AEO for Healthcare: How Practices Get Cited in AI Answers

  • Healthcare SEO
AEO for Healthcare

Answer engine optimization (AEO) for healthcare is the work of making a medical practice’s details easy for AI answer engines to find, verify and cite. Google’s guide says optimizing for generative AI search is still SEO, so the foundations are unchanged: crawlable pages, consistent practice data and clear expert content.

Key takeaways

  • We recommend funding AEO inside the SEO plan. Google says there are no additional requirements to appear in AI Overviews or AI Mode, and a page must be indexed and eligible to be shown with a snippet.
  • We read a June 2026 vendor study by Halcy as a case for consistency beyond your own site: practice-owned websites drew under a quarter of ChatGPT’s citations in it.
  • Our November 16, 2026 check of 8 ranking pages found that none mentions AI Mode and only 3 mention the Health Insurance Portability and Accountability Act (HIPAA). We read both as gaps worth filling.

How does AEO differ from SEO and GEO?

We treat AEO and GEO as parts of SEO, not as separate programs. Generative engine optimization (GEO) is the same work aimed at text written by a large language model (LLM). Google’s guide treats both labels as ordinary SEO work. The labels differ in what the patient sees and what the practice controls:

  • SEO. The patient sees a ranked list of links. The practice controls crawlable pages, relevance and page experience.
  • AEO. The patient sees a direct answer that may link to sources. The practice can shape plain-text answers, consistent practice facts and citable third-party profiles.
  • GEO. The patient sees a generated answer that may draw on retrieved pages. The practice controls the same inputs, plus monitoring which sources each model cites.

This page replaces SEO Rank Health’s October 2025 post on Search Generative Experience (SGE). Google’s documentation calls the feature AI Overviews, so this page uses that name.

Where do patients meet AI answers?

We recommend planning for at least four places instead of tuning for one engine, because a practice cannot choose which one a patient opens. Patients now meet generative artificial intelligence (AI) answers in Google AI Overviews, Google AI Mode, ChatGPT search and Perplexity. Google shows AI Overviews in results and offers AI Mode for queries that need further exploration, reasoning or complex comparisons.

A KFF poll of 1,343 U.S. adults, fielded February 24 to March 2, 2026, found that 32% of the public reports turning to AI chatbots for physical or mental health advice in the past year. We read that as enough demand to justify checking what these tools say about your practice. The 32% figure measures chatbot use for health advice, not provider selection, so it is a reason to check and not a traffic forecast.

What do AI engines read about a practice?

For a practice, the website is one source among several, and the third-party pages an answer engine reads about it deserve attention too. In one public dataset, practice-owned websites were 23.0% of ChatGPT’s citations, and the other 77.0% went to other source types.

Halcy founder Nathan Woo published the dataset in MGMA Insights in June 2026. It covers 4,950 queries across 200 practices, 20 metros and 10 specialties, run in May 2026 through OpenAI’s application programming interface (API), with 33,340 citation events. These were the six largest source types:

  • Hospital system and academic medical center staff rosters: 27.5% of citations.
  • Practice-owned websites: 23.0%.
  • Specialty associations and credentialing authorities: 22.1%.
  • National directories such as Healthgrades, Zocdoc, WebMD, RealSelf, Vitals and Doximity: 8.3%.
  • State medical boards and state .gov sites: 5.1%.
  • Review aggregators: 3.8%.

Hospital rosters and specialty-association credentialing pages together carry 49.6% of the cited surface, according to the MGMA Insights article, more than twice the practice-owned share. We read that as the strongest reason to confirm every clinician’s roster entry and association listing.

The MGMA Insights article reports that Google Business Profile (GBP) has zero direct citations because Google does not expose reviews as crawlable URLs. Google’s own guide says Business Profiles can help products and services be visible in AI responses, so we recommend keeping the profile accurate for Google’s surfaces.

The MGMA Insights article calls its analysis exploratory and correlational, and the study tests one platform, covers the U.S. only and comes from a commercial vendor, so treat these shares as places to check first.

The practical read is consistency. Keep one fact sheet with the practice’s name, address and phone number (NAP), specialty, National Provider Identifier (NPI) and hospital affiliations, then check every directory listing and the Google Business Profile against it.

Which crawlers should a practice allow, and what does llms.txt do?

