Entity SEO: how a person becomes a verifiable brand.

Entity SEO is the discipline of referencing a person, brand or organization across authoritative sources in such a way that search systems and large language models represent it as a unique, verifiable entity — with a stable ID, consistent facts and observable co-occurrence with topics — inside their own knowledge graph. For people, this is not a marketing exercise but a technical process: those who do not exist as an entity are not cited, do not appear in Knowledge Panels and are not named in AI answers. In 2026, a ranking without entity status is a mere address, not authority.

This piece shows why classical personal branding and entity SEO sit at two different levels, which five building blocks turn a person into a verifiable brand, how to build the Wikidata and sameAs graphs operationally — and what a 90-day protocol looks like that converts a profile person into a referenced subject-matter entity.

What is an entity, and why does a person have to become one?

In the semantic web, an entity is a uniquely identifiable thing, such as a person, a brand, a place or a work, that search systems recognize through a stable ID and verifiable facts. A person has to become one because Google and language models can only reliably tell entities apart, cite them and show them in Knowledge Panels. What matters is not the name but the stable ID and the set of verifiable facts that cluster around that ID. According to Google, the Knowledge Graph held more than 500 billion facts about five billion entities as of May 2020 (Danny Sullivan, Google blog). Google says this information comes from hundreds of sources across the web, including licensed data, and that knowledge panels are generated automatically (Google Knowledge Panel Help). For a person, that means: without consistent evidence on several independent sites, they remain one namesake string among many for these systems.

As long as a person does not have a Machine Identifier Number (MID) in the Knowledge Graph, they are just one of many namesake strings to Google. An LLM trained on Common Crawl, Wikipedia and licensed trade media cannot tell "Jane Doe, CMO of XY" from "Jane Doe, teacher" if there are no disambiguating fact points. The consequence: no citation, no authority, no Knowledge Panel.

The operational job of entity SEO is therefore threefold: identify a person uniquely (disambiguation), evidence them with consistent facts (verification) and semantically couple them to topic entities (association). Only those three layers translate a biographical existence into a machine-readable brand.

The difference between personal branding and entity SEO

Personal branding, as codified in the 2010s, addresses humans: how do I come across on LinkedIn, what is the tone of my bio, how consistent is my visual system? Entity SEO addresses machines: can a crawler disambiguate my identity, can a knowledge graph validate my facts, can an LLM reproduce me as a source?

The two layers do not contradict one another, but they do not replace one another either. A charismatic LinkedIn presence with 80,000 followers — without a Wikidata entry, without a consolidated sameAs graph, without Schema.org implementation — produces human perception, but not machine referenceability. Conversely, a technically perfect entity representation without substantive content produces no authority because the co-occurrence with topic entities is missing.

The strategic break: personal branding optimizes perception per contact. Entity SEO optimizes referenceability per query. In a market where AI answers become the first exposure to subject-matter experts, the second layer is the scarcer — and more expensive — resource.

A note on evidence: every cohort and correlation figure in this article (the operator cohort of 41 executives, the Entity Density Score, LLM recall) comes from our own advisory data and has not been replicated externally (evidence level C). How we collect and aggregate audit data is described on our benchmarks page.

The five building blocks of a verifiable person entity

From advisory practice with executives at Turkish Airlines, Volkswagen, Johnson & Johnson and ThyssenKrupp, a stable five-block model has emerged. Each block is necessary; none is sufficient on its own.

1. Stable ID

A machine-referenceable identifier: Google KG MID, Wikidata Q-ID, ORCID (for academic authors), LinkedIn Public ID. Without a stable ID, there is no disambiguation — and therefore no citation.

2. Canonical facts

A consistent canon: exact name, year of birth (or career start year), current position, affiliation, education, previous roles, publications, awards. These facts must be formulated identically across every profile — down to punctuation.

3. sameAs graph

7-10 curated, genuinely authoritative profiles that link bidirectionally to the entity home and carry identical NAP data. Candidates: Wikipedia, Wikidata, LinkedIn, Crunchbase, GitHub, Xing, YouTube, speaker bios, publisher author pages, industry associations. Quality over quantity: one strong, consistent source strengthens the trust graph more than a dozen weak directory listings.

4. Topic co-occurrence

Recurring mention of the person together with clearly defined topic entities (e.g., "Jana Richter" + "LLM SEO" + "Generative Engine Optimization"). LLMs do not learn expertise from titles; they learn it from statistical coupling. Anyone who never appears next to the topic is not an authority within it.

