Caseload Studio

How people find a therapist now

The search box changed. What people are looking for did not. A practical look at the new first step in finding care, and what it means for your practice.

A woman on a sofa looking at her phone in the evening

For most of the last fifteen years, finding a therapist looked roughly the same. Someone asked a friend or their doctor, then went to a directory or typed a search into Google, then read three or four websites late at night and picked the one that felt right.

That path still exists. But there is now a step in front of it, and most practices have not noticed yet.

The new first step

People are asking AI assistants about their health, and the numbers moved fast. In Rock Health's 2025 consumer survey of 8,000 US adults, one in three said they had used an AI chatbot for health information, double the share a year earlier. Among adults under 45 the figure was closer to half. A Gallup survey of more than 5,500 adults found a quarter had used an AI tool for health information or advice, and that most recent users did so to research on their own before seeing a provider.

Finding the provider is part of it. One industry survey reported by TechTarget found 47% of patients had used AI to look for a new provider, up from 31% nine months earlier, and that AI now influenced the final choice slightly more often than Google results did. The most common request was simply a specialty plus "near me."

None of these surveys broke out therapy on its own. There is no reason to expect therapy to be the exception, though. A parent worried about a fourteen-year-old who has stopped going to school is exactly the person who types a long, specific, slightly desperate question into a chat box at eleven at night.

What the answer looks like now

Here is the part that changes things for a practice. A search engine used to hand back ten links and let the person decide. An AI assistant reads those pages first and hands back a short answer: a few names, a sentence about each, maybe a reason one fits better than another.

Google's own results now work the same way for many questions. When Seer Interactive tracked informational searches that show an AI Overview, the click-through rate to regular results fell from 1.76% to 0.61% between mid-2024 and late 2025. Ahrefs found a similar pattern across 300,000 keywords. People still search. They just click less, because the summary already answered them.

So the question is no longer only "do I rank on page one?" It is also "when something reads the web and summarizes my field for a worried parent, am I in the summary?"

The answer is only as good as what it reads

AI assistants do not know your practice. They know what the web says about it: your website, your Google Business Profile, your directory listings, articles that mention you. When those sources disagree, the answer gets confused.

That is not a theoretical risk. In the same industry survey, two thirds of people who had used AI to research a provider said they had run into incorrect information: the wrong address, an old phone number, outdated insurance details, office hours that were no longer true. Details like these drift out of sync across listings over the years, and an AI has no way to know which version is current.

If your own details are inconsistent across the web, an AI will repeat the inconsistency to the people looking for you.

The old signals still count

None of this means referrals stopped mattering. In that same survey, family and friends were still among the strongest influences on the final choice, alongside insurance websites and online reviews, and low ratings still filtered people out. What changed is the order of events. A parent might hear your name from a pediatrician, ask an AI what it knows about you, read the summary, and only then visit your site. Every one of those stops is a place to lose them or to reassure them.

What this means for your practice

  • Make your basics identical everywhere. Name, address, phone, credentials, specialties, whether you take insurance. Same wording on your site, your Google profile and every directory.
  • Say plainly what you do and where. A model cannot recommend you for teen anxiety in Pasadena if your site never says "teen," "anxiety" or "Pasadena."
  • Write things worth quoting. Specific answers to real questions, in your own voice, under your own name. Generic copy gives a model nothing to cite.
  • Keep the human path clear. When someone does arrive, they need fees, availability and a way to book within a few seconds.

The search box changed. What people are looking for did not. They want someone they can trust with something that matters. The practices that do well in the next few years will be the ones that make that trust easy to see, for people and for the software those people now ask first.

Curious how your practice shows up today?

The free audit checks your site the way a search engine and an AI assistant read it, and a person reviews every report before it goes out.