We recommend allowing the search crawlers of the engines where you want to appear and deciding separately whether to opt out of model training. The crawlers fall into four groups:

  • Answer and search: OAI-SearchBot, Claude-SearchBot and PerplexityBot. We recommend allowing all three, because a crawler that cannot read a page cannot retrieve it.
  • Model training: GPTBot and ClaudeBot. Opting out is a policy decision, not an AEO requirement.
  • Training and grounding: Google-Extended. Google says this token controls Gemini training and grounding in Gemini Apps, and does not affect inclusion in Google Search.
  • User-initiated fetches: ChatGPT-User, Claude-User and Perplexity-User. Include these when auditing firewall rules.

A practice that wants answers but not training by OpenAI and Anthropic could publish this:

# Answer and search bots: allowed

User-agent: OAI-SearchBot

Allow: /

User-agent: Claude-SearchBot

Allow: /

User-agent: PerplexityBot

Allow: /

# Training bots: opted out

User-agent: GPTBot

Disallow: /

User-agent: ClaudeBot

Disallow: /

# Google-Extended is left out on purpose. Google says it controls Gemini Apps

# grounding as well as training, so blocking it also removes that grounding.

Google documents that a crawler follows only the group with the most specific matching user agent, without combining it with the * group. If your * group blocks portal or admin paths, repeat those rules in each bot-specific group, and check each vendor’s documentation.

The llms.txt file is a proposal, not a standard: a Markdown file that summarizes a site for language models. Google says creating one is fine but neither harms nor helps visibility in Google Search, which ignores these files, so we rank it below crawler access and consistent facts.

On our own site, as of November 16, 2026, our crawler’s robots.txt probe found none of the 13 AI user agents it tests blocked (4 answer agents in our crawler’s grouping, which includes Google-Extended; 3 user-initiated; 6 training).

For a practice, a probe that finds nothing blocked shows that robots.txt is not what stops these bots, and that training agents stay allowed until someone writes a rule. The probes use one crawler’s user agent and do not test firewall rules, so we recommend confirming bot access in server logs.

Which schema still matters after FAQ rich results ended?

We recommend keeping schema.org markup for accuracy, not for FAQ rich results. Starting May 7, 2026, FAQ rich results no longer appear in Google Search, according to the Search Central updates log, and on June 15, 2026 Google removed the feature’s documentation. Neither entry tells sites to remove existing FAQPage markup, and Google’s AI features page says no special schema.org markup is needed for AI Overviews or AI Mode.

Google’s local business documentation says to use the most specific LocalBusiness subtype and requires name and address. For a medical practice, we recommend the matching specific type, such as Physician or MedicalClinic, filled only with facts from the fact sheet: specialty, hospital affiliation and NPI. Treat this markup as an accuracy aid, not a documented ranking input.

Keep visible FAQ sections for patients, and keep FAQPage markup only if it mirrors the visible answers, as this page does. Google’s structured data policies say not to mark up content that is not visible to readers of the page. Our Physician and MedicalClinic markup guide predates the May 2026 change.

How do you measure AI-referred patients without exposing health data?

AI search for healthcare has a measurement problem, and HIPAA sets the limit, so we recommend sources that do not need patient records. The U.S. Department of Health and Human Services (HHS) says tracking technologies on user-authenticated pages generally have access to protected health information (PHI). HHS also says an unauthenticated scheduling page can collect an email address or reason for seeking care, and then the HIPAA Rules apply to the vendor disclosure.

A federal court in Texas vacated one part of the HHS tracking guidance on June 20, 2024. The vacated part said HIPAA obligations are triggered when a tracking technology connects an Internet Protocol (IP) address with a visit to an unauthenticated public page on health conditions or providers. As of November 16, 2026, the HHS page still says HHS is evaluating its next steps.

Our recommendation is to keep advertising and analytics tags off pages where patients enter health information, such as scheduling tools, symptom checkers and portal logins. The exception is a vendor that has signed a business associate agreement (BAA), with your compliance officer confirming the disclosure is permitted. Our HIPAA-safe analytics guide covers tag setup, and this is not legal advice.

Measure AI visibility with methods that do not need patient records:

  • Search Console. As of August 31, 2026, Google says the Generative AI performance report in Search Console is available to all websites worldwide, showing impressions in AI Overviews and AI Mode by page, country, date and device.
  • Analytics. We recommend segmenting ChatGPT sessions, because ChatGPT adds utm_source=chatgpt.com to referral URLs for publishers that allow OAI-SearchBot.
  • Prompt panel. We recommend running a fixed set of patient-style prompts, with no patient details, many times in each engine and reporting how often the practice is named or linked, because answers vary between runs.

