R. Antonio Matta

R. Antonio Matta

ChatGPT SEO for Therapists: What 18 Practice Audits Show

Maybe you’ve typed your specialty and your city into ChatGPT between sessions, half hoping to see your practice’s name in the answer. It wasn’t there.

We asked five AI engines to recommend care in 18 different markets, and a therapist directory showed up in 16 of them (Content Done Write 2026). That pattern sits at the heart of ChatGPT SEO now, at a time when about one in six adults has asked AI for mental health information in the past year (KFF 2026). Some practices did get named first, and we’ll show you what they had in common.

Key takeaways

  • AI answers cited Psychology Today in 16 of our 18 audits, and five organizations were never cited for their own core services.
  • None of the 18 sites blocked AI crawlers. The gaps were accuracy, clarity and presence on other trusted sites.
  • Practices that were named first won specific searches: a language, a therapy method, a population or a small town.
  • Start with accurate practice details and one specific service page, then earn real mentions where AI already looks.

ChatGPT SEO is the work of making a practice’s website and listings easy for AI engines to read, trust and cite. In November 2026, we tested 18 behavioral health organizations across ChatGPT, Gemini, Claude, Perplexity and Grok, and Psychology Today appeared in answers for 16 of them (Content Done Write 2026). Five organizations were never cited for their own core searches, and none of the 18 sites blocked AI crawlers. The practices AI named first won on specific searches, such as a language, a therapy method or a small town, which fits research showing AI search leans toward trusted third-party sources (Chen et al. 2025).

When I ask ChatGPT for a therapist in my city, who does it name?

Most of the time, the answer opens with a directory, which makes local SEO for a practice feel like a different game than it was a few years ago. Psychology Today showed up in 16 of our 18 audits, and it was the very first source an engine listed in 13 of them (Content Done Write 2026). The rest of the cast was familiar too: Zencare, GoodTherapy, national telehealth apps like BetterHelp and Talkspace, specialty associations such as the International OCD Foundation and Postpartum Support International, hospital systems and state or county health pages.

If you’ve watched the intake phone stay quiet on a Tuesday and wondered where the callers went, that number can sting. What it shows is how these search engines build an answer, and once you can see that picture, you can start to work with it.

Government health pages carried real weight in our results. For one Maryland clinic, a county website was the first source in all five engines, and that lines up with national data showing .gov sites appear about three times as often in Google’s AI summaries as in regular results (Pew Research Center 2025). AI search also leans hard toward trusted third-party sources over a business’s own website (Chen et al. 2025).

That’s why AI search visibility is its own question now, separate from where you rank on Google. The answer someone reads at midnight is built from many sources, and your site is only one of them.

How were these 18 practices tested across five AI engines?

We audited 18 behavioral health organizations in November 2026: 11 group practices and private practices, 5 clinics and programs (including an intensive outpatient program), and 2 digital mental health platforms, spread across 12 states and two national markets (Content Done Write 2026). For each one, we ran three local searches a real client might type, about 60 searches in all, across ChatGPT, Gemini, Claude, Perplexity and Grok, with a few Google AI Overviews spot checks.

We also looked under the hood of each website: whether AI crawlers were allowed in, whether the site had an llms.txt file and a sitemap, and whether the practice’s details were written in a form machines can read. Every organization stays anonymous here.

A few honest limits belong on the table. Each search ran once, answers shift by day and location, and in five audits Claude returned answers with no sources at all, which we report as a limit of this run (Content Done Write 2026). Different AI systems also cite very different sources, so a single run is a snapshot of a moving picture (Yang 2025).

I built a good website. Why am I still invisible to the people looking for me?

You did the work. You paid for the site, wrote the about page, maybe kept a blog going for years, and still ChatGPT named someone else. Five of the 18 organizations in our audits were never cited for their own core services, even with working websites and real clinical depth (Content Done Write 2026).

