There is no dashboard for AI visibility. No “impressions” chart, no ranking report, nothing to log into. So the only way to find out whether AI assistants recommend your practice is embarrassingly simple: ask them, the way your client would, and read the answer carefully. We do this test for every practitioner we work with, and the answers are consistently more revealing than any SEO audit.
This article shows you the method, what to look for in the answers, and the patterns that keep showing up in who gets named. Run it on your own practice while you read; it takes five minutes.
How to test your own practice: ask the way a client asks
Open ChatGPT, Gemini, Perplexity or Copilot, and type the question your ideal client would actually type. Not “best therapist in Amsterdam”, which nobody says out loud, but something like: “I have been exhausted for months and my doctor says everything is fine. Who could help me with that, ideally someone who works online?” Then follow up the way a real person would: “Any specific practitioners you can name?” and “What would that cost?”
Do this three or four times with different phrasings: with a location, without one, with a specific complaint, with a budget mentioned. Save the answers. Repeat monthly, because these answers change as assistants re-read the web.
What should you look for in the answers?
Who gets named. Usually directories first, then a small number of individual practices. If individuals are named in your field and area and you are not among them, that gap is the whole finding.
What the assistant says about them. This is the useful part. It will summarize a practice in a sentence: who they help, where they work, sometimes what it costs. That sentence is assembled from what those websites state plainly. Compare it to what an assistant could possibly say about your site.
What it hedges on. When an assistant says “you would need to contact them for pricing”, it just told you that practice hid its prices. When it cannot say who a practitioner works with, that practice wrote atmosphere instead of specifics.
Whether it gets you wrong. Ask directly: “What can you tell me about [your practice name]?” Outdated services, a wrong location or a confident description of something you stopped doing three years ago all point to inconsistencies between your site, your Google Business Profile and old directory listings.
Who does ChatGPT name, and what do those practices share?
Across the practitioner fields we work in, the individually-named practices tend to share four unglamorous traits.
| Trait | Why it wins | What it looks like on the page |
|---|---|---|
| A stated specialization | Matches the specific question asked | “Burnout and stress for women 40+”, not “holistic wellbeing” |
| A clear location | Location is in almost every client question | Town in text on every offer page, plus “online worldwide” if true |
| Visible prices | Lets the assistant answer the cost question | “€95 per 75-minute session” |
| Plain answers to obvious questions | Self-contained passages are quotable | An FAQ section on the page itself |
Notice what is not on that list: design, brand, follower count, years of experience, certifications. Assistants cannot see how beautiful your site is, and they cannot feel your reputation. They can only read what you wrote down. The full playbook for writing it down is in our AI SEO playbook for coaches and healers.
Why do directories dominate, and what does that mean for you?
This is the same pattern we found when we looked at what makes clients book: clarity beats polish. Big directories usually appear first, because they contain thousands of structured, consistent, frequently updated entries: exactly what a machine trusts. That sounds discouraging, but read what it actually implies. Assistants name directories when there is no clearer individual candidate, and they name individuals when a site answers the question better than the directory does. A directory entry cannot say what you specifically do, how you work, and what a session with you is like. Your own site can, and that is the opening.
The practical reading of that is not “give up, list yourself and be a line in someone else’s database”. It is that directories and your own site do different jobs, and you want both. A directory gets you into the candidate pool. Your own site is what decides whether the assistant can say something specific about you once you are in it, and specificity is what turns a name in a list into a recommendation with a reason attached.
You can see this in the answers themselves. Practices described in one clause (“also listed: [name]”) almost always have a thin site or none. Practices that get a sentence (“[name], who works with trauma and EMDR and offers evening appointments”) have that sentence somewhere in readable text on their own pages. The assistant is not being creative. It is repeating what it found, and it found more about them.
So the order is: get into the two or three directories your field actually uses, keep the wording there identical to your own site, and then make sure your own pages state the specialization, the location and the practical details in plain sentences. The directory gets you considered. Your site gets you described.
How long does it take before an assistant notices a change?
This is the question every practitioner asks straight after running the test, and none of the guides currently ranking for it give an answer. Here is the honest one: it depends on which of the three routes the assistant used to find you, and they move at very different speeds.
| Route | Typical lag | What you control |
|---|---|---|
| Live web lookup during the answer | Days | Your own pages, immediately after they are indexed |
| Directory entry the assistant trusts | Weeks | Your profile text there, once you update it |
| Training data baked into the model | Months, sometimes longer | Almost nothing, in the short term |
That table explains most of the confusion. A practitioner rewrites her homepage on Monday, asks ChatGPT on Tuesday, sees the old description, and concludes it did not work. What she is often reading is the directory entry she has not touched in two years, or a summary formed before her rebuild existed. The fix is not to write better; it is to update the other sources too, and then to re-run the test in a month rather than the next morning.
One practical consequence: change the slowest-moving thing first. Directory profiles and your Google Business Profile take weeks to filter through, so edit those the same afternoon you rewrite your site, not three months later. And keep the wording identical across all of them, because when the routes disagree the assistant tends to repeat whichever source it trusts most, which is rarely yours.
Turning the test into a to-do list
After running the prompts, you will have a short and unusually concrete list. If you would rather see the machine-readable half measured for you, our free website check scores it in about a minute and hands you the three fixes with the most impact. Typically it looks like this: state the specialization on the homepage in client language, put the town in text on every offer page, publish session prices, add an FAQ that answers cost and suitability, align the Google Business Profile with the site, and remove the two abandoned directory listings that still describe an old address.
None of that is a redesign. It is an afternoon of writing plus one technical pass for schema markup. And every one of those changes also helps the anxious human who is reading your site at eleven at night, which is the honest reason to do it.
Frequently asked questions
How do I check if ChatGPT recommends my practice?
Ask it the question your client would ask, in natural language, with and without your location, then follow up with “can you name specific practitioners?” and “what does that cost?”. Also ask directly what it knows about your practice name. Repeat monthly and note what changes.
Why do assistants give different answers every time?
They generate answers rather than look them up in a fixed index, and their view of the web updates continuously. That is why a single test proves little and a monthly habit proves a lot: you are watching a trend, not a ranking.
What if the assistant says something wrong about my practice?
Treat it as a consistency audit. Wrong information almost always traces back to an old directory listing, an outdated Google Business Profile or a page on your own site you forgot to update. Fix the sources; the answers follow.
Can I pay to be recommended by an AI assistant?
Not inside the recommendation itself. That is precisely why this is worth your attention: it is one of the few visibility channels where a solo practice competes on clarity rather than on budget.
In closing
Ask the machines what they say about you. It costs five minutes, it needs no tools, and it produces a more actionable list than most paid audits: the exact sentences your website fails to make available. Then write those sentences down, plainly, where both a machine and a nervous human can read them.
Want a second pair of eyes on the result? Run our free Website Check for an instant scan and your top 3 fixes, or see how we build AI-readable practices on websites for coaches, therapists and healers.
