Your patients are asking AI. Do you need to act yet?
People now ask ChatGPT and tools like it about symptoms, diagnoses and medicines. This guide helps you decide one thing: whether that should change what your brand does before the end of the year.
Five minutes · three questions · one page you can forward. Written for pharma marketing and omnichannel teams. No prior knowledge assumed.
Start with one person
A patient leaves the clinic.
They understand the diagnosis. They do not understand what happens next.
That night, they ask an AI assistant.
Not to replace their doctor. To work out what to ask at the next appointment.
Why it matters
Whatever that answer says is now part of your patient’s understanding — whether or not anyone at your company has ever read it.
The jargon, once
“GEO” is a new label for an old question.
You will see it written as generative engine optimization, usually in an agency deck. Strip the label away and it asks one thing:
When someone asks AI about your disease area or your medicine, what does it say — and is the answer any good?
Why it matters
It is a question about the quality of an answer. It is not about buying placement inside ChatGPT, and it is not a ranking you can game.
How big is this really
Not yet the first place people go.
Where U.S. adults went first for health information. YouGov, July 2026 · previous three months.
Why it matters
For most audiences AI is still an extra source alongside Google, a clinician and your own materials.
How big is this really
And climbing quickly.
17%rising to29%
Adults using AI for health information at least monthly, June 2024 to May 2026. A different study asking a different question — which is exactly why you cannot join these two numbers into one curve.
Why it matters
Both things are true at once. The gap between them is where most of the panic comes from.
What a score can tell you
The first number you will be shown.
A GEO dashboard asks a set of questions across services like ChatGPT, Gemini and Perplexity, then counts how often your medicine is mentioned.
100%
Your medicine appeared in every answer checked.
What a mention means
The name appeared somewhere in the answer. That is the whole definition.
What a score cannot tell you
The same answers, read properly.
01
Missing context
Two answers left out an important limitation.
02
Invented evidence
One named a guideline that does not exist.
03
Weak support
Only one link actually backed the claim beside it.
The practical lesson
First ask whether the answer is good. Then ask whether you appear in it.
Why it matters
A mention is not a quality measure. The score was 100% and the answers were still wrong.
Who is asking
Three people. Three different questions.
Why it matters
One "AI visibility" score averages these three people together. The averaging is exactly where the useful signal disappears.
Who is asking
The moment matters more than the therapy area.
Why it matters
"Do people use AI for oncology?" is too vague to plan around. "What does a caregiver ask in the week after diagnosis?" you can actually answer.
Who is asking
Where the early evidence is strongest.
None of these is a ranking. Each is a reason to look, and each carries a limit worth reading.
01
↓
Before you start
How this guide works
Fifteen short screens, about five minutes. You read for the first ten, answer three questions, and leave with one page you can send to someone.
01
Turn the page
Use Next at the bottom right, the arrow keys, or swipe on a phone. One idea per screen.
02
You only tap three times
The three questions in the middle. Everything else is reading.
03
Your answer is kept
It shows at the top and shapes the last three screens. It survives closing the tab, and you can change it whenever you like.
04
Dotted words have a definition
Tap any of them. "Contents" jumps anywhere. "Sources" holds every study, template and checklist.