By Kevin Doherty

Try this. Open ChatGPT. Type something a real prospective patient would actually ask: “best functional medicine doctor for Hashimoto’s in [your city].” Or “naturopathic doctor near me for chronic fatigue.” Or “where do I find a Five Element acupuncturist for anxiety.” Or “integrative psychiatrist who works with trauma.”

If you’re like most holistic and integrative practitioners, you’ll watch ChatGPT recommend other practitioners — sometimes practitioners you know personally, sometimes practitioners you’ve never heard of, sometimes hospital systems that barely practice integrative medicine — while your own practice doesn’t appear at all. You can repeat the test on Perplexity. You can repeat it on Google AI Overviews. You can repeat it on Claude. The result is usually the same. You’re invisible at the moment patients are deciding which practitioner to call.

This is the structural change in healthcare patient acquisition that most practitioners haven’t fully absorbed. ChatGPT crossed a billion users. Perplexity, Claude, and Google AI Overviews now answer patient questions directly within the search interface, often without the patient ever clicking a website. The patient who used to evaluate ten practitioner websites before booking now reads the AI’s synthesized recommendation and books one of the three or four practices the AI mentioned. Everything outside that small set of cited practices becomes invisible regardless of clinical excellence, credentialing depth, or marketing investment.

The fact that your practice doesn’t appear in AI search results isn’t an accident. It’s the predictable result of how most practitioner websites are built — including the ones built by website providers who advertise “AI optimization” as part of their service. The infrastructure that produces AI citation is specific and technical, and the dominant industry providers haven’t built it. This article covers what’s actually happening, why most websites fail, and what the work of fixing it looks like.

How AI Search Actually Decides Which Practices to Cite

Understanding why your practice doesn’t appear requires understanding how AI search systems decide what to cite. The decision process is technical, and most practitioner website providers don’t engineer for it.

AI systems like ChatGPT, Perplexity, Claude, and Google AI Overviews generate responses through a combination of two mechanisms. First: parametric knowledge — what the model learned from its training data, which includes a large slice of the web. Second: real-time retrieval — live web lookups via Bing’s index (for ChatGPT and Copilot), Google’s index (for Google AI products), and various other indices for the rest. Roughly sixty percent of ChatGPT responses come from parametric knowledge, while forty percent involve real-time lookups, depending on query specificity.

For local practitioner queries — “best naturopath in Portland,” “fertility acupuncturist near me,” “functional medicine doctor for autoimmune in San Diego” — AI systems lean heavily on real-time retrieval. The system performs a search, evaluates the top results, and synthesizes an answer drawing from the sources it considers most authoritative. The websites cited in that synthesis become the practices the patient ends up considering.

The evaluation process that determines which sources get cited weights specific signals. Schema markup that identifies the practice and its credentialing. Content depth that demonstrates expertise rather than just claiming it. Answer-first formatting with question-answer structure that maps cleanly to user queries. Original substantive content rather than templated library content. Credentialing data structured for extraction. Entity authority signals that let the AI confidently recommend the practice. The websites that have these signals get cited at substantially higher rates than websites that don’t.

Most holistic and integrative practitioner websites have few of these signals. The dominant industry providers — subscription template services, generic medical website providers, cross-modality designers — haven’t built them. The practitioner ends up with a website that competes for AI citation against thousands of identical templated sites and rarely wins, regardless of clinical excellence.

The Eight Specific Reasons Your Practice Doesn’t Appear

1. Missing or incomplete schema markup

Schema markup is structured data that tells AI systems explicitly what your website is about. Without it, AI systems have to infer everything from raw text, which they do imperfectly. With it, AI systems get a clean structured signal about the practice, the practitioner, the credentialing, and the services offered.

The schema types that produce AI citation for practitioner websites are specific. Physician schema with NCCAOM credentialing data for acupuncturists, IFMCP and A4M certifications for functional medicine practitioners, AANP and accredited school authority for naturopathic doctors, ABoIM board certification for integrative MDs, NBHWC certification for health coaches, SEP certification for somatic practitioners, and the practitioner-type-specific credentialing for every other modality. MedicalOrganization schema. MedicalSpecialty schema specifying the actual specialty. LocalBusiness schema with consistent NAP data. FAQPage schema marking up question-answer content. Article schema for substantive content pieces. Speakable schema for AI voice systems.

