Author Schema and E-E-A-T: How to Build Person Authority for AI

Person schema is the JSON-LD block that tells AI answer engines exactly who wrote your content, what they know, where they work, and where else they exist online. Drop it on your about page, reference it from every blog byline, and tie it to your Organization schema. When ChatGPT, Perplexity, or Google AI Overviews try to decide whether the author of a piece is a credible source, Person schema gives them a structured answer instead of a guess.

E‑E‑A‑T was Google’s framework first. Experience, Expertise, Authoritativeness, and Trustworthiness are the four signals their search quality raters use to evaluate content. Most SEO people treated the framework as cosmetic for years. AI search changed that. When an AI engine has to pick one source to quote inside a generated answer, it cannot fall back on ten blue links. It picks a source, repeats it as fact, and that source either has visible expertise signals or it does not.

This guide walks through what Person schema actually contains, why E‑E‑A‑T signals matter more for AI than they ever did for traditional search, how to build a Person entity that AI models recognize, where to put your schema, what mistakes to avoid, and how to measure whether the work is paying off. If you publish content under your name and you want AI engines to treat you as an expert, the work starts here.

Why E‑E‑A‑T matters more for AI than for SEO

Traditional SEO had a buffer. A page could rank on backlinks, keyword targeting, and on-page signals even if the author was a question mark. The reader landed on the page, read the content, and decided for themselves whether the source was trustworthy. The human did the credibility check.

AI search removed that buffer. The AI engine reads the source on your behalf, synthesizes an answer, and presents that answer as if it were already verified. Most users never click through. Whatever the AI engine decided about the source’s credibility is the entire judgment.

That changes what AI engines have to do at retrieval time. They cannot pass the credibility question to the user. They have to make it themselves, and they have to make it fast, before generating the answer. The mechanism they use is E‑E‑A‑T signals, and the cleanest, most machine readable version of those signals is Person schema linked to a verifiable identity across the web.

Three concrete shifts come out of that:

• AI models will not consistently cite anonymous content for high stakes queries. Medical, legal, financial, and “best of” comparison queries all skew toward sources with declared authors and visible credentials.
• AI models reward documented expertise over generic authority. A site with one named expert who writes deeply about a niche outranks a generic content farm with a higher overall domain rating.
• AI models cross check author claims. When the byline says “by Jane Smith, CMO,” and the Person schema lists her as CMO at the same organization, and her LinkedIn confirms it, and Wikidata has the same job title, the signal compounds. Every consistent reference is a vote.

What Person schema actually contains

Person schema is a JSON-LD object that lives in a script tag in the head of your page. The core fields are short, but populating the right fields correctly is what moves AI models from “we have an author” to “we have a credible identifiable source.” Here is the minimum useful Person schema for a content author:

{
  "@context": "https://schema.org",
  "@type": "Person",
  "@id": "https://yoursite.com/about#person",
  "name": "Jane Smith",
  "jobTitle": "Founder & CMO",
  "worksFor": {
    "@type": "Organization",
    "@id": "https://yoursite.com/#organization"
  },
  "url": "https://yoursite.com/about",
  "image": "https://yoursite.com/images/jane-smith.jpg",
  "description": "Jane Smith is a digital marketing strategist with 12 years of experience in B2B SaaS. She writes about marketing operations and demand generation.",
  "knowsAbout": [
    "Marketing Operations",
    "Demand Generation",
    "B2B SaaS",
    "Salesforce Administration"
  ],
  "alumniOf": {
    "@type": "EducationalOrganization",
    "name": "University of Michigan"
  },
  "sameAs": [
    "https://www.linkedin.com/in/janesmith",
    "https://twitter.com/janesmith",
    "https://www.wikidata.org/wiki/Q123456789"
  ]
}

Every field on that object earns its place. The @id is a stable identifier that other schema blocks reference, so the Person entity stays unified instead of fragmenting across pages. The name and jobTitle answer “who is this and what do they do.” The worksFor link ties the person to a recognized Organization entity, which should have its own @id on your homepage or organization page. The knowsAbout array declares topical expertise in a format AI models can match against query topics. The alumniOf field gives credentialing context. The sameAs array is the bridge to the rest of the web. Each link is a vote that the person is who the schema says they are.

