Patients used to find a dental practice by searching. They typed a query, scanned the map pack, opened three websites in browser tabs, and picked one. That path is still real, but a growing share of patients never take it anymore. They ask an AI model a question and take the model’s recommendation.
The change is not that a new platform has appeared alongside Google. The change is that the first step of the patient journey, the discovery step, has moved into a conversation with a language model. The consequence for practices is that the discovery step is now happening in an environment most practices are not paying any attention to.
Where the Discovery Step Actually Lives Now
A patient with a specific dental concern used to type “cracked molar dentist near me” into Google. That same patient in 2026 is often typing something closer to a full sentence into ChatGPT, Perplexity, or Google’s AI Mode: “I have a cracked back tooth, I need someone who takes Cigna and can see me this week, who is good in the area.” The AI returns a short list of practices with a paragraph on each. The patient picks one, opens the practice’s website in a new tab, and books.
The practice on the receiving end sees a direct visit to the site. There is no keyword to analyze in Google Search Console, because the query never touched Google in a way the console tracks. The traffic looks like a name-typed URL or a click from a source the analytics tool cannot label cleanly. The practice’s marketing dashboard shows a booking with no attributable origin.
What This Means for Attribution
Traditional attribution frameworks assume patients enter through a channel the practice can identify. Paid search shows up as paid search. Organic search shows up in Search Console with a query and a landing page. Direct traffic used to be small and often meant a returning patient. In 2026, direct traffic is growing, and a portion of it is patients who arrived from an AI recommendation the practice cannot see.
This is a familiar pattern in the DIGI Search archive. NextGen TV™ produces the same attribution problem: patients absorb a TV ad and later arrive through organic search or by typing the practice name directly. The Invisible Engine case study documented how a Naperville practice generated over $213,000 in first-year revenue through organic channels fed by TV air cover that showed up nowhere in a last-click model. AI recommendation looks like a second, parallel version of that same effect, operating at the moment of decision rather than at the moment of awareness.
For practices, the practical implication is that measurement needs to include aggregate signals: total booking rate from direct traffic, quality of patients arriving without a labeled source, and answers to “how did you hear about us” at intake. A practice tracking only paid channel ROI is now missing two important sources of new patients rather than one.
What Patients Are Doing Differently
The behavior of a patient arriving from AI recommendation differs from the behavior of a patient arriving from a Google search result in three visible ways.
First, they arrive further along in the decision. The AI recommendation did the pre-selection work that used to happen when a patient opened three tabs and compared. Patients are not comparing five practices before they call. They are calling the practice the model named, and only opening a comparison tab if something on the website stops them.
Second, they book faster. The gap between site visit and appointment scheduled shrinks because the patient is closer to a decision when they land. The site’s job is now to confirm what the AI already told them, not to convince them.
Third, they arrive with better questions. Patients who read a paragraph about the practice inside an AI answer often already know the general specialty focus, whether the practice accepts new patients, and often something about the provider. The intake call skips the introductory step and moves directly to scheduling.
The Reciprocal Requirement
For a practice to be the one the AI names, the practice has to look like a recognized entity to a language model. That means clean structured data, consistent business information across sources, substantive content that answers the questions patients ask in real language, and citations in the sources AI models draw from. GEO for Dental SEO covers the entity work in more depth.
The practices doing this work well in 2026 are the same practices that took SEO seriously in 2018 and reviews seriously in 2022. The specifics of the discipline have changed. The underlying pattern, investing in visibility infrastructure before it becomes obvious that it was necessary, has not.
The Question Worth Asking
The honest question for a practice owner in 2026 is not “how do we rank in AI search.” That question is already the wrong question. The question is: when a patient in this practice’s area asks a language model for a dentist, who does the model name, and if it is not this practice, what would it take to become one of the names?
Bring that question to a discovery conversation with the DIGI Search team.
