AH074 - How Fast is the Way Consumers Search for Care Evolving? Almost Beyond Comprehension, with Carrie Liken

Transcript

Lightly edited for clarity.‍

[00:28] Justin Venneri: Hello and welcome to another episode of the Astonishing Healthcare Podcast. This is Justin Venneri, your host and Senior Director of Communications at Capital Rx. I'm excited to welcome Carrie Liken back to the studio. She joined us last year for Episode 14, the Rise of Conversational AI: Healthcare Search and Patient Journeys Will Never Be the Same. Of course, one of those things that we're going to link is the reason we're here. Earlier this year, Carrie wrote a Substack that was pretty polarizing, I think, and was along the lines of the death of the website as we know it or an obituary for the website. What was the title again?

[01:02] Carrie Liken: It was an Obituary for the Website, I think. Yeah.

[01:06] Justin Venneri: And then the follow-up she just wrote on AI Mode really caught my attention, given some of the stuff we're working on here at Capital Rx and our recent acquisition of Amino Health on the care navigation side. But, that article is pretty powerful and on trend broadly. So, Carrie, before we get into it, welcome back to the show.

[01:23] Carrie Liken: Nice to be here. Thanks for having me.

[01:24] Justin Venneri: Catch us up on what you're doing. If you want to take, you know, a minute or two on your background for anyone that didn't listen to that first episode, and I think you're in a new role, congratulations.

[01:33] Carrie Liken: No, thank you. So I wrote this Substack and I started the Substack in January because I have been for, let's see, since probably 2007 or so, really, really interested in the intersection of healthcare technology and consumer behavior. And so I really got interested in that for the first eight years where I was at Google, and I was on the healthcare team there. Then I moved over to Yext for another eight years, and I got to help build the healthcare team there, but got to talk to a lot of healthcare organizations about the same concept of how people are finding information and how to get that information out to places where they're looking.  

And then more recently, I joined Snowflake, and I'm on the healthcare team at Snowflake. And what's interesting about that is that Snowflake doesn't necessarily work exclusively with marketing teams, which is what I had been working with from 2007 to last year. But what it does is it organizes data and it's a data cloud. And data is really important for all organizations, but especially for healthcare organizations. And I always say you can't have an AI strategy without a data strategy first. And I used to talk about that at Yext, and I talk about it now. And the reason we're talking today has a lot to do with Google and their data strategy. So it's been an interesting evolution.

[02:54] Justin Venneri: We were talking about the rise of conversational AI, and in preparing for this. And you know, it's funny; this is truly, whether people believe it or not, this is pretty much an unscripted discussion we're having here today. I was like, hey, I just read this. Let's hop on!

[03:09] Carrie Liken: [Laughter] I can confirm – I received no script from you.  

[03:13] Justin Venneri: Let's roll with it and see what happens. You know, we're playing around with AI Mode in Google and really just trying to understand, like what's happening with search. How are people tracking their traffic, their sessions, like all of this stuff is changing. And John Asalone and AJ were on the podcast about the acquisition of Amino, our care navigation solution - now Judi Care, but they were talking about how consumers really still do often rely on Google to find the care they need.

So what do you see? Here we are, end of June 2025. Like, what's happening with search? Why are websites as we know them dying? Talk to us about your Substack and your thoughts on this.

Reference Links & Related Content

• AH070 - Inside Capital Rx's Acquisition of Amino Health: Creating the Health Benefits Platform of the Future, Today‍
• Maybe we should write an obituary for the website (3/20/25)
• ‍AI Mode is the training ground for AI Overviews (and yes, websites are still dying) (6/7/25)‍

‍

[03:49] Carrie Liken: Sure. I think if it's okay, I'll start with the evolution of search because I think that's important, at least from my 2007 days to today. It's important to understand how that has changed because there are big changes, fast changes that are happening in the last six to eight months that we'll get to when we talk about exactly what's going on with Google. So early on, when we all had the opportunity to actually finally search.

So, this is in, let's say, early 2000s, we could look for things about our healthcare. For example, we could go to Google and you'd type in and practice typing in so you could get the best result, and you would get blue links. So it was usually “cardiologist near me” or “what is a cold?” - very simple ways of searching for something. And throughout the years that I was at Google, the search evolution of how people looked for information and how they actually typed in that information changed, and it became a little bit more complex. And that was because Google was able to deliver better results. And so people could actually type in a more complex way. And Google would retrieve the information through the blue links and through evolution of their search results too, through the knowledge graph and other ways of presenting information to give people the answers they needed very simply and easily search, sometimes even on the results page itself. So you wouldn't even have to click off into a blue link.  