For a deeper walkthrough of how Person schema connects to Organization, Article, and FAQPage schema inside a unified graph, our Schema markup for AEO guide covers the full integration pattern with code examples for each schema type.

The four E‑E‑A‑T signals AI models look for in an author

The Google framing maps cleanly onto signals you can declare in Person schema and supporting content. Here is what each letter means in practice for AI search.

Experience (the first E)

Experience is first hand knowledge of the topic. The AI signal: have you actually done the thing you are writing about? For a marketing operations leader writing about Salesforce administration, the experience signal is years of hands-on Salesforce admin work, documented somewhere outside the article itself. The Person schema fields that carry experience are knowsAbout, which declares the domain, description, which tells the story, and worksFor with a stable role history.

Show experience inside the content with details only practitioners would know. Specific configuration steps. The error you saw three years ago that no documentation explains. The decision matrix you actually used on a real project. AI models are tuned to recognize that depth and weight it differently than reword-of-Wikipedia content.

Expertise (the second E)

Expertise is credentialing and demonstrated skill. The schema fields that carry it are jobTitle, alumniOf, hasCredential if you have formal certifications, and award for industry recognition. On the page itself, expertise shows up as a substantive author bio, links to your body of work, and citations of your prior published work.

Many marketers confuse expertise with seniority. A 25 year career as a generalist does not signal expertise on Person schema as cleanly as a 5 year career with a tight topical focus. AI models match knowsAbout against the topic of the query. Tight focus wins.

Authoritativeness (the A)

Authoritativeness is third party recognition. Who else considers this person an authority? The schema fields are limited here because authoritativeness is mostly off-page. But sameAs is the crucial vector. Every link in your sameAs array to an authoritative profile (LinkedIn, Wikidata, a conference speaker page, an industry association membership, a journal author page) increases the authoritativeness signal AI engines can verify.

Authoritativeness also lives in third party mentions outside your own site. If Search Engine Journal cites you as an AEO expert, that mention will not appear inside your schema, but it will appear in AI training data and in the Knowledge Graph that AI engines reference at retrieval time.

Trustworthiness (the T)

Trustworthiness is the meta signal. The other three feed it. Schema fields that explicitly support trust include a stable url for the person, an image (a real photo, not stock), a clear association with an organization via worksFor, and a complete identity disclosure rather than a pseudonymous byline. Trustworthy authors do not hide.

Trust also lives in what your content gets right over time. If you publish predictions and they pan out, that builds trust. If you correct mistakes openly, that builds trust. AI models with web access can see your track record. Authors who built their entity early and updated content as they learned more send stronger trust signals than authors who appeared yesterday.

Building the Person entity, step by step

Schema by itself is the tip of the iceberg. The Person entity is everything underneath. Six steps to build it.

Step 1: Write the canonical about page

Start with a stable URL on your own domain. Options: /about, /about-me, /team/jane-smith. Whatever you choose, do not change it later. That URL is the identifier AI models will store and reference. Inside the page, put a substantive bio with name, current role, years of experience, areas of specialization, education, notable work, and a real photo.

The bio should be at least 250 words of specifics. Vague bios (“Jane is passionate about marketing”) are useless. Specific bios (“Jane led marketing operations at three Series B startups, including X, Y, and Z. She has spoken at MarTech Conference twice and writes a weekly newsletter on demand generation”) carry the entity signal AI engines can verify against other sources.

Step 2: Add Person schema to the about page

Use the template above. Make the @id stable and tied to the about page URL. Match every claim in the schema to a claim on the visible page. If the page says “Jane has 12 years of experience,” the schema description should say the same thing. Inconsistency between schema and visible content is a credibility signal that goes the wrong direction.

Step 3: Connect the about page to every byline

On every blog post, the author byline should link to the about page. The Article schema for the post should reference the same Person @id rather than redefining the Person object inline. Now AI models see: this post has an author, the author has a dedicated profile page, the profile page has structured data, and the structured data points to other authoritative profiles. The graph closes.

Step 4: Build the sameAs web

LinkedIn first. Wikidata second. A personal site third if it is separate from the company site. Then industry specific profiles like Crunchbase, ORCID, or GitHub depending on your field. Each link should describe the same person in the same way. If your LinkedIn headline says “VP of Marketing at X” and your schema says “CMO at X,” AI models cannot resolve which one to trust, and your entity weakens.