That quiet absence matters because of who’s asking. About half of adults living with a mental health condition in one U.S. survey had turned to an AI chatbot for support, and nearly all of them had used ChatGPT (Rousmaniere et al. 2026). Across all adults, about one in six used AI for mental health information last year, and the share is close to three in ten among people under 30 (KFF 2026).

The way people click has changed too. When an AI summary sits at the top of a Google page, people click through to a regular result about half as often, and they almost never click the links inside the summary (Pew Research Center 2025). Being named inside the answer now carries more weight than ranking just below it, and that shift has a real impact on SEO for every practice that depends on new clients finding them.

Think about the person on the other end. Prospective clients already search therapists online, and one detail they find, or can’t find, can decide whether they call (Pomerantz & Dever 2021). Good SEO strategies and steady SEO practices still matter. Your effort was real all along. The place where people now ask the question simply reads your work through a different lens, and that lens can be learned.

Which site problems kept showing up in the ChatGPT SEO audits?

Here’s the good news first: none of the 18 sites blocked AI crawlers in their robots.txt file (Content Done Write 2026). Getting crawled is the first gate, since content has to be crawlable and indexed before any AI feature can use it (Google Search Central 2026). The gaps we found were quieter, the kind of technical SEO tasks that never make it onto a busy clinical director’s list.

Your details lived only in sentences. In four audits, the address and services appeared only as page text, with nothing in the hidden labels search engines read. Three Squarespace sites shipped a built-in business block with the address and hours left blank, and others mislabeled the practice, including one tagged as a health club (Content Done Write 2026).

Leftovers were still public. Six sites still had test posts, duplicate drafts, unused theme pages or builder setup text live where search engines could find them. One site sent AI crawlers an empty page that needed JavaScript to load, so every page read like the homepage.

Descriptions were missing. Several sites had no meta description, the short summary under a search result, on a third or more of their pages, including service and team pages.

Specific services had no home. Child therapy, a second office or telehealth in another state often had no page of its own, so there was nothing specific for an engine to quote.

A blank hours field on a clinic page tells a machine something false, and it tells a worried parent nothing at all. These are good SEO best practices for people and for engines, since good content and SEO have always moved together. AI search optimization builds on that same foundation, since content people find useful, written from first-hand experience, is what these systems reward over time (Google Search Central 2026).

Do I really need llms.txt and schema markup, or is that hype?

You’ve probably heard both pitched as must-haves, and parts of the SEO community still sell them that way. In our audits, 14 of 18 sites had no llms.txt file and 14 had no FAQ markup, and those numbers can make you feel behind (Content Done Write 2026).

The guidance from Google is clear on two points. Google Search doesn’t use llms.txt, and structured data isn’t required for Google’s AI features, though Google still recommends it for richer search results (Google Search Central 2026). Other engines haven’t published anything similar, so for SEO purposes llms.txt stays a low-cost, low-priority extra.

Schema matters most when it’s wrong. A blank address block or a “health club” label tells every machine something untrue about you, so fix accuracy first and add new markup later. Engines also differ in what they cite, which is one more reason to fix the basics that every engine can read (Yang 2025).

Psychology Today keeps getting named first. Can a small practice still earn LLM visibility?

Yes, and this is where the practices that got named first come in. Every one of the eight organizations that an engine named first won on a specific search (Content Done Write 2026). Think of a Mandarin-speaking therapist in New York, a culturally responsive therapist in Detroit, a postpartum day program, exposure and response prevention in Boston, a psychiatric nurse practitioner in a small New England city, a therapist in a Columbus suburb, an outpatient clinic in a Baltimore County town and online OCD therapy.

The broad searches, like “therapist in [city]” or “anxiety therapist in [city],” went to the directories almost every time. That fits what researchers have found about AI search: it favors trusted third-party sources and well-known brands, so niche providers have to work around that bias on purpose (Chen et al. 2025).

Picture the referral list at your front desk. When a parent asks for help with a specific worry, the name someone circles is the provider who fits that worry, and AI answers seem to work the same way. A complete, accurate directory profile is part of your footprint, and so are association listings, hospital and county referral pages and local press.