Fast forward to during my Yext years, fast forward all the way till I would say when ChatGPT was launched, Google hadn't really made any changes at all to its results page to fundamentally really change things. They would present different information so that you could get information right on that results page. But they weren't changing the way it was actually. People could search for it and deliver it. ChatGPT comes along and all of a sudden people are able to talk to technology. And this is what we talked about in the last podcast.

[05:40] Justin Venneri: Yeah, it was great.

[05:42] Carrie Liken: Yeah, with Christian [Ward], where we have this opportunity now to ask really long questions. The technology understands, can respond, and basically summarize. So you never have to hunt through blue links anymore to get information, but you could keep going, keep, keep having that conversation back and forth. Okay, so now that happens, and Google sees a threat, they say, well, if people are going to find that, this is actually better and easier to find an answer because they don't have to do the work anymore. People can just literally read the little summary that ChatGPT is providing. What do we do? So they got scared.  

They then really started to amplify, making big changes to their search experience. And over the course of a year and a half or so, they had introduced ways of getting information out in a summarized way on the results page. When somebody searched and they started to see that, they called this AI Overviews. They started to see that people when searching for long queries, so long questions, you could actually then get the information through a summary. And Google said, “Oh, this is great.” Okay, so we have this AI Overview, and people are finding what they need. Generally, they're not clicking over to websites anymore because people can find exactly the answer in that overview. But then Google said, I don't actually, I can't quote what Google said, but in my mind I'm thinking this is what Google said, “We need to continue to amplify this experience. We need to make it even more conversational because there's a lot of competition (Perplexity, and ChatGPT, and Claude, and Grok, and others).” You can go back and forth and access real-time information from the web. So Google has that real-time information. And they said, “Let's go ahead and launch an experimental experience.”

It's called AI Mode. So this is a mode on Google that has now been released to everyone in the United States. And I think they're rolling it out in a phased rollout. It's funny because I have access to it, but then if I talk to some people, they don't yet have access to to it. But there's a little link on the Google search result where you can click on that link and it says AI Mode. You can type in a long form question just like you would go to ChatGPT or Gemini to do that.

[07:57] Justin Venneri: I actually had it pop up on something I was toying with the other day, and it was like, would you like to look in AI Mode for this?

[08:04] Carrie Liken: There you go.

[08:04] Justin Venneri: So they're kind of prompting you in different ways, which is neat.

[08:07] Carrie Liken: And I noticed they're prompting it. I have the search app on my phone, and they now have a little piece of the search app where you could click on it and go directly to AI Mode. So they're now starting to think how can we get more people to use it? So if you go to it, you can ask a long-form question, very long question, and get a really good response, but then you could keep going. So, on Google now you can have a back and forth, back and forth, back and forth with a question, read the responses, keep going and digging deeper into the responses, keep asking questions, keep understanding what it is that you want to understand, and have that conversation on Google.

It goes back to the fact that Google said we want to keep people on our website on google.com we want to keep people here. We don't want to lose users. They're seeing that they're losing some users to ChatGPT and to other places. And so they said, how do we keep people in this environment and in this ecosystem? So they present this. And then what I've learned over the last, I would say month or two is that what is successful on AI Mode, they now will be pulling into the typical search experience. So AI Mode is going to continue to be experimental. You'll have that conversation, but they'll move anything that's been successful or well used or well liked, they'll move that into the traditional search experience.

[09:31] Justin Venneri: Oh, interesting. Okay. So it'll blend over time?

[09:33] Carrie Liken: Well, I can't speak for Google because I don't work there, but it sounds like it based on other podcasts I've listened to and resources I've read that that is the intention is that it will probably blend and become one. Yeah.

[09:46] Justin Venneri: So what does that mean for the content we create? Others create, especially health systems. We have a lot of health system partners and health plan partners. Like, how do we create content for people or organize our data so that people can find it and it still is credited somehow?

[10:03] Carrie Liken: Yeah, that's a good question that I have been talking to a lot of health systems about because everyone's been asking that main question, what does this mean for my content? I've spent tons of time, money, and energy on creating content in order to bring people to our website. And what we found is. And I'll. I'll talk a little bit about, I did what I call my mini non-scientific research study.

