How Does Person Schema Help AI Identify Specialists and Their Expertise?
Person schema helps AI identify specialists only a little, and indirectly. It lets you state who a specialist is in a form machines can parse: name, role, employer, credentials, specialties, identifiers and profiles on other sites. Google uses author and profile markup to tell people with similar names apart. But no AI assistant documents using Person markup when it writes an answer, and tests show assistants that fetch a page read its visible text and ignore JSON-LD. What identifies an expert to AI is a clear, visible profile page, backed by independent sources such as licensing registers, professional bodies, publications and press coverage. Use the markup to tie those sources together consistently, and put most of the effort into the profile and the corroboration.
What can Person schema say about a specialist?
| Property | What it states | Example value |
|---|---|---|
name |
The name as used professionally | "Dr. Lena Marsh" |
jobTitle |
Current role | "Consultant Dermatologist" |
worksFor |
Employer, ideally linked to your Organization markup by @id |
The clinic |
hasOccupation |
Occupation, with optional classification code | Dermatologist |
hasCredential |
Licences, board certifications, qualifications | A specialist registration, with the issuing body in recognizedBy |
knowsAbout |
Topics of expertise | "psoriasis", "patch testing" |
alumniOf |
Where they trained | Medical school, residency programme |
memberOf |
Professional bodies | A specialty association |
award |
Awards and honours | A named award |
identifier |
Registry numbers | A licence or registration number, an ORCID iD |
sameAs |
The same person elsewhere | Registry entry, LinkedIn, ORCID, Wikidata |
Schema.org's own definition of knowsAbout is careful: it indicates a topic the person knows about, "suggesting possible expertise but not implying it". That is a fair summary of the whole type. Markup states claims. It does not prove them.
What does an example look like?
A specialist profile page can carry a ProfilePage with the person as its main entity:
{
"@context": "https://schema.org",
"@type": "ProfilePage",
"mainEntity": {
"@type": "Person",
"@id": "https://www.example-clinic.com/team/lena-marsh#person",
"name": "Dr. Lena Marsh",
"jobTitle": "Consultant Dermatologist",
"worksFor": { "@id": "https://www.example-clinic.com/#organization" },
"knowsAbout": ["Psoriasis", "Eczema", "Patch testing"],
"hasCredential": {
"@type": "EducationalOccupationalCredential",
"credentialCategory": "Specialist registration",
"recognizedBy": { "@type": "Organization", "name": "[National medical regulator]" }
},
"memberOf": { "@type": "Organization", "name": "[Dermatology association]" },
"sameAs": [
"https://[regulator register URL for this doctor]",
"https://www.linkedin.com/in/[profile]"
]
}
}
Every statement in the markup should also appear on the visible page, in words.
Does Person markup change how AI answers describe experts?
There is no evidence that it does directly. Google says no special structured data is needed to appear in AI Overviews or AI Mode. When SearchVIU had five assistants, including ChatGPT and Gemini, fetch test pages, none of them used the JSON-LD; they worked from the visible text. An Ahrefs study of pages that added structured data, compared with matched controls, found no measurable lift in AI citations across the schema types it pooled.
Google does use the markup in ordinary search. Its Article documentation recommends author.url or sameAs to help it identify an author, and ProfilePage markup for profile pages. Google's Knowledge Graph draws on many sources, and markup that agrees with them may help Google connect the right facts to the right person. Since Google's AI features draw on Google's systems, that is the indirect route. How author markup relates to citations specifically is covered in whether named authors, expert bios and Person schema increase AI citations.
What do AI systems actually use to recognise expertise?
The same things a careful journalist would check:
- A visible profile page with the person's role, specialties, qualifications, experience and a photo, reachable from your team page.
- Official registers, such as medical, legal, accounting or engineering licence lookups, where the person can be found under the same name.
- Professional bodies that list them as members or fellows.
- Publications, such as papers with an ORCID iD, books or articles under their byline.
- Press and speaking, such as quotes in coverage, conference speaker pages and podcasts that name their role.
- Consistent profiles, such as LinkedIn and directory listings with the same name form, employer and title.
Person markup is most useful as the connecting thread. Its sameAs and identifier properties point to these sources, so a system that does read the markup can match them without guessing.