The writing itself matters too, because strong SEO content is what an engine can quote. In controlled tests, pages that cited their sources, included quotes and added real statistics were up to 40% more visible in AI-generated answers (Aggarwal et al. 2024). That’s where content marketing and SEO meet, since a page written in a clinician’s real voice, with real evidence, gives an engine something worth quoting and gives a reader something worth trusting.

Why would Gemini name my practice when ChatGPT won’t?

Each engine behaved like a different referral source. Gemini cited the organization’s own site in 10 of our 18 audits, Claude in 8, ChatGPT and Grok in 2 each, and Perplexity in none (Content Done Write 2026). ChatGPT often built its answers on state health pages and national associations.

Research on hundreds of thousands of AI citations found the same thing: different providers cite different sources, and citations bunch up around a small number of outlets (Yang 2025). So a miss in ChatGPT doesn’t mean you’re missing everywhere, and a win in Gemini doesn’t mean the work is done.

Test each engine on its own, and test again later, since answers shift from one run to the next (Content Done Write 2026).

Where do I start with SEO for ChatGPT this month?

  1. Run your own test. Ask ChatGPT, Gemini, Claude, Perplexity and Grok the three questions your best-fit clients would type, and write down who appears (Content Done Write 2026).
  2. Make your details match everywhere. Use the same practice name, address, phone, services and clinicians on your site, Google Business Profile and directory listings, since Business Profiles can help services show up in AI answers (Google Search Central 2026).
  3. Clear the leftovers. Delete or hide test, duplicate and theme pages, and add meta descriptions to your service and team pages.
  4. Correct the hidden labels. Fill in or remove empty business blocks, choose an accurate practice type and make sure the labels match what’s visible on the page.
  5. Build one specific page. Write about the service, population, language or location you serve best, from first-hand clinical experience, with sources and real numbers (Aggarwal et al. 2024).
  6. Earn real mentions where AI already reads. Complete your directory and association profiles and ask to be added to hospital and county referral lists, because AI search leans on trusted third-party sources (Chen et al. 2025).
  7. Retest every quarter. Keep your notes in one place to track your SEO performance, and your log becomes your own SEO audit over time.

How can I use ChatGPT for SEO without losing my clinical voice?

ChatGPT SEO has a second meaning, and it deserves a plain answer too: using ChatGPT for SEO work itself. ChatGPT is a powerful tool for the busy parts of SEO tasks, and for a small practice with no marketing team, the use case is real. Here are practical ways to use ChatGPT for SEO in 2026, while the clinical judgment stays with you.

  • Research. Use ChatGPT as a research tool for SEO keyword research and content ideas. Ask ChatGPT to generate the questions families ask about a postpartum program, and keep the ones you hear in session.
  • Drafting. Using ChatGPT to create content outlines saves real time. You can use ChatGPT to create an outline, use ChatGPT to write a rough first pass, or ask ChatGPT to write three versions of a meta description. ChatGPT can also turn a dense paragraph into plain language when you ask ChatGPT to rewrite it.
  • Reviewing. Use ChatGPT to summarize your own service page, so you can see whether the main point comes through. ChatGPT helps you spot jargon a worried reader might trip over.
  • Technical SEO tasks. ChatGPT for technical SEO works best on checklists. ChatGPT can assist with those lists, explain what a schema field means and help you use ChatGPT to optimize page titles. Have your web person confirm anything that touches code.

Content generated by ChatGPT can sound polished and still miss what makes your practice yours, while AI search tends to reward first-hand experience and real sources (Aggarwal et al. 2024). Rely on ChatGPT for speed, and let SEO specialists, a human SEO expert or a clinician on your team decide what’s accurate before anything goes live. ChatGPT can generate options and ChatGPT can create drafts quickly, and a person still has to confirm what’s true, since first-hand content is what search systems look for (Google Search Central 2026).

