AI visibility for plastic surgeons: how ChatGPT, Google AI and Perplexity decide whom to recommend.

A practice can be visible to AI at three levels. It is known when an engine describes it correctly if asked about it by name. It is associated when the engine connects it to a procedure and a place, for example listing it among surgeons who perform deep plane facelifts in Houston. It is recommended when it is named among the few practices a patient is told to consider.

Most practices are known. Far fewer are recommended, and recommendation is the level that leads to consultations. AI visibility work is judged on recommendation: whether you are named, how often, how consistently, and whether what is said about you is true.

Patients increasingly read the answer instead of clicking through. Pew Research Center found that Google users who saw an AI summary clicked a traditional search result in 8% of visits, against 15% when no summary appeared, and clicked a link inside the summary in 1% of visits.

The answer is also short. In DeepContour’s study of 117 Houston practices, each patient question produced a shortlist of two to seven practices across three runs, and 68% of verified practices were never named at all. A practice outside the shortlist does not get a lower position; it gets no mention.

The engines do not publish how they choose which surgeons to recommend, so everything below is observation, not a list of ranking factors.

• Answers lean on third-party sources. In the Houston study RealSelf was cited in all 36 answers, Healthgrades in 24, and Zocdoc and the Houston Chronicle in 17 each. Directories and reviews reached far more answers than any single practice website.
• Answers vary between runs. SparkToro and Gumshoe ran 12 prompts 2,961 times across ChatGPT, Claude and Google’s AI and found less than a 1-in-100 chance of the same list of recommendations appearing twice. In Houston, one in six recommendations appeared in only one of three runs.
• Answers can get facts wrong. A 2026 study in Aesthetic Surgery Journal Open Forum asked four AI tools to name board-certified plastic surgeons across all 50 states; 94.1% of 1,000 results were plastic surgeons, and most errors were otolaryngologists and general surgeons.
• Google says its AI features rest on core Search. Google describes its generative AI features as “rooted in our core Search ranking and quality systems” and says no new machine-readable files, AI text files or markup are needed to appear in AI Overviews or AI Mode.

Ask the questions your patients ask, in a clean session, with your city named, at least ten times each, and record every practice the answer names. Your recommendation rate is the share of runs that name you. Repeat it on each engine your patients use, because ChatGPT, Google AI, Gemini, Perplexity, Claude and Copilot answer differently.

The full step-by-step method, with a template, is in how to track whether ChatGPT recommends your plastic surgery practice.

No one outside the AI companies controls which surgeons an answer names, so getting recommended by ChatGPT, Google AI or Perplexity means improving the evidence answers visibly draw on, in this order, and measuring each change rather than assuming it worked.

1. Make your facts identical everywhere. Surgeon names, practice name, every address and phone number, board certifications, procedures and financing should match across your site and every directory. Conflicting listings give an answer two versions of you to choose from.
2. Complete the profiles answers cite. Start with the platforms that appear in your market’s answers. In Houston those were RealSelf, Healthgrades, Zocdoc and WebMD, plus local press.
3. Give each priority procedure its own page. Answer the questions patients ask about that procedure with you, specifically: who performs it, their training and board certification, what consultation involves, and what it costs where you publish prices.
4. Make the site readable without JavaScript and open to AI crawlers. Allow GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended in robots.txt, and check that your main content is in the page HTML.
5. Don’t rely on a single file. In the Houston study, 31 of the 42 practices that published an llms.txt file were never named.
6. Measure monthly, the same way each time. Use the same questions, engines and run counts so a change in the number reflects a change in the answer.

DeepContour reports six measurements, each answering a question a surgeon would ask:

• Answer Presence. Does AI name your practice at all? (Questions where you are named ÷ questions asked.)
• Share of Answer. When AI recommends surgeons, how many of those recommendations are yours? (Your recommendations ÷ all surgeon and practice recommendations observed.)
• Recommendation Persistence. Does AI recommend you consistently, or did you appear once? (Runs in which you are recommended ÷ total runs.)
• Entity Accuracy. When AI talks about you, are the facts right? (Facts verified ÷ facts checked.)
• Citation Authority. What evidence is AI relying on when it talks about you?
• Procedure Ownership. Which procedures does AI connect to you strongly enough to recommend you?

Definitions, run counts and limitations are on the methodology page.

The two overlap: Google describes its AI features as rooted in core Search ranking and quality systems, so sound SEO remains part of the foundation.

A useful audit for a plastic surgery practice reports, by procedure, market and engine:

• Answer Presence across the questions patients ask
• Share of Answer against the practices named alongside you
• Recommendation Persistence, with the number of runs stated
• Which procedures AI connects to you, and which it gives to competitors
• The sources each answer cites or draws on
• Every fact stated about you that is wrong, missing or inconsistent
• Where competitors are recommended instead of you, and the evidence behind them

When comparing providers, ask:

• Do you repeat each question, and do you report how many runs each number is based on?
• Do you report each engine separately?
• Do you check the facts answers state about my practice?
• Do you report a “rank”? Rankings in AI answers do not survive a second run.
• Do you guarantee recommendations? No one can.
• Do you also represent practices that compete with mine in my market?

DeepContour’s private AI answer snapshot covers each item above. DeepContour does not represent directly competing practices within the same defined market.