Most practitioner websites have either no schema or basic incomplete schema. The websites that have comprehensive practitioner-type-specific schema architecture get cited at substantially higher rates than those that don’t.

2. Licensed library content distributed across hundreds of identical sites

The dominant content model in practitioner website services is licensed library content. AcuPerfect’s plans bundle articles like “About Acupuncture” and “What to Expect on Your First Visit” distributed across the entire subscriber base. ChiroMatrix and similar chiropractic providers do the same. Generic medical website providers (WebToMed, Doctor Multimedia) license content across many medical specialties. The economic logic from the provider’s perspective is clear: produce content once and distribute it across many subscribers to dramatically lower per-client production cost.

The problem for AI citation is structural. AI systems explicitly devalue content that appears across many identical sites. The system’s job is to cite authoritative original sources. When the same article exists on three hundred practitioner sites with minor customization, the AI system treats all three hundred as duplicate content and rarely cites any of them. The practitioner with library content competes against thousands of identical pages and almost never wins citation share for queries that matter.

3. Insufficient content depth

AI systems weight content depth heavily when choosing what to cite. The websites that get cited consistently have substantial content depth — typically 8,000 words minimum of substantive authority content addressing the conditions, specialty, and clinical reasoning the practice actually focuses on.

Most practitioner websites have substantially less. A typical chiropractor website has 1,500-3,000 words across all pages combined. A typical acupuncture website has 2,000-4,000 words. A typical naturopathic or functional medicine website has 3,000-6,000 words. The website with 6,000 words competes against websites with 30,000 words for AI citation, and the deeper website almost always wins.

Content depth doesn’t mean keyword stuffing or repetitive padding. It means substantive treatment of specific clinical territories — pillar articles addressing the practice’s primary specialty, condition-specific articles addressing the conditions actually treated, authority pages establishing the practitioner’s training and clinical philosophy. Original content, written specifically for the practice, addressing the actual clinical work with depth.

4. Credentialing buried in plain text instead of structured data

Holistic and integrative practitioners hold credentialing depth that distinguishes them from the broader unregulated wellness landscape. NCCAOM Diplomate status (Dipl.Ac., Dipl.OM, Dipl.CH, Dipl.ABT). IFMCP certification through the Institute for Functional Medicine. AANP membership and accredited school graduation (Bastyr, NUNM, SCNM, CCNM, BINM, UB). ABoIM board certification for integrative MDs. NBHWC certification for health coaches. SEP three-year training for Somatic Experiencing Practitioners. Specialty board certifications across psychiatry, internal medicine, and other specialties.

This credentialing matters substantially for AI citation because AI systems weight authority signals heavily — particularly when the system needs to distinguish licensed practitioners from the broader wellness landscape using similar terminology. A licensed acupuncturist with NCCAOM credentialing is different from a dry needling practitioner. An IFMCP-trained physician is different from someone using “functional medicine” terminology without formal training. An ABoIM-board-certified integrative MD is different from a cosmetic-focused MD using “wellness” terminology.

Most practitioner websites list credentials on the About page in plain text. AI systems can extract some of this through inference, but inference is less reliable than structured data. The website that surfaces credentialing through Person schema with structured certification data, MedicalSpecialty schema specifying the actual specialty, and links to certifying body verification provides authority signals AI systems extract reliably and weight in citation decisions. Most practitioner websites don’t.

5. Missing FAQ-formatted content with answer-first structure

AI systems preferentially cite content formatted in question-answer structure. The “answer capsule” — a short, direct, declarative answer to a specific question — is the structural unit AI systems extract most reliably for citation. When a website has FAQ-formatted content with clear answer capsules under each question, AI systems can extract those capsules cleanly and use them in synthesized responses.

