An attorney bio page earns AI citations in 2026 when it carries schema.org Person markup wrapped in ProfilePage, a sameAs array pointing at the lawyer’s state bar directory profile plus Justia, Avvo, Martindale-Hubbell and LinkedIn, and credential detail specific enough to verify against an outside record. InterCore Research ran 540 high-intent legal prompts through Perplexity Sonar in July 2026 and logged 18,900 source citations: a legal directory was the first source cited in 77.8% of answers, and the single most-cited law firm site appeared just 36 times across the whole set. Directories win because they publish structured, disambiguated attorney identities that the Google Knowledge Graph and every major model can reconcile, and the average firm bio page publishes a headshot and four paragraphs of prose.
That gap is the whole story. Attorney bios are usually the highest-traffic page type on a law firm site, with legal marketing agencies reporting that visitors spend 55% to 75% of their on-site time inside them, and Moiré Marketing’s research found 78% of in-house counsel rely on attorney profiles when hiring outside firms. Yet those same pages are the weakest entity assets a firm owns. The ABA’s 2026 Profile of the Legal Profession counts 1,374,720 active US lawyers, so name collisions are the rule, and a page that cannot prove which “Michael Chen” it describes gives ChatGPT, Perplexity, Google AI Mode and Microsoft Copilot nothing to anchor to.
Why do AI engines ignore most attorney bio pages?
Because the page states claims without supplying anything an engine can verify against an external record. InterCore’s July 2026 study found that 34.2% of the firms Perplexity named had no page of their own among the cited sources, meaning the model knew the firm entirely through Justia, Avvo, Super Lawyers and Lawyers.com listings. The firm’s own bio existed. It just was not usable as evidence.
Three failures cause it. The first is prose without structure. A bio that says an attorney “has spent more than two decades helping injured families across the region” contains no parseable fact. No year, no jurisdiction, no bar number, no court. Retrieval systems chunk pages and score chunks for specificity, and vague prose scores near zero on every extraction pass.
The second is orphaned identity. The bio names the lawyer but never links outward to any independent profile, so there is no path from the page to the state bar record, the Justia profile, or the LinkedIn account. The engine gets one unconnected node instead of a corroborated entity. Justia was cited in 66% of InterCore’s answers precisely because a Justia profile carries a bar admission, an admission year, a jurisdiction, and cross-links, all in a predictable shape. We break the directory dynamic down in legal directories for AI visibility.
The third is missing or wrong markup. Schema.org deprecated the standalone Attorney type, and plenty of law firm sites still ship it, or ship LegalService on a bio page, or ship nothing. Without markup the engine guesses which entity the page describes, and on a page about one of 1.37 million American lawyers, guessing means skipping.
Most firms have never seen what an AI engine actually knows about their attorneys. Get your free AI visibility audit and see which of your lawyers the models can identify by name, and which ones they cannot place at all.
What schema does an attorney bio page need in 2026?
ProfilePage as the page type, with mainEntity pointing to a Person node that carries jobTitle, worksFor, alumniOf, hasCredential, knowsAbout, and sameAs. Google Search Central documents ProfilePage as the markup for a page primarily about one person affiliated with the site, which is precisely what a bio page is, and it is the wrapper almost no law firm uses.
The Person node does the heavy lifting. jobTitle should be the practice-specific title the market actually searches, “Family Law Attorney” or “Board-Certified Personal Injury Trial Attorney,” not “Partner.” worksFor points to the firm’s LegalService entity by @id so the lawyer and the firm resolve as connected nodes rather than two strangers. alumniOf takes CollegeOrUniversity entries for the JD and undergraduate degree. knowsAbout takes practice areas as separate values, and it is the property that lets a model answer “who handles construction defect litigation in Charleston” without reading your prose.
hasCredential separates a real legal entity from a marketing page. Each bar admission becomes an EducationalOccupationalCredential with a credentialCategory, a recognizedBy pointing at the licensing body, and a dateCreated for the admission year. Board certifications get the same treatment. The full firm-level implementation is in our legal schema markup guide.