5. Verification chain

Every claimed fact must be evidenced independently at least once: a press release, conference agenda, interview, or a trade article with author schema. Google's Knowledge Vault research project (Dong et al., KDD 2014) combines extractions from many web sources with prior knowledge from existing knowledge bases and assigns each fact a calibrated probability. That does not document an inclusion threshold for the production Knowledge Graph, but the logic suggests that facts backed by several independent sources are adopted more reliably than facts claimed only once.

The entity home — your central node

The entity home is the one page in your own domain that every other source points to and that defines every fact authoritatively. Technically, it is the root element of the sameAs graph; in content terms, it is the single source of truth. Without that page, every distributed profile entry is a satellite without an anchor planet.

Required components of an entity home: a URL placed canonically at /about/ or under the person's name; a complete Person schema in JSON-LD; a biographical prose version with topic co-occurrence; reliable cross-links to every external profile; visible credentials and publications. The pattern: a homepage plus a dedicated About area as a layered entity home.

The minimal but complete Person schema looks like this:

{
  "@context": "https://schema.org",
  "@type": "Person",
  "@id": "https://www.jana-richter.com/#jana-richter",
  "name": "Jana Richter",
  "givenName": "Jana",
  "familyName": "Richter",
  "jobTitle": "Head of Digital Strategy",
  "worksFor": {
    "@type": "Organization",
    "@id": "https://www.example-agency.com/#org",
    "name": "Example Agency"
  },
  "knowsAbout": [
    "Entity SEO",
    "Generative Engine Optimization",
    "Content Strategy",
    "LLM SEO"
  ],
  "alumniOf": "Freie Universität Berlin",
  "url": "https://www.jana-richter.com/",
  "sameAs": [
    "https://www.linkedin.com/in/jana-richter-seo/",
    "https://www.wikidata.org/wiki/Q...",
    "https://www.crunchbase.com/person/jana-richter",
    "https://www.youtube.com/@janarichter"
  ]
}

Two details are non-negotiable. First: @id must be a stable, domain-owned identifier — it becomes the primary reference across the entire website graph. Second: knowsAbout must explicitly name the topic entities the person claims expertise in. This property is the machine-readable expertise claim.

sameAs graph: the trust multiplier

The sameAs graph is the totality of external profiles that reference the same person. It works like a citation network: the more authoritative sources state the same facts in identical form, the higher the probability that a knowledge graph will absorb the person as a consolidated entity. The rule of thumb: 7-10 curated, genuinely strong identity profiles, maintained consistently and linked bidirectionally. Quality over quantity.

NAP consistency (name, affiliation, position) is the most common break. A person appears on LinkedIn as "Dr. Maria Schmidt", on Crunchbase as "Maria Schmidt, PhD", in a podcast interview as "Maria Schmidt-Berger". For humans, the same person — for an entity resolver, three candidates with unclear identity. The disambiguation loss is measurable: in our operator cohort of 41 executives (own data, evidence level C), NAP cleanup alone produced a 28% increase in cross-profile clustering rate in a sample of entity queries, including queries via the Bing Entity Search API. Microsoft retired that API on August 11, 2025; new measurements run on other entity sources.

The profiles that deliver the highest marginal utility in practice are less social-media-driven than expected. In descending order: Wikidata, Wikipedia (where the notability threshold is reached), Crunchbase, ORCID, LinkedIn, GitHub, verified publisher author pages, industry-association member directories, conference speaker bios, YouTube channel About pages. Instagram and TikTok contribute late from an entity perspective — they address personal branding, not entity verification.

Wikidata & Wikipedia: the structural backbones

Wikidata is the structural backbone of the open semantic web. Its structured data is released under CC0 (Wikidata: Licensing), so search engines, knowledge graphs and AI applications can reuse it freely; how much of it flows into any given LLM's training data is not disclosed by the model providers. In our experience, a Wikidata item with a stable Q-ID is the largest single lever for a person without a Wikipedia entry. The bar is lower than Wikipedia's, but it exists: under the Wikidata notability policy, an item must describe a clearly identifiable entity that can be described using serious, publicly available references. Items without such references can be removed through a deletion request.

The critical Wikidata properties for a person are: P31 (instance of: Q5 human), P106 (occupation), P108 (employer), P69 (educated at), P856 (official website), P2002 (X username), P6634 (LinkedIn personal profile ID), P2013 (Facebook username), P2397 (YouTube channel ID), P1416 (affiliation), P800 (notable work). Every one of these properties should carry a reference, such as P854 (reference URL), to a primary source; unreferenced statements are easy to challenge and more likely to be removed by other editors.