Most practitioner websites have minimal FAQ content. What FAQ content they do have is typically operational — “Do you accept insurance?” “What are your hours?” — rather than clinical. The clinical questions patients actually ask AI systems before booking — “Is acupuncture effective for IVF support?” “What’s the difference between functional medicine and integrative medicine?” “Can a naturopathic doctor prescribe medications in [state]?” “Does Five Element acupuncture work for anxiety?” — rarely appear in FAQ format on practitioner websites, marked up with FAQPage schema, with substantive answer capsules.

Without FAQ content marked up with proper schema, the practitioner forfeits one of the strongest citation signals available. AI systems can’t extract what isn’t structured for extraction.

6. No Speakable schema for AI voice systems

Speakable schema marks specific content sections for AI voice systems — Google Assistant, Alexa, Siri, ChatGPT voice mode, and the rest. When patients ask voice queries about practitioners, AI systems extract content from Speakable-marked sections preferentially.

Most practitioner websites have no Speakable schema. The marketing language has gotten ahead of the technical reality — “voice search optimization” appears on website service sales pages where the actual implementation isn’t built. Voice queries about practitioners are growing as a percentage of total queries, and the practitioners with proper Speakable implementation gain citation share that practitioners without it surrender by default.

7. Weak entity authority signals

AI systems decide which practices to cite partially based on entity authority — the system’s confidence that this practice is a real, recognized, authoritative entity in its space. Entity authority gets built through several signals. Wikipedia presence (one of the strongest signals — Wikipedia accounts for roughly 7.8% of all ChatGPT citations). Consistent NAP (name, address, phone) data across the web. Co-citation with established names in the field — being mentioned alongside authoritative practitioners or organizations strengthens entity recognition. Citations on third-party authoritative sites — directories, professional associations, specialty organizations.

Most holistic and integrative practitioners have weak entity authority. They have no Wikipedia entry (which is a long-term project given Wikipedia’s notability requirements). Their NAP data is inconsistent across directories. They aren’t co-cited with established names in their specialty. Their professional association memberships and specialty board certifications aren’t surfaced in ways that build entity authority. The practitioner with twenty years of clinical excellence and substantial credentialing depth gets represented online as a low-confidence entity by AI systems that can’t verify the claims.

8. Content structured for last-decade SEO rather than AI citation

The SEO best practices that worked through the 2010s and early 2020s — keyword density, link building, on-page optimization for ranking factors — were optimized for keyword-based search engines that returned ranked lists of links. The optimization that produces AI citation is structurally different. AI systems extract answer capsules, not lists. They weight content depth and originality, not keyword density. They weight authority signals through structured data, not raw inbound link counts. They weight conversational and question-formatted content, not keyword-optimized landing pages.

Most practitioner websites are still built for last-decade SEO. The website service providers serving practitioners haven’t updated their underlying methodology. The practitioner pays for “SEO optimization” that’s optimized for the search environment of 2018, while the actual search environment has shifted toward AI synthesis where the 2018 optimization doesn’t produce results.

What Working AI Search Visibility Actually Looks Like

The practitioner website that gets cited consistently in ChatGPT, Perplexity, Claude, and Google AI Overviews has specific structural properties. None of these are individually difficult — but most providers don’t implement them, and getting them right requires deliberate engineering rather than the standard practitioner website template.

Comprehensive schema architecture. Physician schema with practitioner-type-specific credentialing data marked up structurally. MedicalOrganization schema. MedicalSpecialty schema specifying the actual specialty. LocalBusiness schema with consistent NAP. FAQPage schema marking up clinical Q&A content. Article schema for substantive content pieces. Speakable schema for voice systems. The schema isn’t decorative — it’s the structured signal AI systems extract from to make citation decisions.

Substantive original authority content. Pillar article (typically 2,500+ words) on the practice’s primary specialty. Three to five condition-specific articles (typically 2,000+ words each) targeting the conditions actually treated. Authority page (typically 1,500+ words) establishing credentialing, training, and clinical philosophy. Total: 8,000-12,000+ words of original substantive content owned by the practice. Not licensed library content. Not generic articles distributed across many sites. Original content written specifically for the practice and the patient demographic the practice actually serves.