So what we found in this study is that I did two things, and this goes back to your original question of what we should do with content. Everyone pre-AI Overviews in AI Mode, everyone said if I build the content and I build the website, they will come to use a baseball analogy or a Field of Dreams analogy. So get the content out there, make sure it's SEO-optimized. And in a traditional SEO sense, people will come because as long as they're searching for something that is around what it is that I offer about content around cardiology or about pharmacy, you know, opportunities and locations and doctors, whatever it might be, they will come.

But with AI Overviews, what we're finding is that's not the case. And especially with AI Mode, it's not the case. And so this mini non-scientific research study that I did, I wanted to really understand what actually is the implication of this, because I started to hear organizations say that some of their traffic was starting to decline from their websites, but it wasn't necessarily due to what it's like. I don't know what that could be, but let's just keep monitoring it. When Google launched AI Overviews, which is just the search, the summaries on the typical google.com or on the search engine results page, they basically showed that that summary will give enough information to somebody where they don't really need to click through to any websites. So all of that information is delivered on the homepage. So what does that mean for healthcare?

Well, in my study I took 24 conditions and I wanted to see how long I could just hang out on Google and search for information and the summaries via AI Overviews. This was study number one. There were two studies. How long would it take Me before I would actually have to go to a website and how could I get to that website? And what I found was for basically all 24 conditions, I never really had to go to the website at all. Anything that I asked a question about around anything from migraine to glioblastoma to what is yellow fever, didn't have to go to a website to find this information. So it just meant that that content that used to draw people into the website now is basically just being summarized by Google and delivered to me without having to actually get me to the health systems website or even the local organization's website.

I thought, okay, well that might mean that the website might be dying and the website as we know it might actually not exist anymore or not serve its purpose anymore. Because when you look at a healthcare website, you see that there are pretty typical things that a healthcare organization has on the website. You have conditions, you have procedures, you have locations, you have doctors, you have some, you know, some other things like about us, and then you have options to log into your patient portal, pay your bill, those types of things. But maybe a website doesn't need to be or can't be structured like that anymore, and there shouldn't be as much emphasis on generic information.

Because the other thing that I found in that first study was that predominantly the content that was showing when I was looking for things like what is migraine, and you know, the 24 different symptoms or conditions, it was mainly Cleveland Clinic content, mainly Mayo Clinic content, mainly Hopkins content. So what's the point of a healthcare organization creating content that's essentially the same, and it's never going to be shown or prioritized. So I thought, okay, this is interesting. And this created a lot of buzz in the industry because I was threatening, I think the concept of, “Oh geez, we've spent 20 years on building something that now we need to figure this out and change this, what are we going to do?”

But things got worse because then Google launches AI Mode, which is the conversational back-and-forth that you can have in this experimental environment. This is where not only I was thinking, how is this going to change how people get to a website, but also will anybody ever need to visit ever again to do anything on a website? So I wanted to see what that impact was. And I did my second non-scientific research study, took the same 24 symptoms and conditions, and I had a conversation with Google essentially with AI Mode to go from what is all the way to can you help me book an appointment? So I wanted all of the different searches and all of the different questions that could be answered typically was about three to four, sometimes maybe four to five different back-and-forths that I had with AI Mode in order to do that.

[15:06] Justin Venneri: I mean, in a way, that's that initial part of the journey, right? To figure out what's going on and where can I go to get help with this, and then if possible, book the appointment, right?

[15:16] Carrie Liken: Yeah. And the only thing I couldn't do was book the appointment. Everything else was solved within that back-and-forth with AI Mode.

[15:24] Justin Venneri: Interesting.

[15:25] Carrie Liken: Which is a really big deal because again, organizations say, well, what do I do about my content then? How do I make sure that it shows up? And I think these are big questions that organizations should be asking. And organizations are concerned about it. And I don't think anybody really knows. I think the answer is we're not sure.

[15:42] Justin Venneri: One question I have is, can you tell yet how is credit given for being a source locally for information if the model is searching that site to deliver the results back to you? I think that's tracked differently now. Right. And that's one thing that in general people are struggling with is trying to figure out how to track that traffic on your site. So it's like the content still needs to be there so someone can find you, that you're a resource, and you're credible. But how are you getting credit for that? It's kind of weird quandary.

[16:12] Carrie Liken: It is weird. And I know Google does this in stages. Even when I was there, they would launch things, but then they wouldn't let the public know about the analytics of it until they had figured it out and they would launch it well before they had figured it out. So right now I'm not aware they might be, but I'm not aware that you can actually see how you show up. Let's say you were to go into some backend tools. Whether it's referrals or whether it's webmaster tools, you likely are not able to see it. But also the citations themselves, the links that are presented. There are far fewer links in each of these conversations. Far fewer links are presented to someone. You get a great summary.

