Are Patients Finding You Through ChatGPT and AI Search?

It usually shows up as an offhand comment. A new patient is filling out the intake form, and when the front desk asks how they heard about the practice, they shrug and say, "ChatGPT told me about you." Not a friend, not Google, not the ad you are paying for. An AI assistant recommended you by name, and here they are.

The first time it happens it feels like a fluke. The tenth time, it is a pattern, and it raises an uncomfortable question for any practice owner who tries to run on numbers: if patients are finding you through AI, how would you ever know? You open your analytics, look for the source, and find nothing. The visit that started in a chatbot leaves almost no trace. That is not a glitch. It is how this new front door works, and it is worth understanding before it becomes a big share of your new patients.

This is really happening, and the data is not subtle

It is easy to wave this off as a Silicon Valley story that does not touch a dermatology office in Ohio. The numbers say otherwise. In July 2026, Search Engine Land reported on a Previsible study that tracked 6.77 million sessions across 166 analytics properties from late 2024 through May 2026. Traffic coming from AI assistants grew almost ten times over that stretch. And of all that trackable AI referral traffic, ChatGPT alone drove about 92 percent of it.

That is the traffic they could measure. The bigger shift is that AI is no longer just a novelty people play with. A 2026 American Medical Association survey found 72 percent of physicians now use AI in their own practice, up from 48 percent a year earlier. Patients are moving in lockstep. When the tool you use to check drug interactions is the same tool your patient used to decide which office to call, the line between search and recommendation has already moved.

Two launches in early 2026 made it concrete. Perplexity Health arrived in March, connecting a person's own records, wearables, and lab results, then answering health questions and pointing them toward care. Weeks later, OpenAI rolled out a clinician version of ChatGPT for verified US providers. The assistants are getting more capable and more health aware by the month, and patients are the ones typing the questions.

The patient who says "an AI sent me" is not a curiosity anymore. They are the leading edge of how people will pick a doctor, and most practices have no way to count them.

Why your analytics cannot see most of it

Here is the frustrating part. Even when a patient does click through from an AI answer to your website, that visit often lands in the wrong bucket. Some assistants pass along a referrer, so you will see sources like chatgpt.com or perplexity.ai if you go looking. But a large share of AI clicks arrive with no referrer at all, and your analytics files them under direct traffic, the same place it puts people who typed your web address from memory. Your ChatGPT patients get camouflaged as if they already knew you.

Then there is the click that never happens. This is the real blind spot. When a patient asks an assistant for the best knee specialist in their town, the answer often includes your name, a one line summary, and your phone number, right there in the chat. Plenty of patients read that and simply call. No click, no website visit, no session, nothing for any tool to record. The assistant did the referring, your phone rang, and as far as your reports are concerned it came from nowhere. It is the modern version of word of mouth, except the mouth is a machine and you cannot see it moving.

This is why the old habit of judging your marketing purely by what shows up in a dashboard is starting to fail. The dashboard was built for clicks. A growing slice of patient decisions now happen before the click, or instead of it. If you only trust what you can chart, you will conclude AI is doing nothing for you, right up until you realize a fifth of your new patients mention it by name.

The three ways patients now use AI to pick you

It helps to picture the actual moments. There are three that come up again and again.

The direct recommendation. Someone types "best pediatric dentist near me" or "who treats plantar fasciitis in Tampa" into ChatGPT and gets a short list with reasons. Whether your name is on that list depends on the same signals that shape Google rankings, which is why some practices show up and others get skipped. We dug into that gap in why ChatGPT keeps recommending hospitals instead of your practice, and it is fixable.

The AI answer inside Google. Many patients never leave the search page. Google's AI generated answers summarize and cite sources right at the top, and a patient may read enough there to make a choice without clicking a single blue link. We covered how much patients actually trust these in whether patients trust Google AI overviews.

The informed second opinion. With tools like Perplexity Health connecting a patient's own records and labs, people arrive at the decision already briefed. They are not asking whether they need care, they are asking who to see for it, and they want a provider whose online presence matches the confidence the assistant gave them. If your website looks thin next to that, the trust breaks.

