AI SEARCH · TELEHEALTH PATIENT JOURNEY

How Do Patients Use ChatGPT to Choose a Telehealth Provider?

David TerrellFounder, ClinicAdsOctober 27, 20266 min read

Telehealth patients use ChatGPT in four ways: to research a condition, to check price and insurance, to compare brands, and to test whether a provider is legitimate. The comparison and legitimacy prompts name specific brands, so those are the two a telehealth brand can win or lose.

AEO, GEO, AI visibility, and AI SEO are four names for the same work: getting a brand named and cited inside AI answers. This post uses AEO. The [plastic surgery version of this question](/blog/how-patients-use-chatgpt-to-choose-a-plastic-surgeon) follows a local, in-person surgical decision. A telehealth patient has no local pack to fall back on, compares national brands side by side, and worries about whether an online prescriber is real, so the journey turns on comparison and trust instead of a consult.

KEY TAKEAWAYS
  • KFF's March 2026 poll of 1,343 U.S. adults found that 32 percent used an AI chatbot for health information in the past year, and 77 percent are concerned about the privacy of medical information given to AI tools.
  • No independent survey found while writing this post measures how many patients use ChatGPT to choose a telehealth provider, so the four-stage journey below is a ClinicAds planning model, not a measured funnel.
  • A telehealth patient typically asks four kinds of prompt: condition research, price and insurance, brand comparison, and legitimacy checks. The comparison and legitimacy prompts name specific brands, so those are the two a brand can win or lose.
  • ClinicAds plans telehealth paid media at a $90 to $250 CAC with payback under 3 months, so a brand missing from comparison answers pays for patients it could have reached without ad spend. These are agency averages, not guarantees.

How many patients use AI chatbots for health questions?

About a third of U.S. adults used an AI chatbot for health information in the past year. KFF's poll fielded February 24 to March 2, 2026, with 1,343 adults, found 32 percent, and 77 percent of adults were concerned about the privacy of medical information they give AI tools.

No independent survey found while writing this post isolates patients choosing a telehealth provider with ChatGPT. The KFF number covers all health questions, so it shows the channel is large enough to test, not how many telehealth patients use it.

What does a patient ask ChatGPT about a condition?

A patient first asks what a condition or treatment is, who it suits, and what the risks are, for example how GLP-1 medication works or whether low testosterone can be treated online. ChatGPT answers from general medical sources, so a brand is rarely named at this stage.

The vocabulary carries forward. A patient who learns a drug name or a lab term will use it in the next prompt, and condition pages on a brand's site that use the same terms and cite clinical sources are the ones a model can match later.

How do patients check price and insurance in ChatGPT?

Patients check cost by asking what a treatment costs per month and whether insurance covers it, such as how much online GLP-1 treatment costs without insurance. ChatGPT returns a range and a caution that pricing varies by brand and medication, and it may cite brands that publish a price.

A brand that publishes monthly price, what is included, and whether it accepts insurance on a crawlable page gives the model a figure to quote. A brand that shows price only after a long intake form gives it nothing, and the answer names competitors that did publish.

How do patients compare telehealth brands in ChatGPT?

Patients compare brands by asking for the best provider for a condition or a head-to-head, such as best online provider for hair loss or one named brand versus another. The answer names roughly 3 to 6 brands, and a patient usually starts intake with a brand that appeared.

This is the one discovery stage with no local fallback. A brand left out of the comparison is outside the patient's consideration set before any ad runs. The sources behind those names are covered in the ClinicAds post on [where each engine gets telehealth recommendations](/blog/chatgpt-perplexity-gemini-ai-overviews-telehealth-recommendations), and the [telehealth program](/telehealth) treats AEO as part of search visibility alongside paid media and speed-to-lead.

How do patients test whether a telehealth provider is legitimate?

Patients test legitimacy by asking whether a named brand is safe, whether its clinicians are licensed in their state, and whether the medication is FDA approved or compounded. Online prescribing carries a trust question an in-person practice does not, and the answer decides whether intake starts.

Physician skepticism feeds these answers. An Omada Health survey, vendor-sponsored and fielded in early 2025, found 67 percent of primary care doctors believe third-party telehealth GLP-1 access could put patients at risk. A brand that states clinician licensure, state coverage, medication sourcing, and how care continues after the first visit gives the model reassuring facts to cite.

Which prompt maps to which stage of the patient journey?

Each prompt type maps to one stage, and each rewards a different asset. The table shows the four stages, a typical prompt, whether the answer names a brand, and what decides the outcome. It is a ClinicAds planning model, not a measured funnel.

The four prompt types in a telehealth patient's ChatGPT journey. ClinicAds planning model, not a controlled study.
StageTypical promptNames a brand?What decides the outcome
1. Condition researchHow does online TRT treatment work?RarelyGeneral medical sources; condition page wording matters for later matching
2. Price and insuranceHow much does online GLP-1 treatment cost per month?SometimesPublished monthly price, inclusions, and insurance policy on the brand's own site
3. Brand comparisonBest online provider for hair lossYes, 3 to 6 namesReview sites, comparison publishers, Reddit threads, and press mentions
4. Legitimacy checkIs this brand legit and licensed in my state?Yes, one named brandClinician pages, state coverage, medication sourcing, and complaint history

What should a telehealth brand do about it this month?

A telehealth brand should run the four prompt types in ChatGPT for its lead condition, log the answers, and fix the earliest stage where it is missing. Record whether the brand is named in comparisons, whether quoted prices match the site, and whether the legitimacy answer is accurate and complete.

The fix follows the stage. A missing price is a quick page edit. A missing comparison name is a third-party sources problem that takes longer. Add one question to intake asking whether an AI assistant helped the patient choose, because intake is the only place this journey becomes visible. At a $90 to $250 CAC and an LTV of $600 to $2K, each patient an AI answer sends without ad spend improves payback.

  • Run the four prompt types monthly for the lead condition
  • Publish monthly price, inclusions, and insurance policy on crawlable pages
  • State clinician licensure, state coverage, and medication sourcing plainly
  • Ask new patients at intake whether an AI assistant helped them choose
FREQUENTLY ASKED

Do patients start telehealth treatment from a ChatGPT answer alone?

Rarely alone. Patients use the answer to build a shortlist and then check reviews, Reddit, and the brand's site. ChatGPT shapes which brands are considered; the brand's own pages and third-party reviews decide who gets intake.

Can a telehealth brand see which ChatGPT prompts led to a patient?

Not directly, because ChatGPT does not report prompts to businesses. A monthly prompt log and an intake question about AI assistants approximate the effect.

Does publishing price help a telehealth brand show up in AI answers?

It helps for price prompts, because the model can quote a figure from the brand's own page. It does not by itself earn a place in a comparison, which depends on reviews and third-party mentions.

Is there independent data on patients using ChatGPT to pick a telehealth provider?

Not found while writing this post. KFF measures AI chatbot use for health information in general, at 32 percent of adults in March 2026. The four-stage journey here is ClinicAds' planning model.

Does ClinicAds sell AEO as a separate service?

No. ClinicAds treats AEO as part of the search visibility system inside its three-system scope, alongside paid media and speed-to-lead, and does not sell GEO, AEO, or AI SEO as separate line items.

See what ChatGPT tells a new patient about your brand

ClinicAds will run the four prompt types for your lead condition and send back whether your brand is named, whether price and licensure are quoted correctly, and which fix comes first.