What does the current SERP for “aeo for healthcare” show?

Our reading is that the pages ranking for this query mention ChatGPT (7 of 8) and AI Overviews (6 of 8) but mostly skip AI Mode and HIPAA. A practice weighing AEO advice should ask any adviser, including us, how it handles both.

On November 16, 2026, we took the top organic results for this query from Ahrefs (United States) and excluded one LinkedIn post. We fetched each of the remaining 8 pages once with our crawler, without JavaScript rendering. A term counts if it appears anywhere in the page source with tags stripped, including navigation, footer and script text. Matching ignores case.

  • AI Mode, 0 of 8 pages. None mentions it, although Google’s documentation and Search Console report cover AI Mode alongside AI Overviews.
  • HIPAA, 3 of 8. Fewer than half mention it. We treat AEO advice that does not mention HIPAA as incomplete for a healthcare provider, because referral measurement is where AEO work meets HIPAA.

The study covers 8 pages for one query on one day. November 16, 2026 fell in the first week of Google’s November 2026 core update, which began on November 10, so rankings may still have been moving. Term counts match exact phrases and measure mention, not accuracy or depth, and anything added by JavaScript is invisible to the check.

What should a practice do in the first 30 days?

The order matters: access and facts come before content and measurement, because better writing cannot fix a page that crawlers cannot reach or a fact that conflicts across listings. Each step has an owner and a check a second person can repeat.

  1. Week 1, crawler access. Allow answer bots in robots.txt and any firewall or hosting rules. Owner: web developer. Verify: server logs show no 403 responses and at least one 200 response to OAI-SearchBot, Claude-SearchBot or PerplexityBot.
  2. Week 1, practice facts. Write one fact sheet with legal name, address, phone, hours, services, insurers, clinicians, NPIs and hospital affiliations. Owner: practice manager. Verify: a second person compares each live profile with the sheet and finds no differences.
  3. Week 2, listings and markup. Match every listing and the Physician or MedicalClinic markup to the sheet. Owner: marketing manager and web developer. Verify: dated screenshots of each listing, and a validator report with no errors.
  4. Week 3, service pages. Rewrite top service pages to answer the patient’s question first, with a named clinician reviewer. Owner: content lead and clinician. Verify: page source contains the answer, and the reviewer approves it in writing.
  5. Week 3, measurement and compliance. Review the Generative AI performance report and audit tags on scheduling and portal pages. Owner: analytics owner and compliance officer. Verify: every tag on those pages has a signed BAA or is removed.
  6. Week 4, prompt panel. Run a fixed panel of patient-style prompts many times in each engine. Owner: SEO lead. Verify: a sheet logs presence rate per prompt and engine, and the panel reruns in 30 days.

For outside help, the SEO audit page describes a full review of technical health, content, backlinks and competitors. The E-E-A-T checklist covers clinician bios and review dates.

FAQ about AEO for healthcare

What is AEO vs SEO?

SEO aims to rank pages in search results, while AEO aims to make a practice’s information the source that answer engines quote or link. Google’s guide says optimizing for generative AI search is still SEO, so AEO works best as an extension of SEO.

What does AEO stand for?

AEO stands for answer engine optimization, and its sibling term GEO stands for generative engine optimization. Google’s guide gives both expansions and treats both as ordinary SEO work. AEO for healthcare applies the same work to medical practices, clinics and hospitals.

What is AEO and how does it work?

AEO prepares a practice’s pages and profiles so AI answer engines can retrieve and cite them. Google says AI Overviews and AI Mode may use query fan-out, which issues multiple related searches across subtopics and data sources. To be a supporting link, a page must be indexed and eligible to be shown with a snippet.

Is AEO replacing SEO?

No. Google says the best practices for SEO remain relevant for AI features in Google Search. We treat AEO as an extension of SEO that adds consistency work across profiles and new measurement, such as a prompt panel, rather than a replacement.

How do you get recommended by ChatGPT?

No tactic guarantees a recommendation. We recommend two things: allow OAI-SearchBot, since blocking it keeps a site out of ChatGPT’s search answers, and keep the practice’s facts consistent across the sources ChatGPT cites. Judge progress by how often the practice appears across many runs of the same prompts.

Written by

Oleg Silin

Head of SEO & Content

Agency co-founder who leads SEO and content for 20+ clients and 30+ websites, specializing in recovering traffic lost to Google algorithm updates.

Get a Free Strategy Session

We get to know your business and goals to find the approach that works for you.