Prospect Finder
← All posts
Feature

What AI tells patients and providers about aesthetic products

Chatbots now answer questions about injectables and devices. Which brands they name, where answers come from, how accurate they are, and what reps can say.

By Peggy Chung, CEO, Prospect Finder

Key takeaways

  • In a June 2026 index of AI answers about aesthetics, Botox scored 95 out of 100 for citation share. RealSelf, a patient review site, came third at 86, ahead of every filler brand. AbbVie/Allergan and Galderma accounted for 80% of drug and device brand citations.
  • Assistants draw on different sources. In an analysis of about 825,000 citations in AI answers to U.S. health questions, ChatGPT leaned on medical and academic sites, while Google's AI products cited YouTube and Reddit far more often.
  • In a peer-reviewed test, physicians rated about 77% of ChatGPT-4's answers on neurotoxins as medically accurate, and about 63% of its answers on fillers.
  • Chatbots aren't bound by a product's FDA labeling, so their answers can describe unapproved uses. FDA rules still apply to anything a rep says in response.

Patients increasingly arrive at a consultation having already asked a chatbot about their options, and providers have the same tools on their phones. The answers they get decide which brands come up and what claims sound familiar. Those answers don't come from the manufacturer, and nobody reviews them for compliance.

How many patients ask AI before they book?

About three in ten, among engaged aesthetics consumers. NewBeauty's Summer 2026 State of Aesthetics report found that 30.3% of respondents had used AI to research treatments, and 52% said they were comfortable doing so. The respondents were NewBeauty's own BeautyEngine community, surveyed in Q2 2026, so the share among all patients is likely lower. AI still ranked fifth among their research sources, behind search engines, social media and physician websites.

Which aesthetics brands does AI mention, and why?

The ones patients already talk about online. 5W AI Communications, a public relations firm that sells AI-visibility audits, scored 25 aesthetic brands by how often ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews cite them. Botox led with 95 out of 100, followed by CoolSculpting at 87.

RealSelf came third at 86, ahead of Juvéderm (83), Dysport (78) and Restylane (77). RealSelf sells no treatment. It is a marketplace of patient reviews and doctor Q&A. Its rank suggests the assistants reward brands that patients write about in large volume, not only brands that run large ad campaigns. The accompanying release put the split among drug and device makers at 47% AbbVie/Allergan, 33% Galderma, and 20% for Merz and Revance combined.

5W's method hasn't been independently reviewed, and its scores weight how often a brand is cited, across how many assistants, and how authoritative the citing sources are. The pattern is still useful for reps at smaller or newer brands. A patient who asks a chatbot about wrinkles or volume loss will probably hear four or five names, and yours may not be one of them.

Do all AI assistants use the same sources?

No, and the gap is large. LLM Pulse, an AI search analytics company, analyzed about 825,000 citations in AI answers to unbranded U.S. health questions. PubMed Central, the National Institutes of Health's archive of journal articles, was the single most-cited domain. ChatGPT drew most heavily on government, academic and medical sources. Google's AI Overviews and AI Mode cited YouTube and Reddit much more often.

The same question can therefore get a journal-based answer in ChatGPT and a forum-based answer in Google. Providers and patients don't necessarily use the same tool, so a provider may have read a different answer than the patients in the waiting room.

How accurate are AI answers about injectables?

Mostly right, with weak spots. In a study in the Journal of Cosmetic Dermatology, physicians rated ChatGPT-4's answers to 14 common patient questions, seven about neurotoxins and seven about fillers. About 77% of the neurotoxin answers and about 63% of the filler answers were judged medically accurate. The reviewers often disagreed with each other about individual answers, and they found the chatbot struggled with procedural detail. The study tested an older model, and current versions may do better.

Mistakes in AI health answers aren't limited to small studies. In January 2026, after a Guardian investigation flagged misleading health summaries, Google removed AI Overviews for some medical searches. The investigation's examples involved liver test ranges and cancer diet advice, not aesthetics, but the cause is the same. These systems summarize whatever they find, and they don't check it against a product's labeling.

That last point matters most for reps. A chatbot doesn't know which indications a filler or energy device is cleared for. It repeats what clinicians and patients write online, and aesthetics is full of discussion of off-label techniques.

What can a rep say when a customer quotes AI?

Less than the chatbot did. FDA rules on promoting medical products apply to what a company and its reps say, whatever the source of the question. In practice:

  • Stay on label. If a provider asks about a use they read about in an AI answer and it isn't in the product's cleared or approved labeling, a rep shouldn't discuss or promote it. Say that it falls outside the labeling, and offer to pass the question on.
  • Route off-label questions to medical affairs. Most companies have a medical information process for unsolicited questions about unapproved uses. FDA's final guidance on scientific information on unapproved uses, issued January 6, 2025, sets out when firms can share that kind of information with health care providers. It is generally handled outside the sales conversation.
  • Know whether your company corrects misinformation. FDA's July 2024 revised draft guidance describes how firms can voluntarily correct false or misleading internet content about their own products that others created, without that correction counting as promotion. It is still a draft. Whether FDA would treat a chatbot answer, which changes from user to user and day to day, the same way as a website or social media post hasn't been settled.

Follow your company's compliance policy. This article describes the rules in general terms and isn't legal advice.

What to do with this

  • Ask the assistants yourself, every month. Put the questions your customers ask into ChatGPT, Perplexity, Gemini and Google. Try "best filler for lips", "Morpheus8 vs microneedling" or "is [your product] safe". Note which brands appear, what is said about yours, and which sources are cited.
  • Flag errors to your company, not the customer's chatbot. If an assistant gets your product wrong, send the question, the answer and its sources to marketing or medical affairs. The fix, if there is one, is a company decision.
  • Bring sources a provider can check. ChatGPT leans on medical and academic sources, so a provider who uses it may be reading summaries of journal articles. Bring the label and the studies behind it.
  • Treat patient reviews as part of your market. RealSelf's third-place score suggests AI assistants weigh what patients write. Practices that use your product and collect detailed reviews help keep it in the answers.
  • Rehearse the moment a provider quotes AI. The hardest version of this conversation is a provider repeating an off-label claim from a chatbot answer, and the time to practice staying on label is before it happens. Prospect Finder's Pitch Simulator lets reps rehearse a meeting with an AI prospect built from a real practice in their territory, one that pushes back the way providers do. Book a demo to try it.