Each AI engine builds a plastic surgeon recommendation from a different mix of sources. Temso's 2026 study of healthcare provider prompts found the top-20 cited domains of any two engines overlap by only 13 percent on average, so a surgeon named by ChatGPT can be missing from Gemini on the same rhinoplasty question.
AEO, GEO, AI visibility, and AI SEO are four names for the same work: getting a surgeon named inside AI answers. This post uses AEO. The ClinicAds post on sources AI assistants cite for plastic surgeons sorts sources by type, and this one sorts by engine without repeating that table. It states plainly that no engine-level study of plastic surgery exists, so the figures below come from adjacent datasets.
- The engines do not share one source list: Temso's July to September 2026 study of 15,769 online healthcare provider responses found an average top-20 domain overlap of 13 percent between engine pairs, and 0 percent between Gemini and Microsoft Copilot.
- Yext's analysis of about 6.8 million citations found Gemini sent 52.2 percent of its citations to first-party websites, while listings supplied 48.7 percent of ChatGPT's, so a surgeon's own site and directory profiles carry different weight by engine.
- No independent study found for this post splits citations by engine for plastic surgeons specifically, so every figure here is cross-industry or cross-healthcare and is labelled that way.
- For a surgeon the surfaces that matter in every engine are board and society lookups, RealSelf, hospital privileging pages, and the practice's own procedure pages, which a per-engine log shows in order of repetition.
Do ChatGPT, Perplexity, Gemini, and Google AI Overviews recommend the same plastic surgeons?
No. The engines retrieve from overlapping but different pools, so one procedure prompt can name different surgeons in different engines. Temso measured the overlap between each pair of five engines on healthcare provider prompts and found it averaged 13 percent, from 43 percent between Google AI Mode and Google AI Overviews down to 0 percent between Gemini and Microsoft Copilot.
For a surgeon, this means one audit of one engine proves little. A practice with strong procedure pages and few third-party listings can appear in one engine and vanish in another. The table below shows the public evidence per engine, what kind of dataset it came from, and the surgeon-specific surface to fix first.
| Engine | What the data shows | Dataset | First fix for a surgeon |
|---|---|---|---|
| ChatGPT | Listings supplied 48.7 percent of citations; 46 percent of healthcare provider sources were commercial and 20 percent institutional | Yext, about 6.8M citations, cross-industry; Temso, 2026, healthcare | Match the surgeon's name, practice address, and phone across the board lookup, society directory, RealSelf, and Google profile |
| Gemini | 52.2 percent of citations went to first-party websites; 72 percent of healthcare provider sources were commercial | Yext, same sample; Temso, same study | One page per procedure naming the surgeon, board certification, technique, and recovery in plain sentences |
| Perplexity | Counted inside Yext's 86 percent brand-managed finding; no separate share published | Yext, same sample; Temso did not cover it | Log the numbered sources under each answer and fix the ones that repeat |
| Google AI Overviews | Not split out by Yext; shares an index with Google organic results | ClinicAds observation, not study data | Keep the Business Profile and procedure pages indexable and current |
Which sources does ChatGPT lean on when it names a plastic surgeon?
ChatGPT leans on listings and mixed commercial and institutional pages. Yext found listings supplied 48.7 percent of ChatGPT's citations in its cross-industry sample. Temso found 46 percent of ChatGPT's healthcare provider sources were commercial and 20 percent institutional, the highest institutional share among the engines where it reported one.
For a surgeon, institutional sources include hospital pages and board or society lookups, and commercial ones include RealSelf and local editorial lists. A practice absent from those pages gives ChatGPT nothing independent to corroborate its own site.
- Confirm the surgeon is listed with current credentials on the certifying board and society directories
- Complete the hospital or surgery center privileging page that names the surgeon
- Keep the RealSelf profile current, with procedures and photo counts that match the practice site
Which sources does Gemini use for plastic surgeon recommendations?
Gemini favors first-party websites. Yext reported 52.2 percent of Gemini's citations went to a business's own site, the highest of the three engines it studied. Temso also found 72 percent of Gemini's healthcare provider sources were commercial, which points to provider pages and platforms rather than health authorities.
That makes the procedure page the most valuable owned asset for Gemini. A page earns the citation when it states the procedure, the operating surgeon, board certification, typical recovery, and a price range or what drives the price. A page reading call for a consultation gives the engine nothing to quote.
What about Perplexity and Google AI Overviews for surgeons?
The public evidence is thinner for these two. Yext folded Perplexity into its 86 percent brand-managed finding without a separate share, and it did not study Google AI Overviews. Temso did not include Perplexity. No study found here measures either engine on surgeon prompts.
ClinicAds fills the gap by direct reading. Perplexity prints numbered sources under every answer, and AI Overviews attach linked sources to the summary. In both, run the prompt, record the domains that appear, and treat repeats as the surfaces to fix first. That is an observation method, not a published benchmark.
How should a surgical practice audit sources engine by engine?
A practice should log sources per engine for one fixed prompt set, because a combined list hides which engine each fix helps. Fifteen prompts across the top three procedures is enough for a first read, repeated monthly because answers change between runs.
- 1. Write 15 prompts: best surgeon for a procedure in the city, revision specialists, cost questions, and surgeon comparisons
- 2. Run each in ChatGPT, Perplexity, Gemini, and Google AI Overviews, logged out, with the date recorded
- 3. Record whether the surgeon is named and every domain each engine cites
- 4. Mark each domain as owned, society or board, review platform, hospital, press, or competitor
- 5. Fix the most repeated surface per engine first, then rerun the same 15 prompts 30 days later
What is the cost case for fixing the sources an engine uses?
The cost case rests on booked consults and consult-to-surgery rate. ClinicAds plans surgical paid media at $80 to $150 per booked consult, with 5 to 10x return on $5,000 to $10,000 a month. A listing or procedure page fix that adds three booked consults a month without ad spend is worth $240 to $450 in avoided media cost, before any surgery that follows. These are agency averages, not guarantees.
Fix the surface the practice's own log shows repeating, not the one a study ranks first. Cross-industry shares tell a practice where to look, and the log tells it what to change.
Does a plastic surgeon need a different AEO plan for each AI engine?
No, but a different first fix. Procedure pages, accurate directory profiles, and post-op reviews feed all four engines. The engine split only decides which to do first, and the practice's own source log shows that.
Is there independent data on which sources AI engines cite for plastic surgery?
Not found for this post. Temso covers healthcare providers broadly and Yext covers several industries, so a practice should confirm either pattern in its own prompt log. Both companies sell visibility products.
Which engine should a surgical practice check first?
Check whichever engine already names competitors for the practice's lead procedure. Start with ChatGPT and Google AI Overviews for reach, then Gemini and Perplexity.
How often do the answers change?
Often. The same prompt can return different surgeons between runs, so log each prompt on separate days before concluding anything.