{
"@context": "https://schema.org",
"@type": "ProfilePage",
"mainEntity": {
"@type": "Person",
"@id": "https://example-law.com/attorneys/dana-reyes/#person",
"name": "Dana Reyes",
"jobTitle": "Board-Certified Family Law Attorney",
"worksFor": { "@id": "https://example-law.com/#legalservice" },
"alumniOf": [
{ "@type": "CollegeOrUniversity", "name": "University of Texas School of Law" }
],
"knowsAbout": ["Child custody litigation", "High-asset divorce", "Marital property division"],
"hasCredential": [
{
"@type": "EducationalOccupationalCredential",
"credentialCategory": "Bar Admission",
"recognizedBy": { "@type": "Organization", "name": "State Bar of Texas" },
"dateCreated": "2009"
},
{
"@type": "EducationalOccupationalCredential",
"credentialCategory": "Board Certification",
"recognizedBy": { "@type": "Organization", "name": "Texas Board of Legal Specialization" },
"dateCreated": "2016"
}
],
"sameAs": [
"https://www.texasbar.com/AM/Template.cfm?bar-profile-url",
"https://lawyers.justia.com/lawyer/dana-reyes",
"https://www.avvo.com/attorneys/dana-reyes.html",
"https://www.martindale.com/attorney/dana-reyes",
"https://www.linkedin.com/in/danareyeslaw"
]
}
}
How do you build a sameAs array that AI engines trust?
Tier it, and never include a URL the firm has not claimed. A sameAs array asserts that all these profiles describe one person, and engines weight those assertions by the independence and authority of the target. Five verified links beat twenty scraped aggregator pages every time.
Tier 0, the licensing record. The state bar directory profile: State Bar of Texas, State Bar of California, the North Carolina State Bar, whichever applies. This is the only sameAs target that constitutes proof of license rather than a self-published claim, and it should always be first in the array. Add federal court admissions where the court publishes a public roster.
Tier 1, the legal graph. Justia, Avvo, Martindale-Hubbell, Super Lawyers, Lawyers.com, Best Lawyers, and the relevant practice section page at the American Bar Association or a state bar association. These are the domains that dominated InterCore’s citation counts, and linking to them is how you tell an engine that the entity in its directory index and the entity on your site are the same person. Claim and correct each profile first, because a sameAs pointing at an outdated Avvo listing propagates the error instead of fixing it.
Tier 2, the general graph. LinkedIn, a Google Scholar page if they publish, an ORCID iD for lawyers with academic output, a Wikidata item where notability supports one, and any authored column page at the ABA Journal, Above the Law, or a state legal weekly. Wikidata matters more than its traffic suggests because it feeds the Google Knowledge Graph directly, the substrate several assistants query to confirm a person exists.
How do AI engines tell two lawyers with the same name apart?
By triangulating three signals: structured identity on the page, cross-source agreement, and unique numeric identifiers. With 1,374,720 active US lawyers per the ABA, name collisions are constant, and there are hundreds of practicing attorneys named some version of David Miller. The engine does not resolve this by reading better. It resolves it by finding a stable identifier that appears in more than one place.
Give it one. Publish the state bar number in visible page text, not just in schema, alongside the admitting jurisdiction and the admission year. That triplet is functionally unique and appears identically on the bar’s own public record, so an engine comparing your page against the licensing database gets an exact match instead of a fuzzy name match.
Then reinforce it with anchors an engine can cross-check: the courthouses and counties the attorney appears in, the practice sections they belong to, the law school and graduation year, and the firm’s LegalService @id. Where an attorney has real third-party coverage, a Wikidata item with a P106 occupation statement and links back to the bar profile makes the identity durable across engines rather than dependent on any one index. The mechanics of building a person entity models recognize are in author bylines for AI search.
Which credentials and proof details actually move citations?
The ones with a date, a number, a jurisdiction, or a named third party attached. Everything else is decoration. An engine extracting facts from a bio is looking for statements that survive a verification pass, and “highly respected trial attorney” does not survive anything.
Here is the difference in practice.
Before: “Dana Reyes is a highly experienced family law attorney who has helped countless families through difficult times. She is passionate about protecting children and has been recognized for her outstanding work in the community.”
After: “Dana Reyes is a family law attorney admitted to the State Bar of Texas in 2009 (Bar No. 24XXXXX) and board certified in Family Law by the Texas Board of Legal Specialization since 2016. She has tried more than 60 contested custody matters in Travis and Williamson County district courts, chairs the Austin Bar Association Family Law Section, and was selected to Texas Super Lawyers in 2023, 2024, 2025 and 2026. She teaches the annual UT Law CLE session on high-asset property division and has been quoted on custody jurisdiction in the ABA Journal and Texas Lawyer.”
Same person, same length range. The second version contains fourteen verifiable facts, six named institutions, and four dates, and it gives a model complete, quotable sentences, which is what actually gets lifted into an AI answer. Add the two sections most bios skip entirely: a publications and speaking list with dates and venue names, and a representative matters block written within your state’s bar advertising rules.