So let's say I ask, “What is migraine? What are symptoms of a migraine? How do I know if I have a migraine? Who are the doctors who might treat a migraine? Can you help me book an appointment with a doctor?” Or even before that question, maybe you could ask, “Are there doctors near me who treat migraine or who are specialists in migraines? Can you help me?” But that whole progression, each one of those answers to that question or that piece of that conversation, they don't have 10 blue links anymore. You're presented sometimes one, sometimes four, sometimes five. And so the big question becomes how do you get in there? How do you make sure you get in there?  

And what I saw when I was going through all of this was that the closer you got to a location query, so the closer you got to I want to find something near me or I want to find something that is, I'm based in Boston, so I'm going to find something that's Boston related. Then you could start to pull in a local organization, but before that they were pulling in random things along with the typical Cleveland Clinic, Mayo Clinic type content.

[17:56] Justin Venneri: Yeah, got it. It's like the national, most authoritative.

[17:59] Carrie Liken: Yes, yes.

[18:00] Justin Venneri: That's interesting because we pull a lot of data into Judi, our platform, and obviously it's a closed environment because it's a member searching for something on an app now with Judi Care. But it has like, it can rank based on quality, here's your in-network options. That's, I think, helpful but that's unique because it's for the insured population, if you have that resource available, you can find stuff hopefully -- and hopefully this will get better over time for everyone being able to find care for the providers, because then it's I think more targeted leads, quote, unquote, if you can see my air quotes. It's like just driving your potential patients, new patients, to your practice. So I think it's really fascinating.  

How do you see things evolving from here as AI Mode evolves and as search evolves? I think I've seen some interesting stuff about the backend, but it's not as much how you show up, but it's more what the LLMs -- the large language models -- what they're searching and what they're sourcing. Some seem to like Reddit more, some seem to like Google. So I think that's fascinating too. How are you thinking about that?

[19:04] Carrie Liken: Great question. I have been thinking about it, and I don't think people will like the answer that I'm going to give. But I'm going to say that I reserve the right to change my mind.  

So right now, what I believe is happening is we are all responding to changes that are happening that are big, they're large-scale changes. Search has not changed in 20+ years, and all of a sudden now it's really changing. And consumer behavior is responding to this frictionless environment where you can just get the information in the summarized way, and there's a dispersion of tools, basically of people as well. Like they could go anywhere and get information from anywhere.  

But the way that the industry is responding is they're saying, well, we did it this way for SEO. So now let's look at these changes and see if we can fit in what we've done and just see if that model can just,  we'll just adjust a little bit. So we need to optimize to show in AI Overviews and we need to optimize in order to make sure that we're part of AI Mode. And this is where I think it might be, again, controversial, is that maybe we don't. Maybe that we just need to stop.  

So you could look at it in two ways. One is if you go that direction and you say, don't worry about it. Google's already determined what they are prioritizing as far as showing up. So, the generic content, the Cleveland Clinic, the Mayo Clinic, all of this high level, very trusted content that they've deemed as trusted, you can't compete there and maybe you shouldn't compete there.  

So maybe then what you need to do is build brand on the other side and make sure that you're showing up in places where you don't have to compete as much anymore. You'll never compete with Cleveland Clinic's content. They have had a content strategy that is, bar none, unbelievable for 20 years or more.

[20:52] Justin Venneri: Yeah.

[20:52] Carrie Liken: So you might not be able to do that. So you need to then start to try to figure out how to tap into people in other places. Whether you're tapping it into social, you're tapping it into newsletters, lots of other ways of marketing to people where maybe it's not all about the website anymore.  

But where it does need to be about the website is people need to know you. Once they get to you, they need to be able to transact. And this is the second piece, is even if you don't think about the content side and optimizing for SEO, which is now this new GEO -- Generative Experience Optimization -- you need to be able to convert people. And I think many organizations are behind the times when it comes to conversions anyway, conversions especially being things like online scheduling. But it could be anything. I mean, it could be an HRA, it could be signing up for a newsletter, it could be paying a bill, it could be, you know, any number of things. I'm just going to say online scheduling just because that's a, it's an easy way to talk about it.

[21:45] Justin Venneri: Yeah, it's an important part of the care journey for a patient to be able to get to the doctor that they see or the NP or whoever.