Wikidata is unfamiliar to most marketers, so a quick note. Wikidata is the structured knowledge base that feeds Wikipedia and the Google Knowledge Graph. You can create a Wikidata item for yourself even if you are not famous enough for Wikipedia. The item needs to be supported by independent sources: a few articles you wrote, a few mentions of you elsewhere, a LinkedIn profile. Once the item exists, link to it from your Person schema sameAs array.

Step 5: Generate third party citations

This is the hardest step and the most important one. Person authority depends on being mentioned, quoted, or referenced on sites you do not control. Guest posts on industry publications. Quotes in news articles. Podcast appearances with show notes that name you. Conference talks with speaker pages. Each one is a mention that AI engines can verify, separately, from a separate domain.

Five to ten high quality third party mentions in 12 months can move you from “the schema says you exist” to “the web confirms you exist.” A hundred low quality directory listings will not. Aim for relevance and authority, not volume.

Step 6: Cross link to and from authoritative sources

If you write a guest post for Search Engine Journal, link to your about page from the author bio there. When someone else writes about your work, ask them to link to your about page rather than your homepage. Each inbound link to the about page strengthens that URL as the canonical Person identifier, and reinforces the entity graph AI engines are building behind the scenes.

Where Person schema should live on your site

Person schema has three natural homes on most sites: the about page, the article byline, and the Organization schema reference. Each plays a different role in the graph.

The about page is canonical. This is where the full Person schema lives, complete with sameAs, knowsAbout, and the long description. The @id on this schema is the URL of the about page with #person appended. Once set, do not change it.

Author bylines on individual posts should reference the same @id rather than redefining the Person object. This is cleaner JSON-LD and avoids the duplication problem that has historically confused AI parsers. In the Article schema for a post, the author property can be a short Person object that includes only @id and name, pointing back to the full Person object on the about page.

The Organization schema should reference the Person as a founder, employee, or both, using the same @id. This wires the Person and Organization graphs together. If your organization has multiple authors, each one gets their own Person schema with their own @id, and the Organization schema can reference an array of them.

A common mistake is putting a full Person object inline on every post. AI models will still parse it, but the entity gets fragmented because there is no single canonical declaration to anchor against. The @id reference pattern keeps the entity unified, which is what AI engines need to confidently associate work with a single individual.

Common Person schema mistakes

Five mistakes show up in almost every audit.

1. Vague jobTitle

“Marketing Professional” or “Consultant” tells AI models nothing useful. AI engines try to match jobTitle against query topics. “VP of Demand Generation” or “Salesforce Solutions Architect” gives them a hook. Be specific. The job title should describe the work you actually do, not the level you have reached.

2. knowsAbout as a brag list

The knowsAbout field should hold four to eight things you actually know deeply, not 25 buzzwords. AI models that see knowsAbout arrays with thirty plus items downgrade the credibility signal, because nobody is genuinely an expert in thirty things. Focus is a signal of expertise. Sprawl is a signal of generalism.

3. sameAs links to inactive profiles

A LinkedIn profile last updated in 2019 is worse than no sameAs link at all. AI engines verify sameAs links and stale profiles signal dormant identity. Either update the profile or remove it from the schema. Same rule applies to abandoned Twitter accounts, broken personal sites, and conference pages that have been taken down.

4. Person schema with no Organization link

Person entities are stronger when they are tied to a recognized Organization. If your organization has weak entity signals, prioritize building Organization authority in parallel with Person authority. A strong Person entity attached to a phantom Organization looks suspicious to the Knowledge Graph and weakens the verification chain.

5. Inconsistent name spelling

“Jane Smith” on the about page, “Jane M. Smith” in the schema, “J. Smith” in the byline. Each variant fragments the entity. Pick one canonical name and use it everywhere across schema, bylines, LinkedIn, Wikidata, and social profiles. If you use a middle initial sometimes and not others, choose one form and stick with it.

The supporting signals AI needs beyond schema

Person schema is necessary, not sufficient. The schema declares the entity. The supporting signals prove it.

