Clover Health Investments, Corp. (CLOV) Earnings Call Transcript & Summary

Good afternoon. I'm Cal Sternick, I'll be hosting today's discussion with Clover Health. I'm pleased to have with me here today Clover's Founder and CEO, Vivek Garipalli; Chief Technology Officer, Andrew Toy; and Vice President of Investor Relations, Derrick Nueman. So Andrew, I think I'll turn it over to you for a presentation, and then we'll move on to Q&A.

Sounds good. Thank you, Cal. Thanks for everyone for joining us today. So my name is Andrew. I'm the President and CTO of Clear Health. And today, what I'll talk to you about a concept that's really important to us, that is central to our entire philosophy, which is the idea of the software-defined physician. It's an approach that we're bringing to health care, where we truly believe that there's a gap between what software can do in other industries and what it's done for health care. But in particular, what it's done to enable physicians. There's just so much opportunity here if we can move to a world where software and physicians come together and we are truly enabling better care at the site of care. And putting incredible power at the fingertips of all of the clinicians who are working so hard to serve each and every one of us. So next slide, please. Our standard disclaimer. You can read that if you like. Next slide. As I said, I'm here today, and Vivek is here as well and will be joining us for the Q&A session. Next slide. So many of you and many of us have seen this quote from Marc Andreessen from a few years ago, that software is eating the world, right? And unfortunately, I believe that health care has given it a little bit of a case of indigestion. And what I mean by that is the potential for software to really help us all in our day-to-day lives is really just not being fulfilled. If you think about how much you use technology in your personal life, each and every day. And think about how much you use software in your life on your phone, on your desktop, all of us are doing this conference virtually right now. The boundary of what is enabled by a focus on technology and software. When it empowers the human being is well ahead of what we see in health care and certainly what we see when it comes to actually powering the decisions of a physician and clinician. We've all been in situations where you've sat in front of your PCP or you're talking to a doctor. And most of us will have great experiences overall with our clinician. Trust of our clinician is really high. But at the end of the day, you can see them struggle to actually use the tools that are provided to them by the health system in terms of their EHR, the scheduling system. There's a disparate number of tools. I was once in a health care appointment that I had for myself recently where I could hear people in the corridor walking up and down yelling what browser do I use to open up the EHR. Is it meant to be Edge? Is it meant to be Internet Explorer? And this was at a very high-performing academic medical institution. And all of these things just get in the way of what clinicians and what physicians want to do, which is deliver amazing care and have data at their fingertips, protocols at their fingertips, we truly just had to be assisted as opposed to a burden as opposed to something that is forced upon them. So if you go to the next slide. What we see is at Clover, we want to move to a world where we have software-defined physicians. So what do we mean by software defined? Well, if you think about other areas of technology, we have things like software-defined networking now, right? Where instead of hard wiring all of our switches, hard wiring all of our capabilities in a network center. We are moving that into firmware, into software, so we can actually improve the performance of networks, improve routing on networks, improve throughput and data flow in our data centers, all dynamically all through updating software. That's what the revolution that software-defined networking brought. I would argue to everybody watching this today that something like a Tesla and what they brought with their Model S was more than just electric cars, but it brought the idea of software-defined vehicles, right? Like when was the idea that you could actually receive an update overnight for your Tesla. And suddenly, you have more horsepower. Your 0 to 60 time got better, your battery got more efficient. It's not just the experience you had in the cabin, but actual new capabilities are released to existing vehicles, existing infrastructure through the release of new software, right? And so when we move to a software-defined world, what we see is, it truly changes the relationship between someone who is engaging with something that they're very familiar with, whether it be connecting to your network or whether like get to the Internet. Whether it be to driving your car and suddenly it can change, right? You received an update to your phone and suddenly your phone can do new things that it couldn't do before. And so that is what software definition really enables. And so what if we could move to a world where we have the software-defined physician? And that doesn't mean that we're going to rob the physician of agency, actually far from it. What it means is that we need to empower physicians and see software as an additive capability that we bring to them to sort data, make it efficient for them to make decisions. We hear feedback from them and build that into the future versions of the software. It's about iteration rate, it's about patch rate, release rate. All of these things have to come together in order to make something software defined. So we believe that we're well on the way, if not having achieved this at Clover with Clover Assistant. When we give Clover Assistant out to our physicians on a wide network, remember, not employed physicians. Our whole model is that we don't need obscure abstract contracting to influence physicians. We don't need to provide payment models that are confusing or create moral hazards. We can bring together payment, we can bring together value-based care, and we can bring together software and package it all together and make it very easy for any physician to engage with the system. And we can measure, learn from what's going on with all of these physicians in the field, make updates to our software. And because we can harness the power of the cloud and modern software techniques, update Clover Assistant on a regular basis to make sure that the latest protocols, the latest data, the latest feedback is all integrated to make