Rae Woods (00:14):
From Advisory Board, we are bringing you a Radio Advisory, your weekly download on how to untangle healthcare's most pressing challenges. My name is Rachel Woods. You can call me Rae. A few weeks ago, we talked about the new financial reality for regional health plans. That conversation focused on the financial reset that plans are actively navigating. They're repricing their business, responding to higher medical costs, and relearning how to generate sustainable margins in a market where frankly, their old growth engines have stalled. But here's the thing, resetting the math only gets you so far. Health plans are facing another challenge. The cost and administrative burden of running a plan just keeps rising. And the old cost-cutting playbook, I'm talking about outsourcing, offshoring, reducing the workforce, automating tiny pieces of repetitive work, they've all reached their limits. And that's why seemingly every health plan is talking about the potential of AI.

(01:16):
But we'll just use AI is not an operating strategy. It doesn't tell leaders where AI can create value, what a plan should build or own, why promising pilots fail to scale, or whether making an old process faster actually gets us closer to true transformation. And that's why this week I've invited vice president of the payer practice within Optum AI, Troy Anderson, and Advisory Board's director of health plan research, Sally Kim. They're going to talk about the mandate for health plan operational excellence and where AI can actually help. Troy, Sally, welcome to Radio Advisory.

Sally Kim (01:55):
Thanks, Rae.

Troy Anderson (01:57):
Yeah, thank you.

Rae Woods (02:01):
You are both in the market every day talking with health plan leaders and talking to them specifically about the challenges that they're facing right now. Sally, can you describe the state of health plan pressure in 2026?

Sally Kim (02:17):
I know in the last episode we talked a lot about the margin pressure and the revenue side, but there's also the admin cost side. There's been a nearly doubling of health plan admin costs in just the past 10 years, largely by high labor costs, seeing more claims come in, increased vendors, which means increased vendor management, and just the regulatory complexity of it all.

Rae Woods (02:43):
But they've always been dealing with administrative costs, right? They've been dealing with it as long as they've existed as plans. And there are a lot of levers that health plans should have access to in order to address those high administrative costs. What are those historic levers and are they going to cut it?

Sally Kim (03:01):
The typical levers are the same as in any business. It was outsource, offshore, or hire more FTEs. That's no longer possible anymore because the low-hanging fruit has been captured, margin pressures are increasing, and all of the claims and service work is becoming more and more complex.

Troy Anderson (03:21):
One of the core processes that I think a lot of plans are focused on, because it just creates a lot of friction and it's in the media all the time, is on prior authorization. And I tried to really transform this process back in 2018, 2019 timeframe using some of these traditional levers. And we were using more traditional machine learning algorithms and just trying to focus on where the labor was spent. So ensuring that non-clinical resources did the non-clinical work and that clinical resources were focused in the clinical space. And we struggled because machine learning algorithms are a black box, so you can't really use those from a regulatory standpoint. And so that didn't work. And I think that's the problem that a lot of plans are running into right now is that the traditional levers can help you with the simple things, but as soon as you start getting into nuance and complexity, which is where most of the cost is now, so we had to rely on human beings to manage that complexity.

Rae Woods (04:16):
And this is where the conversation starts to get into what are the new levers? And there's no way that we can talk about new levers for controlling costs without talking about artificial intelligence. And here's the thing, AI can mean a lot of things. Troy, you were even talking about the older version of AI, the traditional machine learning, the automation, the analytics. When you talk about AI for health plan operations and specifically the work that you do, what do you mean exactly?

Troy Anderson (04:46):
We're definitely talking about the new AI capabilities that weren't available 3, 5, 10 years ago. The challenge for them is to figure out how not just to use them as pilots, but to actually insert them into core workflows and to really reimagine and transform their operations in ways that drastically reduce costs and create a much better experience for the people that we serve.

Rae Woods (05:09):
I want to ask a bullish question. The old tools in the health plan toolbox, the outsourcing, the offshoring, the staffing changes, the basic automation, they worked up to a point, and Sally already made the case that they're not going to work anymore, but why should we really believe that generative AI will work above and beyond the traditional tools?

Sally Kim (05:33):
AI is still just a tool. Sure, it is the hottest new tool. It is faster, it is more scalable, but because it is still just a tool, if the processes inherently are wrong, then you can just be scaling bad processes.

Rae Woods (05:50):
Yeah, you're making a bad process faster, not fixing the process in the first place.

Sally Kim (05:54):
Yeah.

