Combined with content that actually demonstrates that authority, this signal compounds.
Keep knowsAbout tight. 3-7 topics. If you list 30, the AI ignores the list. Topic authority means depth in a few areas, not surface presence everywhere.
The consistency rule
Every entity-defining fact must match across every source. AI engines cross-reference relentlessly. The five facts that matter most:
- Legal name + brand name — both must be consistent across Wikipedia, LinkedIn, Crunchbase, Google Business Profile, your homepage, and your invoices
- Founding year — pick a canonical year and stick to it across every source
- Headquarters location — city + country at minimum; full address if you have one
- Founder names — exact spelling, with each founder linked via their own sameAs to their LinkedIn/Wikipedia
- One-line description — the elevator pitch, identical or near-identical across all entity sources
If those five facts disagree across two sources, you have entity drift. AI engines either pick the more-cited version (usually Wikipedia or LinkedIn, not your homepage) or treat the entity as ambiguous. Both outcomes hurt citation.
When and how to build a Wikipedia entry
Wikipedia is the highest-value entity source for AI citation. Wikipedia and Reddit together command the largest share of LLM citations across major AI engines. But Wikipedia has strict notability requirements, and a rejected or deleted entry hurts more than no entry at all.
You qualify for a Wikipedia entry if you can demonstrate significant coverage in independent, reliable sources — meaning multiple substantive articles in named publications, not press releases or sponsored content. Concretely: two or more 800+ word articles in named tech press (TechCrunch, The Verge, Bloomberg, etc.) or peer-reviewed academic citations.
If you qualify, the right approach is to find an experienced Wikipedia editor (paid or volunteer) to draft and submit the entry. DIY rarely works — Wikipedia editors aggressively delete entries that read as promotional or fail to cite independent sources properly.
If you don't qualify yet, focus on building the citation surface first. One year of consistent third-party press coverage, then revisit.
Implementation: shipping entity foundations this week
- Day 1. Audit your five entity-defining facts (name, founding year, HQ, founders, description) across Wikipedia (if any), LinkedIn, Crunchbase, Google Business Profile, and homepage. Pick canonical versions where they disagree.
- Day 2-3. Update all sources to match canonical versions. This often means editing LinkedIn company page settings, requesting Crunchbase profile updates, and fixing your homepage copy.
- Day 4. Add
sameAsarray to your Organization schema with at least 5 entity URLs (Wikipedia if you have one, LinkedIn, Crunchbase, X/Twitter, plus one industry-specific). - Day 5. Add
knowsAboutwith 3-7 topics to your Organization schema. - Day 6. Add
aboutwith WikipediasameAsURIs to your top 5 pillar content pages. - Day 7. Run a Reffed audit. Compare your entity subscore to baseline.
What comes next
Lesson 1.3 covers llms.txt and crawler access — the technical layer that determines whether AI engines can actually read the pages you're optimizing. Even perfect schema doesn't help if your robots.txt is blocking AI crawlers.