the physicians the absolute best version of themselves that they can be. We can enhance the quality of care. We can make sure that we put the latest data at their fingertips, and we are continually releasing, continually learning in a way that only the cloud-based software and high engaged software enables us to do. And so this is really the role of Clover Assistant, is to bring about the era of the software-defined position and bring all the benefits to health care that, that entails. Next slide, please. So Clover Assistant, what is it really doing? The key thing is it's being used every day now by PCPs at the point of care, engaging with the data set that Clover has, where we bring together a number of different facets of health care data. You can see that on the left-hand side there, we synthesize all of that data for the PCP. So we go out, fetch it, clean it, synthesize it, load it into our data warehouse so that they don't have to. Then we make sure that we pump that through our expert system where we apply rules and machine learning to generate insights about what the most important things a physician can actually do at the particular moment are. And then we really -- most importantly, we translate that to actual action by the physician. So none of this is important in terms of how much data you have or what insights you can run with your data scientists if it's not being used every day and engage with every day by real-world physicians. And not just physicians that are in your own group, we have to bring this capability to every physician that's out there. That's why Clover is focused on bringing Clover Assistant to every physician in the country because we're not just about saying, are you in our network, we're not just saying do we have a capitated value-based contract with you, are you employed by us or do we own you. What we're saying is we want to bring this capability to everyone because great care means great outcomes lowered cost per society and better results. So as we drive more actions on Clover Assistant, that's how we measure ourselves, right? What -- how many people are we helping with Clover Assistant, how many physicians, how many patients, how many actions are we driving every day. And then as we drive those actions, we can fold that back into our feedback loop, learn continuously, deploy continuously, improve continuously, right? It's not about just how much data you have, it's not just about how many physicians you have, it is about that rate of iteration that drives great software. It's about that fact that you have to be continuously learning, whether it's within your ML models, whether it's with our product managers, whether it's through your feedback cycle with the clinicians and bringing that model to everybody. Next slide, please. And so you don't have to -- look, I'm not going to go through every single one of these boxes. But I've talked about the fact that we are synthesizing data on a longitudinal data platform, and we've built a large stack to make sure that we can do that. Data in health care is not clean, right? Like everyone knows this. It's one of the stumbling blocks for the larger tech companies is how do we make sure that we connect to, and make -- have interoperability, pull together and have a central view of data, one of the major challenges that we have there. Next is how do we create the right insights and connect back to the care management protocols of the clinician. The many, many different forms of managed care that exists out there, like Medicare or in Medicaid and all those different variants. How do we tie all that together to make sure that we are producing the right insights for clinicians. And that's a combination of rule-based learning and machine learning that we brought together into an expert system, large amount of investment into making sure that's right. And finally, I already talked about this, provider engagement and making sure that we are having a lot of usage, making sure that our customers and our providers are delighted and when they use the software and that they are actually engaging with all of those insights that we are servicing. These are the key parts of Clover Assistant. We've built all of this in the cloud, it's cloud native. It is also running for the same version for all of our clinicians across multiple, dozens and dozens and dozens of health systems, are all running the same version of a cloud-based Clover Assistant. We think that alone is a huge achievement that when we release code and we've released, I believe, about 10 significant feature packs in the last 6 months, not counting bug fixes and operational fixes. We are constantly releasing this across our infrastructure and having everyone run on the same version. That alone we're very proud of and is part of our fundamental view of how we build software to enable physicians on a day-to-day basis. It's how we bring scale, it's how we bring it to the wide network, it's how we bootstrap new provider groups extremely quickly and bring the model to as many people and many clinicians as possible. Next slide, please. So we truly see this as driving health care efficiency. So speaking in terms of the medical care ratio, the medical loss ratio, which actually influences like how much money is going into health care and how much is being spent. What we see is that when we look at year 1, year 2, year 3, this is speaking to how long a clinician has been using Clover Assistant. So the first year, we're seeing that there's already a significant improvement in that MCR at around 13%. And this is in the last couple of years. These were measured during the pandemic, but around 13%, which is very exciting. We also see that, that improves year-over-year. So the second year a clinician is using Clover Assistant, that goes up to a 17% improvement in the MCR and even up to a 20% improvement in the MCR in year 3. Remember, this is measured with the clinician being at the center of the analysis. So we're saying when a physician goes on to Clover Assistant, the longer they've used it, the more they're using it for their care planning. And the more their panel has been managed by Clover Assistant, the better the improvement in the medical care ratio. And I think that, that is really what we would expect to see, right? Because once we get -- reduce the variability in care, if we're improving evidence-based care planning, if we're using data and putting at the fingertips of doctors, if they're synthesizing that in their heads and making care decisions on the basis of what's being suggested, we would expect that, that would