What does a finished 2026 attorney bio page look like?
Run this checklist against every lawyer on the site. A page that clears all twelve items is a page the engines can cite.
- URL at /attorneys/firstname-lastname/, one page per lawyer, never a shared team page.
- H1 is the attorney’s full name exactly as it appears on the state bar record, with suffix if used there.
- First 40 words state name, role, admitting jurisdiction, admission year, and primary practice area in plain sentences.
- Bar admissions block listing each jurisdiction, admission year, and bar number, plus every federal court admission.
- Board certifications with the certifying body named in full.
- Education with institution, degree, and year for JD and undergraduate.
- Representative matters or case types written to bar advertising rules, with case types and venues rather than adjectives.
- Publications and speaking as a dated list with real venue and publication names.
- Memberships and leadership with section names and role dates.
- Outbound profile links in visible page text to the bar profile, Justia, Avvo, Martindale-Hubbell, Super Lawyers, and LinkedIn, matching the sameAs array exactly.
- ProfilePage plus Person schema validated in Google’s Rich Results Test and the Schema Markup Validator, with worksFor resolving to the firm’s LegalService @id.
- Author linkage so every blog post, practice area page, and FAQ the attorney contributed to carries a byline that links here, and the bio links back to that content.
Validate the markup, then test the outcome. Ask ChatGPT, Perplexity, Google AI Mode and Microsoft Copilot who the attorney is by name and city, and record whether the answer names the correct firm, jurisdiction, and practice area. That result, not the schema, is the metric.
Frequently asked questions
Should each attorney have their own bio page or is a team page enough? Each attorney needs a dedicated URL. A combined team page forces every lawyer’s identity into one document, which makes Person schema ambiguous and gives Google AI Overviews no single canonical page to attribute a citation to. One page per lawyer at /attorneys/firstname-lastname/ also gives you a clean sameAs target to reference from Justia, Avvo, Martindale-Hubbell and LinkedIn profiles, which is how the corroboration loop closes.
How long should an attorney bio be for AI visibility? Between 600 and 1,200 words for the main narrative, plus structured blocks for admissions, education, publications, and matters. Legal marketing guidance has long recommended 300 to 500 words, and that range is now too thin. Retrieval systems chunk pages and score each chunk, so more verifiable facts means more citable chunks. Length only helps when every added sentence carries a date, a named institution, a venue, or a number.
Is Attorney schema still valid on bio pages in 2026? No. Schema.org deprecated the standalone Attorney type, and it should be replaced with Person markup carrying a practice-specific jobTitle, plus worksFor pointing at the firm’s LegalService node. Wrap the page in ProfilePage per Google Search Central’s documentation. Firms still shipping deprecated Attorney markup are sending a type that validators flag and that no current engine treats as a person entity.
How many sameAs links should an attorney bio include? Five to ten, all claimed and accurate. Start with the state bar directory profile, then Justia, Avvo, Martindale-Hubbell, Super Lawyers, and LinkedIn. Add Google Scholar, ORCID, or Wikidata only where the attorney genuinely has that presence. Quality beats volume: one stale Avvo listing with a wrong practice area does more damage than five missing links, because the engine sees contradiction rather than absence.
Do bar numbers on a public bio page create any risk? No. Bar numbers are public record, published by every state bar in its own directory and printed on court filings daily. Putting the number in visible text and in hasCredential markup is the clearest disambiguation signal available for a lawyer with a common name. Confirm your state bar’s advertising rules on how admissions and certifications must be phrased, then publish the number alongside jurisdiction and admission year.
How fast do bio page changes show up in AI answers? Expect four to twelve weeks. Google’s index refresh and Knowledge Graph reconciliation drive most of it, and assistants that query live search, including Perplexity and Microsoft Copilot, reflect changes faster than models relying on training data. Directory corrections at Justia and Avvo often propagate before your own site changes register, which is another argument for fixing the external profiles first.
The bottom line
Your firm already paid for these pages. They are built, they are indexed, and they absorb more visitor attention than anything else on the site. What is missing is the machine-readable layer that lets an engine confirm a specific licensed human stands behind the claims, and that layer costs a few hours per attorney. Skip it and you stay in the 34.2% of firms AI describes secondhand, through a Justia listing you do not control, inside an answer format Ahrefs measured cutting clicks to the top organic result by 58% as of February 2026.
Start with the lawyers who bring in the most business and work down the roster. Get your free AI visibility audit and see exactly which bio pages the engines can verify today and which ones read as strangers.
Tagged