[21:56] Carrie Liken: Correct, yeah. And it's just not a possibility right now for many organizations. There's a lot of backend challenges to activate scheduling, but then also the online presentation of scheduling is very difficult as well.

And so, Google can't accommodate online scheduling yet. ChatGPT can't accommodate online scheduling yet. So people still need to get to the website in order to be able to do that. So they're either making a phone call, which is painful, or they want to book online. And right now that's not very friction free. So I feel like that needs to happen, and that needs to be the future.  

And then we could always get into, which we probably don't need to get into today, but what's the future look like? The future doesn't even look like humans searching for information anymore. It's about the agents. So agents will be looking for information on behalf of the human and then they'll also want to be able to book on behalf of the human too. And that's where the online scheduling will be very important.

[22:56] Justin Venneri: Got it. What would you say would be the first or second thing, whoever's responsible for the provider directory at a health system or the directory -- because we see a lot of issues with directories for health plans. That can be very frustrating for a member when there's out-of-date information there, when those changes happen, is it, is it really just a data quality exercise?

[23:18] Carrie Liken: 100%, yes. When I was at Yext, boy, this was a huge issue, just getting your provider data to be accurate. If you have bad data going in, you have bad data going out.  

So again, two different things there. If you have bad data going in, if you're trying to get that information out to Google, let's say in AI Mode, AI Overviews, then it could be picking up bad data from what you're sending to it. That's not good. But then on the flip side, let's say somebody needs to come to your site directly. Same thing, bad data in, somebody looking for that information, they're going to get the bad information. Then that leads to inefficiencies in the system, and you know, upset patients and bad patient experiences. So it's all about the data.

[24:03] Justin Venneri: And last question. Thank you so much for sharing your thoughts on this, Carrie. It's been a fun discussion and I'm looking forward to your answer to this and it's been a while since you've answered this question. It's like a little over a year.  

What's the most astonishing thing you've seen relating to our discussion today that's safe to share, of course? Did AI Mode shock you in any way?

[24:19] Carrie Liken: Yeah, I think it's not related to healthcare, I think. I'll go back to the agents. This is something that I'm really excited about.  

When I see the demos of what agents are able to do, you basically can talk by typing -- let's say Project Mariner is an example of this, Google's Project Mariner -- where you can go to Project Mariner and you can type in find me a Monet print on Etsy. I would like it to have pink and yellow and blue tones. I would love it to be framed, and you can go ahead and purchase it. Just let me know before you go ahead and make that transaction. What that will do is the browser will go to Etsy and it'll also shop around to other places where you could find it, highlight a few options, ping you when the options are available, you look at it, you're like, that one's good to go and it has all of your information, payment, et cetera, to be able to go ahead, process it, buy it and then send it to you. It makes me think, boy, that's so -- it's astonishing, right? It's really exciting. But what could that look like in health care?  

That's what I think is really exciting because there could be a point in time where I don't have to wait for, I don't have to look at my calendar because I could be prompted, oh, you have a primary care. You know your one-year annual appointment is coming up. Would you like me to go ahead and book an appointment for you? And it goes and it does, it puts it on my calendar. Or maybe I received a referral to go to a dermatologist and the agent goes ahead and takes care of that directly from understanding, maybe, reading my email or reading my patient records that I have been referred to Dr. Smith and the agent goes directly to Dr. Smith's agent, finds that information, finds the availability, looks at my calendar and availability, goes ahead and matches it, books it, puts it on my calendar and gives me any of the prep that I need prior to my dermatology appointment. That is mind blowing to me. I don't know when we'll get there, but that is exciting, I would say.

[26:20] Justin Venneri: I think every Compliance officer's head just exploded.

[26:24] Carrie Liken: I actually think I wrote a Substack about this and I received some feedback that said that I'm a little off my rocker, but I'm going to keep going there and just keep thinking about what that could look like.  

Obviously, yes, I think that there are some major limitations to it but just thinking about what it is outside of healthcare and how it could translate to healthcare is just a fun little thought experiment.

[26:43] Justin Venneri: Absolutely. And this has been a fun discussion. Thank you so much, Carrie. I wish you success at your new role at Snowflake, and I look forward to staying in touch.

[26:51] Carrie Liken: Thanks. It was great to be here.  

‍

‍

‍

‍

‍

If you would like to learn more about Capital Rx’s full-service benefit administration solutions, including our clinical programs, CLICK HERE to get in touch with our team.

‍

Want to stay apprised of the latest Capital Rx news? Sign up for our monthly newsletter!

‍

‍