Wikipedia is the next stage, but underlies strict relevance and neutrality criteria. German Wikipedia checks biographies against its notability criteria (Relevanzkriterien); the article itself must show, with sources, that they are met. In deletion discussions, biographies often fail on missing encyclopedic relevance, meaning too little received expert publication, media resonance or institutional visibility. Those who do not reach the threshold should build a stronger Wikidata base and focus on building references through trade media before even attempting a Wikipedia entry.

"A Wikidata entry without a consolidated sameAs graph is a rumor. A sameAs graph without Wikidata anchoring is a silo. Only the combination of both produces what Google and LLMs accept as an entity."

Co-occurrence with topic entities: how LLMs recognize "expertise"

Large language models do not learn titles, they learn distributions. When across billions of tokens of training material the name "Jana Richter" appears disproportionately often in the same context as "Entity SEO", "Generative Engine Optimization" and "Content Strategy", the model learns that association statistically. This coupling — known in the information-retrieval literature as co-occurrence — is the actual basis for LLM citation in expert answers.

Practically that means: a CMO who writes 40 LinkedIn posts on generic management over 24 months couples themselves to no subject-matter entity. A CMO who, in the same period, publishes six substantive expert contributions within a tightly defined topic corridor — in trade media with clean author schema — produces far stronger co-occurrence. Quantity is secondary; what matters is the semantic density of joint appearances.

The operational discipline: name topic entities explicitly. In knowsAbout, in the About-page intro, in every author box, in the Wikidata P106 additions. The shift from "I work on digital marketing" to "I work on entity SEO and generative engine optimization" is not a rhetorical style — it is a machine-readable expertise claim.

The entity-density formula: when a person counts as authority

Anyone who wants to steer the maturity of a person entity operationally needs a metric. In advisory practice, we use the Entity Density Score (EDS) — a composite metric built from four dimensions, mapping structural entity status onto a 0–100 scale.

EDS = (0.30 × I) + (0.25 × S) + (0.25 × C) + (0.20 × V)

where:
I = Identifier coverage   (Knowledge Graph MID, Wikidata Q-ID, ORCID, LinkedIn ID → 0-1)
S = sameAs consistency    (number of NAP-consistent profiles / 25, capped at 1.0)
C = Co-occurrence score   (LLM cross-model recall on 10 topic queries / 10)
V = Verification depth    (avg. independent primary sources per core fact / 3, capped at 1.0)

Thresholds:
EDS < 25   → namesake string (no entity)
EDS 25-49  → fragmented entity (partial recognition)
EDS 50-74  → stable entity (Knowledge Panel candidate)
EDS ≥ 75   → authoritative entity (LLM citation tier)

The thresholds are calibrated on our own portfolio data (evidence level C), not externally validated. Across our portfolio, EDS correlates with citation rate in AI Overview and ChatGPT Search answers at r = 0.81 — far stronger than classical authority scores like Moz DA (r = 0.22 for people) or Ahrefs Domain Rating, which are barely meaningful for individuals. An EDS of 75+ is, in practice, the entry ticket into the group of people consistently cited in AI answers.

The 90-day protocol: from profile to entity

Entity SEO for people is not a campaign topic — it is a structured project. The following 90-day protocol takes an executive with an existing but unconsolidated online presence from a typical EDS of 20–30 to an EDS of 60–75. It assumes that an entity home (own domain with author infrastructure) exists or is built within the first two weeks.

Days 1-7 — entity audit & NAP consolidation

Inventory of every existing profile (LinkedIn, X, Crunchbase, Xing, GitHub, YouTube, speaker bios, publisher author pages, podcast guest appearances). Consolidate NAP data into a master table, mark inconsistencies, define the canonical spelling.

Days 8-21 — build the entity home

Own domain with an /about/ page as the central node. Complete Person schema with sameAs, knowsAbout, alumniOf, worksFor, award, memberOf. Canonical biography as single source of truth. Publications list with author schema per article.

Days 22-35 — close the sameAs graph

Update the 7-10 curated core profiles to identical NAP data, identical photo and cross-links to the entity home. Ensure bidirectional linking: every external profile links the entity home, the entity home links every external profile.

Days 36-55 — Wikidata anchoring

Create or complete the Wikidata item. Evidence every relevant property (P31, P106, P108, P69, P856, P2002, P6634) with references to primary sources. In parallel: where notability allows, prepare a Wikipedia draft in the user namespace.

