For most healthcare marketing teams, AI search was not a serious line item two years ago. SEO was. Paid search was. Reputation, physician directories, service-line pages and conversion certainly were.
AI was something to watch.
That is changing pretty quickly.
This year, Becker’s asked marketing leaders at 17 major health systems about their priorities for 2026. AI search, answer engine optimization and the changing digital front door came up repeatedly.
Northwell Health CMO Ramon Soto raised a question that gets to the heart of it: as search moves toward AI-generated answers, how does a healthcare brand show up, get attribution and eventually transact?
University Hospitals talked about redesigning the digital front door around an AI-driven patient journey. Universal Health Services included SEO and answer engine optimization among its priorities.
These are not conversations about experimenting with ChatGPT. They are marketing conversations.
And I think there is an important second question hiding behind all of this.
What exactly should we be measuring?
Showing up is not the same as being visible
The first instinct is usually to measure mentions.
Take 100 questions that patients in your market might ask AI. Run them through an AI platform. Count how many times your health system appears.
You now have an AI visibility score.
That’s useful. We do it too.
But on its own, that number doesn’t tell you very much.
Imagine a patient asking for the best place nearby for a knee replacement.
Hospital A is the first recommendation. The answer explains why, mentions its orthopedic program, names relevant physicians and links to useful information.
Hospital B is mentioned in the fourth paragraph as another local option.
Both hospitals appeared in the answer.
Calling that the same level of visibility doesn’t make much sense.
And there is another problem that matters even more in healthcare: the answer can be wrong.
A physician has left. Weekend hours changed six months ago. A location doesn’t provide the service the model says it does. Insurance information came from an old directory.
Technically, the organization is visible.
From a patient’s perspective, that visibility isn’t helping.
Patients are already using AI this way
This isn’t hypothetical patient behavior anymore.
KFF reported in March that 32% of U.S. adults had used AI tools or chatbots for health information or advice during the previous year.
The reasons are interesting.
Among people who used AI for health information, 65% pointed to the speed of getting an answer. More importantly for healthcare marketers, 41% had used AI before deciding whether to see a healthcare provider.
By June, KFF found that 29% of U.S. adults were using AI for health information at least monthly.
AI is also being inserted directly into conventional search. Stanford’s 2026 AI Index, citing RAISE Health data, reported Google AI Overviews appearing on 84% to 92% of the health-related queries studied.
So we shouldn’t assume there is a separate group of “AI patients” who have abandoned Google and moved to chatbots.
For many patients, AI is simply becoming another layer between their question and the healthcare organization.
That layer deserves measurement.
We need to look at more than mentions
At Litode, we’ve been working with a fairly simple framework:
Access → Presence → Position → Accuracy → Outcome → Change
It isn’t meant to create six more vanity metrics for a dashboard. It is a way to diagnose what is actually happening.
Access comes first because an AI system needs to understand the organization before it can represent it well.
Can it make sense of your physicians, specialties, locations and service lines? Is the information consistent across the sources it encounters?
Then comes presence.
Do you actually appear when patients ask the questions that matter to your organization?
That’s the obvious part, and it’s still worth tracking. But once you know that, the next question is much more useful:
How are you positioned?
Being mentioned alongside four competitors isn’t the same as being recommended. Being described as a nearby option isn’t the same as being identified as a leader for a particular type of care.
This is one of the biggest differences between measuring traditional rankings and measuring AI answers. There isn’t always a neat position 1, 2 or 3. The language around the recommendation matters.
Then we get to the part healthcare can’t afford to ignore: accuracy.
Is the physician still practicing there?
Are the hours correct?
Is that service actually available at that location?
Does the answer describe the organization accurately?
Healthcare marketers have spent years trying to maintain accurate information across websites, directories, maps and physician profiles. AI doesn’t make that problem disappear. It adds another place where inconsistent information can surface.
Then follow the patient
This is where I think AI visibility gets much more interesting.
Suppose an AI answer recommends your health system.
Where does the patient go?
Maybe the model cites your service-line page. Great.
Maybe it sends them to a physician directory.
Maybe the source is a third-party site.
Maybe it recommends your organization but gives the patient no obvious next step at all.
Or maybe your competitor appears immediately afterward with a clearer route to scheduling.
If we only count the mention, all of that disappears from the measurement.
Healthcare marketing teams already spend a lot of time thinking about patient journeys after a click. AI means we also need to understand what happens before that click exists — or when there is no click at all.
The metric I’m most interested in is change
There is one more piece that tends to get lost.
Your team finds a problem.
Perhaps a major service line isn’t appearing. Maybe an AI answer repeatedly surfaces outdated location information. Maybe a competitor dominates a category where your organization should be competitive.
So you do something about it.
You improve a page. Fix structured data. Update physician information. Clarify access information. Correct conflicting sources.
Then you wait.
Did anything happen?
This sounds obvious, but it’s surprisingly easy to build an AI visibility program that keeps finding problems without showing whether the work actually changed the answer.
That is why we include change in the framework.
If a marketing team spends time fixing something, it should eventually be possible to see whether AI’s representation of the organization moved with it.
Otherwise, we’re just producing another report.
Healthcare doesn’t need another score for the sake of having a score
There will be a lot of AI visibility products, scores and optimization tactics introduced over the next few years.
Some will be useful. Some probably won’t survive very long.
For healthcare marketers, I think the better approach is to stay close to the patient question.
What did they ask?
Who appeared?
What was said?
Was it correct?
Where could the patient go from there?
And did that answer change over time?
Those questions aren’t particularly complicated. That’s probably a good thing.
Because the goal isn’t to “win AI.”
It’s to understand an increasingly important part of how patients discover care — and make sure your organization is represented accurately when they do.
We made a short video explaining the framework we’ve been using at Litode.
Access → Presence → Position → Accuracy → Outcome → Change
Watch the framework video on YouTube
With HCIC starting in Orlando in a few days, I’m curious how much of the healthcare marketing conversation will move in this direction.
A year ago, much of the discussion was about whether AI would change search.
It seems like we’re getting past that question.
Now we need to figure out how to measure what replaced it.