Rae Woods (05:54):
Which maybe actually takes us back to prior authorization, Troy.

Troy Anderson (05:58):
What gives me the most hope that these tools are going to be effective is their ability to deal with nuance and complexity. Our old tools didn't do anything when it comes to that. It's simple, straightforward, deterministic approaches where either you could automate it or you could move it to a different resource. In this case, you can actually reimagine a process where rather than all prior auths be reviewed by nurses and MDs, it actually has the capability of approving these procedures so that you can map the evidence coming in to the criteria and you can get to an answer immediately, which reimagines the process altogether and I think a big reason why so many plans are focused here right now.

Rae Woods (06:40):
You're right that so many plans are interested in these tools. And I want to be careful with my own language in this conversation and not refer to the health plan industry as a monolith. Sally, when you look across the entirety of the health plan sector, how far along would you guess organizations are in their AI implementation journey?

Sally Kim (07:03):
So we asked over 50 plans where they would rate their AI maturity on a scale of one to five, and over 90% said either two, two and a half, or three. They don't want to say that they're a one, but they're also very hesitant to say they're a four or a five because you might not even want to be a pioneer in this space because that could mean that you make a big mistake, a big PR mistake, or worse, a clinical mistake.

Troy Anderson (07:30):
The whole industry and the tools that are available are accelerating so quickly and you can't help but feel behind if you're not spending every waking hour exploring these tools and their capabilities. And so I think everybody feels like there's room to grow. Most plans are doing pilots and they'll be centralized, starting to prove out that they can do tasks, but not actually stringing together and in workflows. Then in the middle, you're starting to put pilots together to get to some scaled end-to-end transformation on workflows, but you maybe don't have the full alignment all the way through the organization top to bottom. I think to get to a five, plans really need to be fully reimagining their workflow, leveraging tools in ways that most plans haven't even started because you're augmenting existing processes versus reinventing from scratch given these new tool sets. And in addition to that, it is not just the act of reimagining, it's you have to have full alignment from the CEO down around how you incentivize, how you organize, how you govern, and how you deliver this systematically throughout your plan.

Rae Woods (08:35):
What's interesting about how you're rating plans is actually quite similar to how I would rate providers. They perceive themselves as the middle of the road. There are a handful of maybe it's the academics or the providers with real national reach that are getting into that four or five. There are a couple that are just dipping their toe in the water on the one. But what I think is interesting is that when I talk to providers, they think the plans are well ahead of them. How do the plans perceive where the providers are?

Troy Anderson (09:02):
Plans actually believe that providers are further ahead than they are in terms of their implementation, use of AI, and getting to scale. And I think they might be right. I think that a lot of plans in terms of setting up the administration of benefits, there's a lot of complex rules and complex environment in which providers have to navigate and they have to navigate across every single plan. And that problem happens to be something that AI is very, very good at. And so the tools that they've started to implement in terms of ambient listening, in terms of coding have helped them to be able to navigate the health plan ecosystem and get proper reimbursement for every one of their visits. And so I do think that the ability of providers to leverage AI is a more natural fit and maybe not as difficult of a scaled problem as the health plans have.

Rae Woods (09:51):
Oh, interesting.

Sally Kim (09:52):
Yeah, especially in the rev cycle versus payment integrity space, that's probably what's driving this AI arms race because both sides feel behind and they feel like they need to catch up. But from the plan side, I hear a lot of, "We're under more scrutiny from regulators than providers are in the AI space." And then from providers, I hear, "We finally have some way of responding to plans after decades and decades of abrasion."

Rae Woods (10:21):
So everyone is interested in AI. They might have inflated expectations about what the other side of the aisle is doing with artificial intelligence. And part of that is because there are just so many use cases that you could adopt. My question for you though is where are you seeing the most activity? Where are the areas that are rising to the top of the list knowing that health plans are focused on reducing administrative costs?

Sally Kim (10:44):
The reason that plans are using AI for admin costs and operational efficiency first is because that's where there's high costs and volume. It's repetitive tasks that are inflectable by AI. It's within the plan's control. There is relatively easier, or I'll at least say faster ROI measurement, and then there's lower clinical risk in this space. Within operations where we're seeing the focus is in payment integrity and member experience because those are the two big operational cost centers.

Rae Woods (11:18):
Payment integrity is an obvious one for me. Member experience is less obvious for the problem that we are focused on in this episode, which is administrative costs. Why are administrative costs high when it comes to the member experience?