How to make executives recognizable as entities in AI answers
Assistants answer "Who is the CEO of Example?" and "Who is [name]?" from sources that agree with each other. To make an executive recognisable:
- Publish one canonical bio page per executive, with their full name, title, start date in the role, background and a photo. Link it from your leadership page.
- Use one name form everywhere, including middle initial or not, and keep titles current across your site, LinkedIn, press releases and speaker bios.
- Mark up the bio page with Person inside ProfilePage, with
worksForpointing to your Organization's@id. In your Organization markup, reference executives with properties such asfounderoremployee. - Update Wikidata where it applies. If your company has a Wikidata item, its "chief executive officer" statement (P169) should be current and referenced. A person item for the executive is appropriate only if they meet Wikidata's notability policy, which requires serious, publicly available references.
- Earn independent mentions: interviews, conference talks and coverage that name the person with their role.
- Remove stale references. Former executives still listed on old pages or directories cause assistants to name the wrong person.
Test by asking each assistant who holds the role and what they are known for, and note which sources they cite. The Organization side of this is covered in how to add Organization schema so AI can identify your business.
How do MedicalBusiness and Physician schema types help AI understand medical expertise?
They describe the practice more than the person. In schema.org, MedicalBusiness is a type of LocalBusiness for businesses run by health professionals. It has specific subtypes such as Dentist, Dermatology, Pediatric, Physiotherapy, Pharmacy and Physician. Physician covers "an individual physician or a physician's office considered as a MedicalOrganization". It has two more specific types: IndividualPhysician, for a single practitioner, and PhysiciansOffice. Useful properties include:
medicalSpecialty, using schema.org's MedicalSpecialty values, such as Dermatology or Cardiovascular.availableService, for procedures, tests and therapies offered.hospitalAffiliation, for hospitals where the physician or office has privileges.usNPI, the 10-digit National Provider Identifier for US providers.practicesAt, on IndividualPhysician, for the clinics or hospitals where the practitioner works.
A practical pattern for a clinic is a MedicalClinic or PhysiciansOffice for the practice, with its address, hours and specialties. Each clinician then gets a Person on their own profile page, with credentials and a worksFor link to the practice. A solo practitioner with their own listing can use IndividualPhysician, with an NPI where relevant.
As with Person, the markup restates what the page should already say. For medical questions, assistants and Google's systems lean heavily on registers and established directories, such as the NPPES NPI registry and state licence lookups in the US, or national registers elsewhere, plus the Google Business Profile. Keep those accurate first. How local business markup fits in generally is covered in whether LocalBusiness or Service schema helps AI understand business details.
What mistakes undermine specialist markup?
- Credentials in markup that are not on the page, or not verifiable anywhere.
- Long
knowsAboutlists that claim expertise in everything. sameAslinks to the wrong person, a common error with shared names.- Stale titles and employers after role changes.
- Invented or AI-generated experts. Markup for people who do not exist is deceptive, and assistants checking registers will not find them.
- Duplicated profiles, such as separate pages for the same doctor at each location, with conflicting details.
What might an illustrative case look like?
Consider an illustrative physiotherapy group with 14 clinicians across three sites. Assistants asked for a sports injury specialist in its city named one clinician who had left two years earlier, citing an old directory listing. The group gave each clinician one profile page with their registration number, specialties and training, marked up as Person with worksFor linked to the right clinic. It linked each profile to the professional register, updated its directory listings, and asked the old directory to remove the former clinician. Within two months, assistants named current clinicians and cited their profile pages and the register.
What do people ask about Person schema?
Does every staff member need Person markup?
No. Focus on people whose expertise clients look for, such as clinicians, lawyers, advisers and executives.
Should I link to Wikipedia in sameAs?
Only if an article about that exact person exists. Never link to an article about someone with the same name.
Can Person markup create a knowledge panel?
Not by itself. Knowledge panels come from Google's Knowledge Graph, and the subject can claim one if it exists.
Is IndividualPhysician a type of Person?
No. In schema.org it sits under medical organisation and local business types, so use Person for the practitioner's biography.
Published by The Kovmere Team · updated Sep 29, 2026 · More resources