Days 56-75 — co-occurrence publishing

Five to eight substantive publications in authoritative trade media where the person entity co-occurs with the three core topic entities. Goal: every piece references the entity home, carries clean author schema and explicitly couples the name to the defined topics.

Days 76-85 — citation and mention monitoring

Tracking setup: Google Knowledge Graph Search API, Kalicube Pro, Ahrefs Brand Radar, cross-model LLM prompts against 10 persona queries ("Who leads the field on topic X?"). Document the baseline.

Days 86-90 — entity review & delta report

Pre/post comparison. Document EDS calculation, citation rate, Knowledge Panel status, LLM recall rate. Prioritize the backlog for the next quarter: deepen co-occurrence, densify verification sources, add reputation-engineering measures.

In our own data (evidence level C), in 73% of the 41 cases we have supported, the EDS uplift after 90 days landed between 35 and 55 points — with a measurable effect on citations in AI answers and, in 11 of 41 cases, a newly created or visibly stabilized Knowledge Panel. The Knowledge Panel is never the goal, only a by-product of clean entity work.

Typical mistakes on the way to a person entity

• Mistake 1: multiple entity homes. Two competing root URLs (e.g., personal domain plus corporate About page) destroy the graph anchor. One person, one node, one @id.
• Mistake 2: confusing social-media reach with entity authority. 100,000 LinkedIn followers do not replace a Wikidata Q-ID. Reach is a channel, not an identifier.
• Mistake 3: topic drift. Monthly shifting focus topics prevent the statistical co-occurrence from which LLMs derive expertise.
• Mistake 4: unsupported facts in Schema.org. Schema statements that cannot be replicated by an external source lose weight. Knowledge-Graph confidence is a multi-source signal.
• Mistake 5: ignoring Wikidata. No other single lever delivers a comparable marginal utility for non-celebrity subject-matter experts.

On top of that comes the discipline of treating entity work not as a one-off setup. Entity drift — a changed E-E-A-T profile after a job change, new book release, new affiliation — must be reflected in every relevant source within 30 days. Our own cohort data (evidence level C) show: people who leave their sameAs and Wikidata signals unconsolidated for more than 60 days after a career change lose a median 18% of LLM recall within six months.

Why entity SEO belongs to online reputation in 2026

The boundaries between entity SEO, online reputation and brand strategy are blurring. A person cited in AI answers exerts steerable influence; a person misquoted lives with a reputation claim that is hard to correct. As LLMs become the first exposure, the question is no longer whether someone is "findable on Google" — it is how they are framed in generative answers.

The transfer from the brand world is direct: a corporate brand without a consistent knowledge graph is perceived as fragmented; a person without a consolidated entity graph is either omitted from the AI answer or placed in the wrong context. Both cost — revenue, authority, pricing power in the own consulting field, in the speakers' circuit, in the hiring process.

Brands that take this seriously separate entity SEO cleanly from classical personal branding and from the prompt-level-SEO discipline: personal branding produces tone, entity SEO produces referenceability, prompt-level SEO produces answer presence. The three layers are sequential — personal branding without an entity base is aesthetics alone, prompt-level SEO without an entity base is play.

Conclusion: the strategic imperative for executives in 2026

For every executive with a public profile, the question in 2026 is no longer "how do I become more visible?" It is: "Am I an entity to machines — or am I a namesake string?" That distinction decides whether your expertise is cited in AI answers, presented in Knowledge Panels and recognized in recruiting, pricing and authority contexts — or whether it simply does not show up in the increasingly AI-mediated first exposure.

Entity SEO is the technical, structured process of brand-building for people. It is less visible than a campaign, less charming than a keynote video — and at the same time the structurally most durable investment in personal subject-matter authority an executive can make in the next three years. Everything else is decoration on a foundation that machines cannot read.

Sources

• Google (Danny Sullivan): Google’s Knowledge Graph and Knowledge Panels (May 20, 2020)
• Google Knowledge Panel Help: About knowledge panels (accessed September 23, 2026)
• Dong, Gabrilovich, Heitz et al.: Knowledge Vault: A Web-Scale Approach to Probabilistic Knowledge Fusion, KDD 2014 (August 2014)
• Wikidata: Wikidata:Notability (accessed September 23, 2026)
• Wikidata: Wikidata:Licensing (accessed September 23, 2026)
• German Wikipedia: Wikipedia:Relevanzkriterien (accessed September 23, 2026)
• Microsoft Learn: Bing Search APIs Retiring on August 11, 2025 (May 15, 2025)