How to actually tell if AI is sending you patients

You cannot manage what you cannot see, so the goal is to make the invisible a little more visible. You do not need fancy tools. You need a couple of habits.

• Ask every new patient, and write it down. The single most reliable AI tracker in your practice is the question "how did you hear about us?" on your intake form, your booking form, and your phone script. When the answers start including ChatGPT, an AI, or I asked Perplexity, you have real data no analytics tool can give you. Tally it monthly. This one habit is the backbone of knowing where your patients actually come from.
• Check your referral sources by name. In your website analytics, look at referrals for chatgpt.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com. The numbers will be modest, but the trend line matters more than the total. If it is climbing, so is your exposure.
• Watch for branded search bumps. A subtler tell: people hear your name from an AI, then Google you to double check before calling. A rise in searches for your practice name, with no ad campaign behind it, often means something upstream is recommending you.
• Use call tracking, carefully. A tracked number and a simple front desk log can catch the calls that skip your website entirely. Just keep patient privacy in mind when you record or store anything, the same care you would apply to any tool that touches patient data, which we get into in whether Google Analytics is HIPAA compliant for a medical practice.

None of this is perfect. It is not supposed to be. The point is to move from "no idea" to "a rough, honest read," so you can tell whether AI is becoming a real channel for you and act before your competitors do.

What to do so the assistants keep saying your name

Now the good news, and it is genuinely good. You do not need a whole new playbook to earn AI recommendations. The assistants are not inventing their answers, they are pulling from the open web, from reviews, from Google Business Profiles, and from pages that explain things clearly. The work that makes Google trust you is the same work that makes a model quote you. If your local SEO is strong, you are already most of the way there.

A few things move the needle most. Write pages that answer the exact questions patients ask, in plain language, because a model can lift a clear paragraph and cite it far more easily than it can parse marketing fluff. Keep your Google Business Profile complete, current, and consistent with your website, since that is one of the first places assistants check. Keep a steady flow of genuine reviews, because they shape both rankings and the summary an AI writes about you. And make sure the basic facts about your practice, your services, your location, your hours, match everywhere they appear online. This whole discipline has a name now, and we laid it out in our guide to SEO and generative engine optimization for healthcare in 2026, along with a practical look at whether your site even needs an llms.txt file to guide the AI crawlers.

Getting recommended is only half of it. The other half is not fumbling the handoff. When an assistant sends a patient to your site or your phone, that last step has to work. A booking page that loads slowly or confuses people wastes the referral you just earned, which is why a website built to convert and rank matters more, not less, in an AI world. And when the AI hands over your phone number instead of a link, a real answer has to be waiting. A missed call at 8pm from a patient an assistant just recommended is a patient gone. This is exactly the gap our AI receptionist is built to close, answering every call and booking around the clock so nobody the machines send you slips away.

Where EtherealMinds fits

We work only with healthcare practices in the United States, and this is one of the shifts we spend the most time helping owners get ahead of. Not because it is flashy, but because it is early, and early is cheap. The practices that build strong, quotable, trustworthy web presences now will be the default answer when a patient asks an assistant for a recommendation, long before their competitors even notice the channel exists.

In practice that means a few things working together: healthcare SEO tuned so both Google and the AI assistants find and cite you, pages written to answer real patient questions clearly, a Google Business Profile kept sharp, and a patient acquisition system with tracking and an AI receptionist behind it, so the patients these tools send you actually get caught and counted. We also make sure you have a straight answer to the question this article started with, by building the habit of asking every patient how they found you, so the invisible channel stops being invisible. If you want to be recommended by the assistants directly, we go deeper on that in how to get your practice recommended by ChatGPT.

The honest takeaway

AI is not going to replace how patients find doctors overnight, and anyone telling you it already has is selling something. But it is a real, growing channel, it is nearly invisible in the reports you rely on, and it rewards the practices that are genuinely trustworthy and clearly explained online. That is the reassuring part. There is no trick here, no gaming it. The same things that make you a good, findable, well reviewed practice for humans make you the answer a machine gives too. Start by asking every patient how they found you, keep your web presence strong and honest, and make sure the phone gets answered when the assistant sends someone your way. Do that, and the next time a patient says "ChatGPT sent me," it will not be a surprise. It will be the plan working.