Troy Anderson (11:33):
Commercial plans typically have a ton of spend within the member experience in terms of contact center and the app that members have on their phones. And it is one of the few touch points in which a plan can really help their membership, but it's historically been negative.

Rae Woods (11:49):
Yeah, I know. I'm only picking up the phone and calling my health plan if I have a problem.

Troy Anderson (11:54):
That's right. And I'm guessing when you make that call, a lot of times you have difficulty finding the right individual that can help you. And when you do find that individual, you give them your problem and they say, "I'm going to put you on hold for a minute to research and figure out what's going on and then come back and help you with that solution." So AI solutions now, you can speak in plain language what your issue is and it'll navigate you to the right person right there. And then when you're talking to that individual, advanced plans are investing in AI that listens alongside the agent and can surface the right answers while they're talking to the member. And that drives a much more fluid conversation. It also addresses the issue on the first call and without needing to put the phone down to research.

(12:37):
And so the experience that you can drive is one thing, but it's also the cost of the resources that you have put into that contact center and the constant churn of member calls that, if you get after that in a meaningful way, can drive down your overall operating costs as a plan.

Rae Woods (13:46):
If I think about the connective tissue between payment integrity and member experience, these are areas that are higher cost, but they're high cost in part because they're very high volume. If you think about the amount of time that a health plan has to manage those two things, that probably eats up a big proportion of their time and perhaps then their labor, their staff, et cetera. And so clearly ripe for tech-enabled tools to make this high cost, high volume activity more cost-effective, more efficient. And then to your point, Troy, maybe better for the member who's managing through it.

Troy Anderson (14:23):
And the best solution that plans are after is to avoid that call altogether, meaning that you give your membership the tools at their fingertips that can allow them to navigate their plan and answer their own questions without needing to call in.

Rae Woods (14:37):
But connect this back to then where the plans actually are in their AI journey. If most of them are a two or three on a scale of one to five, are there examples of plans that are not just dabbling in deploying AI to address administrative costs in payment integrity and member experience, but really making a difference in their effectiveness, their efficiency and their cost? Who's doing this well?

Troy Anderson (15:03):
You do see a difference between the level of investment that a smaller regional plan can make in something like this and a bigger national plan, because you talked about the volume issue, the capability still costs the same in terms of implementing it and you're going to get more return on investment if you have a lot more volume. And so that dichotomy creates challenge to be an innovator as a smaller plan, but it also creates opportunity to be a fast follower and learn from others and implement the things that are really working well.

Rae Woods (15:33):
I totally hear you, Troy, that the big nationals can do more, spend more when it comes to investing in this kind of tech-enabled innovation. But a few weeks ago, we focused on how the regional health plans were really the ones that are struggling. So how do we square the fact that the nationals might be leading the charge in some of this innovation, but it's the regional plans that are struggling the most with their costs and ultimately with their margin?

Troy Anderson (16:02):
To be an innovator on the leading edge, you've got to spend more and you're going to fail, and there's certain capabilities that are going to succeed and drive an ROI. We can learn from those and work with our partners that are smaller regionals and really right size the type of investment that they're making in order to be right for their plan, in order to give them the most value for what they're implementing. I don't think the regional plans are behind. In some ways they can benefit by being a fast follower and just focus on where there's proven value and make their ROI in that respect.

Sally Kim (16:35):
That's why plans are hesitant to be in that four or five category. Sure, you can be innovative by trying the newest pilots in something that no one has ever dabbled in before, but more realistically, if you're risk averse, like most plans are, you want something that has proven ROI.

Rae Woods (16:54):
Or if you're risk averse, you want to let the fours and fives, maybe that's only a handful of nationals, lead the way, make mistakes, spend their money so that you can be the fast follower and learn what you can do next. Now, that only works if you can actually get ROI when it is time to be a fast follower. So I'm going to ask this question pretty bluntly. How much ROI can plans expect to get from making moves, using artificial intelligence to reduce administrative costs in payment integrity and member experience?

Troy Anderson (17:28):
Generally, our clients are aiming for at least a three to one ROI. And so in order to get there, you need to be selective and you need to have a roadmap and proof points on what gets you there. So that rules out being on the very front edge of just trying to invent new things because you inherently can't really guarantee an ROI in that type of a transformation. We talked a little bit about scale. So one capability that you could invest, say, $10 million to drive a contact center capability. If you have enough call volume, you're going to make that back quickly. If you have a little call volume, it'll take you longer on the same investment to make that back.

