A person searching for a therapist now often types a question into ChatGPT, Gemini, or Perplexity instead of scrolling a directory. These engines answer by pulling from your website, your listings on directories like Psychology Today, and review platforms, then naming a small handful of practices that match the request. If your practice's specialties, location, and insurance details are not clearly stated online, the engine has nothing to name.
The path from a private question to a named practice
Someone looking for help rarely searches a business name first. That means the practice that gets named is the one whose information the engine could find, understand, and trust enough to repeat.
This is a shift from ranking to being chosen. A traditional search engine result page shows many options and lets the person compare. An AI answer compresses that comparison into a handful of named practices, sometimes just one. If a practice's specialties, location, and logistics are not stated in plain language somewhere the engine can access, that practice is skipped over, not ranked lower.
The kinds of prompts clients use for each engine
Prospective clients phrase their requests differently depending on which tool they open, but most prompts combine a concern, a location, and a practical filter such as insurance or session format. Understanding the shape of these prompts helps a practice see which facts actually need to be answered on its own site rather than left implied.
On ChatGPT, prompts tend to read like a conversation: "I'm looking for a therapist in your city who works with teens and takes Blue Cross." On Gemini, which is often used from a phone alongside Google, prompts skew more local: "counselor near me for couples therapy." On Perplexity, which shows its sources more visibly, users often ask comparison-style questions like "what's the difference between a psychologist and a licensed counselor for grief support," then follow up with a location. In every case, the engine is trying to match a concern and a constraint to a real, findable provider.
What sources each engine pulls from to build an answer
Each AI engine builds its answer from a mix of your own website, professional directories, review sites, and general web content, weighted differently by tool. ChatGPT and Perplexity tend to cite named sources directly in their responses, while Gemini often blends web results with Google's local business data. A practice that is inconsistent across these sources gives the engine conflicting signals and a reason to leave it out.
Your website is the source you fully control, so it carries the most weight when it is specific and current. Directory profiles matter next, because they are frequently updated and heavily indexed by these tools. Review platforms contribute context on what clients experienced, which engines sometimes summarize when asked "what do people say about" a practice. If your practice name, address, phone number, and specialties do not match across all of these, the engine has to guess which version is correct, and it may simply choose a competitor with cleaner information instead.
Where your specialties and insurance information need to live
The details clients filter by first, such as issues treated, age groups served, session format, and accepted insurance, need to appear in plain text on your own website, not only in a downloadable form or an intake questionnaire. AI engines read visible page text; they cannot open a PDF behind a "new client" button or infer coverage from a logo image. If a fact only exists in a document a visitor has to download, it will not surface in an answer.
A page or clearly labeled section describing who you see, in what formats, and which insurance plans you accept gives the engine language to quote directly. Write it the way a client would ask about it: name specific concerns you work with, such as anxiety, relationship issues, or life transitions, using the same everyday words a person would type into a search box. Listing insurance panels by name, and noting whether you offer sliding-scale or self-pay options, gives engines a concrete detail to repeat when someone asks about cost.
How to test whether you appear right now
The only way to know how a practice currently shows up is to ask the engines directly, using the same phrasing a prospective client would use, and to review the exact information behind whatever answer you get named in. This kind of testing takes a few minutes and tells you exactly where the gaps are, without guesswork.
Open ChatGPT, Gemini, and Perplexity separately and ask each one a version of "who is a good therapist in your city for your a specialty you offer." Note whether your practice appears, and if it does not, look at what information the engine did use for the practices it named. Then ask a narrower question involving insurance or a specific concern you treat, since general visibility and specific-match visibility are not the same thing. If your practice is missing or the details cited about it are outdated, that points directly to what needs to be corrected on your site and directory listings, not to a problem with the engines themselves.
What to ask a marketer before they touch your visibility
Before hiring anyone to work on how your practice appears in AI search, ask them to show you, in an actual chat window, what ChatGPT or Perplexity currently says about a practice like yours and why. Ask what specific facts on your site they would change first and how they would verify those facts are consistent across your directory listings. Ask how they would describe your services in a way that names concerns and populations served without implying a guarantee of outcome. If they cannot answer with specifics tied to your practice, or lean on vague promises about rankings and traffic, that is a sign they are guessing rather than testing.