Practitioner-type-specific credentialing surfaced structurally. NCCAOM Diplomate status for acupuncturists in Person schema with explicit certification fields. IFMCP, A4M, and ABoIM credentials for functional medicine practitioners. AANP membership and accredited school graduation for naturopathic doctors. NBHWC certification for health coaches. SEP three-year training for somatic practitioners. State licensure data with verifiable license numbers. Credentialing surfaced through structured schema rather than buried in About-page plain text.

FAQ-formatted clinical content with answer capsules. The actual clinical questions patients ask AI systems before booking — addressed in FAQ format, with answer capsules that lead with direct declarative answers, marked up with FAQPage schema. The answer capsules are short (typically 50-150 words), declarative, link-free in the capsule itself with supporting links below, and structured for AI extraction. The website that has thirty to fifty clinical FAQs across the site, each with proper schema, becomes citation-rich in ways the website with five operational FAQs cannot.

Speakable schema for voice systems. Specific content sections marked with Speakable schema so AI voice systems can extract them preferentially. Typically the direct-answer summary at the top of authority pages, the answer capsules in FAQ sections, and key practitioner credentials.

Entity authority signals built deliberately. Consistent NAP data across major directories. Listings on professional association sites and specialty organization directories. Co-citation opportunities with established names in the field through guest content, podcast appearances, or co-authored work where appropriate. Long-term: working toward Wikipedia notability for practices with substantial public footprint, though this is a multi-year project and not a starting-point objective.

Content structured for AI extraction rather than 2018 SEO. Question-formatted headings. Answer-first paragraph structure. Clear topic clusters with internal linking architecture. Direct declarative writing rather than keyword-optimized prose. Substantive depth rather than thin keyword-targeted pages.

The DIY Path vs. The Integrated Solution

A practitioner with substantial technical skill and substantial time can implement most of these elements themselves. The schema architecture can be added through manual JSON-LD insertion or through Yoast and similar SEO plugins (with limitations on the practitioner-type-specific schema fields). FAQ content with answer capsules can be written and marked up. Speakable schema can be added to specific sections. Original substantive authority content can be written by the practitioner over time.

The realistic time investment for a practitioner doing this themselves is substantial. Comprehensive schema implementation: 15-25 hours of focused work assuming familiarity with structured data. Original substantive content (8,000-12,000 words at clinical depth): 40-80 hours of writing and editing. FAQ content with proper schema: 10-15 hours. Site architecture refactoring for AI extraction: 10-20 hours. Total: roughly 75-140 hours of practitioner time, spread across weeks or months while the practice continues operating. Most practitioners genuinely don’t have this time available, which is why the work usually doesn’t get done despite practitioners knowing it should.

The integrated solution path replaces practitioner time with focused production. Modern production tools combined with practitioner-type-specific knowledge make it possible to deliver the full infrastructure — schema architecture, substantive authority content, FAQ implementation, Speakable schema, structured credentialing — in a compressed timeline at a fixed price. Modern Practice Websites was built specifically to deliver this integrated solution for serious holistic and integrative practitioners.

What Modern Practice Websites Was Built to Do

The structural decisions that distinguish Modern Practice Websites from typical practitioner website services are concrete and specific.

Comprehensive practitioner-type-specific schema implementation. Every site includes Physician schema with practitioner-type-appropriate credentialing data, MedicalOrganization schema, MedicalSpecialty schema, LocalBusiness schema, FAQPage schema across all clinical FAQ content, Article schema across all authority content, and Speakable schema for AI voice systems. The schema is engineered for the specific modality — NCCAOM credentialing for acupuncturists, IFMCP for functional medicine practitioners, AANP and accredited school authority for naturopathic doctors, ABoIM for integrative MDs, NBHWC for health coaches, and the practitioner-type-specific credentialing for every other modality covered.

10,000 words of substantive original authority content built into every site. One pillar article (~2,500 words) on the practice’s primary specialty. Three condition-specific articles (~2,000 words each) targeting the conditions the practice actually treats. One authority page (~1,500 words) establishing credentialing, training, and clinical philosophy. Written specifically for the practice in the practitioner’s actual voice. Owned by the practice permanently rather than licensed library content distributed across many sites.