We recommend using ChatGPT, and AI tools like ChatGPT, as an assistant inside your SEO processes. Practices that leverage ChatGPT for SEO well treat it like a capable intern: integrating ChatGPT where it saves time, and implementing ChatGPT drafts only after review. You don’t need a library of SEO prompts or a stack of SEO tools to start, and no AI tool replaces the voice your clients came to trust.

One boundary belongs here. No one can promise that an AI engine will recommend your practice, and we won’t promise it either. In a clinical field, that also means no fake reviews, no testimonials that strain confidentiality and no claims about outcomes you can’t stand behind.

Being findable is part of how I care for the people I serve

The care you give is already real. Making it easy to find is one more way of showing up for the person searching late at night with no one else to ask. Most gaps in our audits were small and common, and our Snapshot uses only your public pages, never client data.

What else am I still wondering about ChatGPT SEO?

What is ChatGPT SEO?

ChatGPT SEO is the work of helping AI engines like ChatGPT find, understand and cite your practice when someone asks for help. It overlaps with generative engine optimization, answer engine optimization and LLM SEO, a field built on research into how content earns visibility in AI answers (Aggarwal et al. 2024). From Google’s side, optimizing for AI search is still search engine optimization, built on content people find useful (Google Search Central 2026).

Can I just use ChatGPT to do my SEO for me?

ChatGPT can help with SEO tasks like keyword research, content ideas, a first outline or a rewrite of a clunky paragraph. Writing content that earns citations takes more, and it’s where your SEO efforts will either improve your SEO or blur your voice, because AI search rewards first-hand experience and clear sourcing (Google Search Central 2026). Use AI tools like ChatGPT as an assistant for AI SEO drafts, and keep a human SEO expert or clinician deciding what’s true and what goes live (Aggarwal et al. 2024).

How can I check whether ChatGPT recommends my practice?

Ask ChatGPT, Gemini, Claude, Perplexity and Grok the three questions your best-fit clients would type, using your city and specialty. Note who appears and which sources each engine cites, then repeat every quarter (Content Done Write 2026). Since engines cite different sources, check each one separately (Yang 2025).

Does a Psychology Today profile help my AI visibility?

It appeared in 16 of our 18 audits, so a complete, accurate profile is part of the footprint AI reads (Content Done Write 2026). It works alongside your own site, and AI search leans on trusted third-party sources like it (Chen et al. 2025).

Is llms.txt worth adding?

Google Search doesn’t use llms.txt, and adding one won’t change your visibility there (Google Search Central 2026). Other engines haven’t said how they treat it. Treat it as a low-priority extra after your accuracy fixes, since 14 of the 18 sites we audited had none and some of them still got named (Content Done Write 2026).

How long does it take for AI answers to change?

There’s no fixed timeline, and getting indexed and shown is never guaranteed (Google Search Central 2026). Answers also shift between runs, so retest every quarter and compare your notes over time (Content Done Write 2026).

References

  1. Content Done Write. (2026). AI visibility audits of 18 behavioral health organizations [Unpublished raw data].
  2. KFF. (2026, March 26). New KFF tracking poll on health information and trust finds one in three adults have used AI chatbots for health advice.
  3. Chen, M., Wang, X., Chen, K., & Koudas, N. (2025). Generative engine optimization: How to dominate AI search [Preprint]. arXiv.
  4. Pew Research Center. (2025, July 22). Google users are less likely to click on links when an AI summary appears in the results.
  5. Yang, K.-C. (2025). News source citing patterns in AI search systems [Preprint]. arXiv.
  6. Rousmaniere, T., Zhang, Y., Li, X., & Shah, S. (2026). Large language models as mental health resources: Patterns of use in the United States. Practice Innovations, 11(2), 139–155.
  7. Pomerantz, A. M., & Dever, J. M. (2021). Googling psychotherapists: Can online information make or break prospective clients’ decisions? Practice Innovations, 6(3), 149–158.
  8. Google Search Central. (2026). Optimizing your website for generative AI features on Google Search. Google for Developers.
  9. Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K., & Deshpande, A. (2024). GEO: Generative engine optimization. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (pp. 5–16).