Placement screen
ChatGPT Ads management for audiology practices
A card under the answer, while you earn a place in it
A labelled ad under ChatGPT's answer while the AI search work earns the practice a place in it.
What is included
- An advertiser account in the practice's name, with the spend billed by OpenAI straight to the practice
- The health-advertiser application prepared with you, including proof of licensure if OpenAI asks for it
- Ads that state facts (who, where, which services, how to book) and make no treatment claims
- A landing page that matches each ad, because OpenAI's standards cover the page as well as the ad
- Enquiries and calls reconciled to booked appointments, and a step-down as ChatGPT names the practice
How it is measured
- Enquiries and booked first-time appointments from ChatGPT ads, reconciled each month
- Cost per booked appointment, rather than cost per click
- Questions where ChatGPT names the practice in the answer, from the fixed prompt set
- The share of the ad budget still covering questions where the practice is not named yet
Every engagement starts with the SEO Audit: where your practice can be found today, and the order to fix it in. You keep it whether or not you continue.
Placement screen
Where a practice can appear, conversation by conversation
Describing a symptom
“My ears have been ringing for a week. Should I be worried?”
- Answer
- Earned. The practice can be the source the answer draws on, if it wrote the clearest page
- Ad
- Never. Nobody's ad shows: OpenAI keeps ads out of personal health conversations
Comparing options
“Should I see an audiologist or buy hearing aids over the counter?”
- Answer
- Earned. A page that answers the question as it was asked
- Ad
- Possible. Where OpenAI judges the conversation suitable for an ad
Asking about cost
“How much do hearing aids cost at a private practice?”
- Answer
- Earned. Plain statements an assistant can lift whole
- Ad
- Possible. Stating facts, with no treatment claims
Choosing where to go
“Which audiologists near me are worth seeing?”
- Answer
- Earned. Only if the practice is the best described and best corroborated nearby
- Ad
- Possible. Holding the space while the practice is not yet named in the answer
The circle is the answer, which only the AI search work can earn. The triangle is the labelled card under it; with the arrow, no response: OpenAI places no ad there. Possible means the policy allows a card, not that one will run. OpenAI decides each placement and approves health advertisers one at a time. The questions are illustrative.
No ad buys a place in the answer. It can only sit under one, labelled.
Presenting complaint
A patient asks ChatGPT which audiologists near them are worth seeing. The answer names three practices, and the space under it can now carry a labelled ad. A practice in neither place never learns the conversation happened.
What an ad can do, and what it cannot
Free and Go users of ChatGPT in the US can see sponsored cards under an answer, matched to what the conversation is about. They are labelled and kept apart from the response, and OpenAI says ads never shape what the answer says. So no ad gets a practice named in the answer. That is AI search work: being the best-described, best-corroborated practice nearby. The ad holds the space under the answer while that work earns a place in it.
Where it can appear, and where it never will
OpenAI keeps ads out of personal health conversations. Someone describing their own ringing ears sees no ad, from you or anyone else. A card can appear when someone is comparing options, asking about cost or choosing where to go, and only where OpenAI judges the conversation suitable. Health advertisers are approved one at a time, and OpenAI may ask for proof of licensure. We prepare that application with you; if the practice is not approved, nothing is spent.
Facts, not treatment claims
OpenAI restricts ads that claim to prevent, diagnose or treat a condition, and the FTC expects any health claim to be backed by evidence. So the ads say what is true and checkable: who the practice is, where it is, which services it offers and how to book. OpenAI's standards cover the page an ad opens as well as the ad, so each one lands on a page that says the same thing.
A small test, measured at the appointment
It is a new channel, so it starts as a small test budget and never becomes the plan. Advertisers see totals such as views and clicks, never the conversations, so we measure what follows: enquiries from the ad's landing page, and calls and bookings in the monthly report. As the fixed prompt set shows ChatGPT naming the practice in its answers, the ads step down for those questions.
Questions
Questions about this piece
Can you run our ChatGPT ads without the SEO?
No. They run alongside the SEO work, mainly the AI search side of it: the ad holds the space under the answer while that work earns the practice a place in it. A practice that only wants ads is better served by an agency that does nothing else, and we will say so.
Will our ad show when a patient describes their symptoms?
No. OpenAI does not place ads near personal health conversations, so nobody's ad shows there. What reaches that patient is the answer itself, which is why the AI search work matters more than the ad.
Can every practice advertise on ChatGPT?
Not automatically. OpenAI approves health advertisers one at a time and may ask for proof of licensure. We prepare the application with you, and if OpenAI declines, nothing is spent.
Who sees what the patient typed?
Nobody outside OpenAI. Advertisers, including us, never see the conversations, only totals such as views and clicks. That is also why there is no ChatGPT version of the Google Ads search terms report.
Whose account is it, and who pays OpenAI?
Yours. The advertiser account is in the practice's name with its own payment details, so OpenAI bills the spend to the practice. Our fee covers the management. If we part company, the account stays with you.
This one rarely works alone.
The audit decides which pieces you actually need and in what order. You keep it whether or not you continue.
Works with