FAQ-formatted clinical content with answer capsules across the site. Clinical FAQ sections on every major page, addressing the actual questions patients ask AI systems before booking, formatted with answer-first capsules and marked up with FAQPage schema for citation extraction.

Custom design that supports content depth rather than fighting it. Visual identity matching the depth of the actual clinical work rather than averaging toward generic wellness-template aesthetics. Typography optimized for substantial reading rather than scannable bullet points. Layout that allows content depth without visual clutter.

One-time fee with full ownership. $1,997 one-time investment. The practitioner owns the design, the content, the schema infrastructure, the technical implementation permanently. No recurring fees. No subscription lock-in. The website becomes a permanent asset on the practice’s balance sheet rather than recurring liability.

Ten business days from payment to launch. Approximately ninety minutes of practitioner time required across the entire build. The integrated production model replaces practitioner time with focused expert process.

How to Evaluate Where Your Practice Currently Stands

Before deciding whether to invest in rebuilding the website or attempting DIY implementation, the practitioner should run honest diagnostic tests on the current site.

Test One: AI search visibility. Open ChatGPT, Perplexity, Claude, and Google AI Overviews. Run five to ten queries a real prospective patient might ask about your specialty in your geographic area. Note which practices get cited and how often your practice appears. If your practice rarely or never appears, the diagnosis is straightforward.

Test Two: Schema audit. Use Google’s Rich Results Test or Schema.org’s validator on your homepage and your About page. Note which schema types are present and which are missing. Most practitioner websites have minimal schema — typically basic LocalBusiness schema and not much else.

Test Three: Content depth audit. Count the total word count of substantive content on your site (excluding navigation, footers, and operational pages). Most practitioner sites land at 2,000-6,000 words. The threshold for AI citation typically requires 8,000 words minimum of substantive specialty depth.

Test Four: FAQ structure audit. Count clinical FAQs (not operational FAQs) on your site. Note whether they have FAQPage schema markup. Most practitioner sites have under five clinical FAQs and no proper schema markup.

Test Five: Credentialing structure audit. Look at your About page. Are your credentials listed in plain text only, or are they surfaced through Person schema with structured certification fields? Most practitioner sites list credentials in plain text with no structured data.

Practitioners who run all five tests typically discover the gulf between their current site and what AI search citation requires is larger than they thought. The website that seemed adequate for last-decade search produces minimal results in AI search because the underlying infrastructure was never built for it.

Where to Start

The honest path forward depends on the practitioner’s available time, technical skill, and willingness to invest in rebuilding versus continuing to operate with diminishing returns.

Practitioners with substantial technical skill and 75-140 hours of available time can attempt DIY implementation. The resources for schema implementation, content development, and FAQ structure are available, though scattered across many sources. The work is real and the timeline is long — typically three to six months from starting to having a substantively rebuilt site.

Practitioners who don’t have that time available, who want the integrated infrastructure delivered without piecing together implementation themselves, who want to own the result permanently rather than rent it through subscription, and who want fast delivery should look at Modern Practice Websites. The detailed scope of what’s built, how it’s built, and what it costs is on the main service page. The investment is $1,997 for the website with 10,000 words of authority content built in, comprehensive schema architecture, FAQ implementation, Speakable schema, and full ownership at the end. Or $3,497 for the website plus the complete Practice Operating System covering ad systems, email automation, patient education systems, and the broader marketing architecture.

For modality-specific guidance, the dedicated hubs cover specific dynamics: chiropractor website services, acupuncturist website services, naturopathic doctor website services, functional medicine website services, and holistic and integrative practitioner website services for everyone else.

The structural change in how patients find practitioners has already happened. The patient who used to evaluate ten practitioner websites now reads ChatGPT’s three-practitioner recommendation. The practices that show up in those AI recommendations win patient acquisition that the practices outside those recommendations lose by default. The infrastructure that produces AI citation is specific and technical, and it can be built. Most practitioner websites haven’t built it yet, which means the practitioners who build it now gain visibility advantage that compounds across the next several years as AI search continues displacing traditional patient research.

Frequently Asked Questions