(18:08):
And so you need to make decisions based on what's right for your plan, and you also need to work on partners. And this is why a lot of smaller plans may not be interested in totally transforming their own processes that they own, but they're interested in finding partners that can be a managed service for them as well, because that can offer an ROI without needing to actually invest in the change in your own infrastructure.

Sally Kim (18:33):
A mistake I see a lot of plans make is trying out AI pilots and having no idea what the ROI is or is going to be. I think Point32 actually is an example of a plan who does ROI measurement for AI projects really well, where at any point they know how much ROI a project has shown already and will continue to show, and that's all required to even get approval for the project.

Rae Woods (18:58):
And it's also why I think the conversation with ROI comes back to scale. That was something that we heard on the provider side when we spoke to Sutter Health, is that the goal from the jump had to be scale. And I'm hearing you say that on the plan side as well. Let me ask you the opposite question. What's standing in the way of meaningful progress?

Troy Anderson (19:20):
I think the biggest thing is the leadership teams and how they're thinking about AI and how they're organizing to drive scale. And so Sally gave a great example of ROI, the interconnection with finance and really having this not be a project that's off to the side, but part of your core operating principles as an organization that it's not just your chief AI officer that's responsible for this, it's your CEO, your CFO, your COO that need to take accountability, and you should have a set of, say, six to seven priorities that are your areas that you're transforming with AI, that you have a plan all the way through aligned with finance, and you have partners that you feel really good about in terms of driving that.

Sally Kim (20:05):
The challenges I most heard last year were around governance and prioritizing use cases. The ones I'm hearing about now are about data modernization and ROI measurement, and then ones I think I'll hear about more next year are token spend and vendor management.

Troy Anderson (20:23):
One of the other conversations we often have is around foundational investments that enable you to scale. Otherwise, you just work on your data layer for three years, so that connection with the use cases that are delivering true ROI and the foundational investments that get you there that drive the scale, that's a big part of the strategy to move it forward.

Rae Woods (20:43):
And it sounds like don't stall between that proven pilot with the ROI and the scaling phase.

Troy Anderson (20:49):
Yeah, 100%. I think without the right alignment of leadership, it's easy to stall in that phase where you've proven it can do something, but now how do I get the alignment of the organization, the change management, the finance support in order to now scale that throughout? And that's a common theme that you see is that a lot of plans have proven AI can work, but they're just having difficulty truly scaling it and getting ROI.

Rae Woods (21:14):
Sally, Troy, I want to give you a moment to speak directly to our listeners because, frankly, there's just so much buzz around AI. There's so much AI slop out there. So what's one actionable step that you want our health plan listeners to take away from this conversation?

Sally Kim (21:32):
I have one high level and then one very, very specific. The high level is don't feel like you have to push along this AI journey just because you feel behind. And then the very specific one is we have a webinar coming up about AI use in member engagement and a separate one on payment integrity. So please listen in.

Troy Anderson (21:52):
Similarly, I think the theme of knowing you're not behind, the great thing about how fast AI is advancing is that anybody can start and it won't be that long before you're caught up because you're going to be starting on the next gen of technology and you're going to be able to start from where a lot of other people made a lot of mistakes. And so for me, it's realizing that to candidly assess your organization and where you're at in your journey and how your leadership team is showing up, and then really focus on changing that. So focus on discussion with your leadership team, focused on the few use cases that you can really drive a return on and the partners that you want to work with to get there.

Rae Woods (22:34):
Troy, Sally, thanks so much for coming on Radio Advisory.

Troy Anderson (22:37):
Thank you.

Sally Kim (22:39):
Thanks, Rae.

Rae Woods (22:44):
There's a subtle but important takeaway in this conversation that I want to make sure you don't miss, and that's that you are probably not as behind when it comes to AI as you think you are. Most health plans and most providers seem to be in this middle ground, figuring out where they can be a fast follower, where they can learn from the national plans or the academic medical centers and start to generate some ROI. When it comes to operational excellence, focusing on payment integrity and member experience is a good place to start. And remember, as always, we're here to help.

(23:35):
New episodes drop every Tuesday. If you like Radio Advisory, please share it with your networks. Subscribe wherever you get your podcasts and leave a rating and a review. Radio Advisory is a production of Advisory Board. This episode was produced by me, Rae Woods, as well as Chloe Bakst, Atticus Raasch, and Abby Burns. The episode was edited by Katy Anderson with technical support provided by Dan Tayag, Chris Phelps, and Joe Shrum. Additional support was provided by Dominique Del Gaudio. We'll see you next week.