A visible body of work

AI models that find your Person schema then look for what you have produced. A backlog of articles, talks, podcasts, courses, or research papers. The body of work should be findable on your own site (an /articles or /author archive page) and indexable. If you have published on third party sites, link to those pieces with proper attribution. The body of work is what turns “you say you know X” into “here are 40 things you have written about X.”

An active presence in your domain

You do not have to be a famous voice in your field to have an active presence. You need to be visibly engaged in your domain on at least one platform. LinkedIn is the safest bet for B2B authors. Substack or a personal blog works for niche experts. The platform matters less than the consistency. AI engines look for evidence that you are currently active in the topic, not only historically connected to it.

Third party endorsement signals

Quotes in industry publications. Awards. Speaking engagements. Memberships in professional associations. Each one creates a verifiable claim outside your control that AI engines can use to support the expertise signal. The endorsement matters more when the source has its own entity authority. A quote in Search Engine Land is worth more than a quote on an unknown blog, because AI engines weight the entity strength of the citing source.

Topical consistency

This is the single biggest determinant of how strong your Person entity becomes. Choose your topical lane and stay in it. If you publish on AEO, B2B marketing, demand generation, and content strategy, AI models see expertise in marketing generally. If you publish on AEO and almost nothing but AEO for two years, AI models build a tight association between your name and that topic. They start citing you specifically for it.

Time

The frustrating one. Entity recognition compounds over time. A Person entity built today will not show up in the Knowledge Graph next week. The signals you create now feed AI training data and Knowledge Graph updates over months. The author who started building their entity two years ago has a head start that no amount of schema can compress in a single quarter.

How to measure whether your Person entity is working

Run these checks quarterly. The numbers will move slowly, but the direction tells you whether the work is paying off.

1. Knowledge Panel check

Search your name on Google with quotes. Look for a Knowledge Panel on the right side of the results. The presence of a Knowledge Panel for a Person is the clearest single signal that Google’s Knowledge Graph has built an entity for you. Most authors do not have one initially. The realistic goal is to earn one within 12 to 18 months of concerted entity work.

2. AI engine recognition

Ask ChatGPT, Perplexity, and Google AI Overviews: “Who is [your name]?” Run the query three ways: name alone, name plus job title, name plus topic of expertise. Compare the responses.

If you get “I do not have information on that person,” you are early. If you get a generic response that confuses you with someone else, you have a disambiguation problem and your sameAs web needs strengthening. If you get an accurate response that cites specific work you have done, your entity is working.

3. SameAs verification

Run your Person schema through Google’s Rich Results Test. Confirm that every sameAs URL resolves and that the target page mentions your name in a way that confirms the connection. Stale links are silent failures. They do not throw errors, but they fail to verify when AI engines crawl them.

4. Third party citation count

Set up a Google Alert for your name. Count the mentions you get in a typical month. Distinguish between low value mentions (directory listings, comment threads) and high value mentions (industry publications, podcasts with show notes, news coverage, conference speaker pages). The high value count should trend up over time. If it is flat, your third party outreach needs more attention.

5. Wikipedia and Wikidata status

Check Wikidata quarterly for your entry. Confirm the data is correct, that the linked sources still resolve, and that no one has made changes you do not recognize. If you become notable enough for a Wikipedia entry, monitor that page for accuracy. Most authors will never have a Wikipedia page, and that is fine. Wikidata is the more important target for most.

The longer view

Person authority is the slowest moving piece of an AEO program. Schema can be added in a week. Content reformatting can happen in a month. Backlinks can be earned over a quarter. But a Person entity that AI models recognize and reach for? That takes time, consistent publication, and visible third party validation accumulating across many domains.

The good news is that most of the work is a one time setup followed by a steady drip. Once the Person schema is on your about page and references are wired into your blog posts, you do not redo it. Once Wikidata has your entry, you maintain it once a year. Once your LinkedIn is fully built out, you keep it current. The compounding signal builds from there.

If you want help building Person and Organization authority that AI engines actually recognize, our Entity and Authority service covers the full process from schema implementation to Wikidata creation to third party citation strategy.

The brands and individuals that win AI visibility over the next three years will not be the ones with the cleverest content. They will be the ones AI models can verify. Person schema is the first piece of that proof. Everything else is what backs it up.