result in better outcomes. And we would expect that over time, the results of that outcome versus control would actually improve, right? Because you wouldn't see all of that benefit that first year. You would see increasing benefit of having being on evidence-based care, the longer you are using the software, and that is exactly what we are seeing when we do our analysis. Next slide, please. And so Clover Assistant usage is growing fast, right? We are consistently growing scale. On the left, you can see live Clover Assistant providers by NPI, which means individual clinicians, signing on and using the platform and growing over time. As a reminder, this came out of beta in mid-2018, and we've been growing through -- after we came out of beta and throughout the pandemic. And we see more and more platform usage proportionate to the number of clinicians that are actually coming on to the platform, which is once again what we want to see, right? You can see a bit of a spike there in utilization on the right-hand side, which is when we brought fee for service under the management of Clover Assistant via direct contracting. That resulted in a significant increase their utilization as we would expect, lots of lives coming on at the start of last year. We will probably see that again this year as the next class of direct contracting for the clinicians come online, and we see additional lives come under the management of Clover Assistant. Very exciting. So we're growing scale across all of Medicare on both the fee-for-service side and on the Medicare Advantage side. As we grow, we constantly measure how many lives are under Clover Assistant management, how many physicians are being empowered and being defined by the utilization of Clover Assistant. And as a reminder, this is our way of influencing and affecting how care is given every day. We've brought together payment schemes. So it's -- we have simple, easy-to-understand contracting, which enables more and more physicians to sign on to use the software. We have simple, easy-to-use software. We're constantly releasing new updates. And it's very straightforward for clinicians to give care. They can use their clinical judgment. They interact with their data. We are not compelling them to agree with the software. It's all there to help them make better decisions on a day-to-day basis, there is nothing we're doing to compel any given outcome, which is why I think -- honestly, I think that's the way physicians want to work, it's the way they want to interact with software. They don't want to be forced to do documentation. They don't want to be forced into one-size-fits-all population health management protocols. They want to give personalized care. They want their software to be helping them be the best form of themselves, and that's what we're bringing to them with Clover Assistant. Next slide, please. And so it's growing rapidly. Like I said, this is one of the key KPIs that we're measuring, which lies on the Clover Assistant management. Things are growing rapidly. We are moving quickly into fee for service. We have grown significantly on the Medicare Advantage side. And we're bringing more and more physicians as the slide -- previous slide showed onto the platform on a day-to-day basis. So more physicians using it, more lives under management for Clover Assistant and releases going out every couple of weeks, making the product better and better and better, both in terms of clinical and physician quality of life as well as improving the protocols. And the data we're showing, the kind of clinical recommendations we're putting in there, the kind of ML models we're shipping into Clover Assistant and which physicians are engaging with every day. Next slide, please. So we really do believe that the software-defined physician is a key part of where we as a society need to go in order to fix health care, right? We don't have enough clinicians as it is. We need to make the physicians we have who are incredibly hard-working. We need to -- we can't ask them to work harder. We cannot be throwing physicians out of network and using contracting games to solve health care. What we need to do is empower our physicians, make them better and give them data. Let them use their clinical judgment, give them superpowers, right, and just focus on physician enablement and bring that to every single doctor in the United States. That's how we get more scale out of our health system. That's how we squeeze up inefficiency. That's how we make physicians' lives better, which is really, really important. That will generate value for providers, right, as we actually help them move towards value-based contracts, but in a much simpler way from a contracting perspective and from a much -- more straightforward way from a physician perspective. Value for patients as we empower their physicians, they can know their physicians have the latest protocols at their fingertips. The latest complete data at their fingertips, and not worry all the time that they have to be the quarterback of their care, their physician and their PCP is the quarterback of care. And then value for payers like the government or like private payers by lowering cost, improving outcomes and reducing variability of care over as broad a network as possible to enable payers to work with as many physicians as possible, right? Because we want to help as many people in the world as we can. So wider networks, lower costs, better outcomes by widening those networks and moving to software as the control plane versus contracting and payments as the control plan. It has to be software-centric is our view to move to that future in that future world. So we're seeing enormous excitement by this. We started in MA. We brought the model to the fee-for-service population and our provider partners. And we have tremendous interest from lots of providers talking about how we can bring this model of the software-defined physician and bring it to other areas of care, other payers, other models of care. And we think the model does scale. At the end of the day, the common variable is that physician, is that primary care physician, across all of these health care constructs. And by investing in that software that then makes that physician better and give them superpowers. We think that will always generate an incredible ROI. So that's our model. That's where we're -- that's how we're thinking. That's how Clover Assistant, we believe, is the core of what we do and the progress to our future. And with that, I'm happy to move towards Q&A. And I'll toss it back to Cal.

Yes. Thanks, Andrew. That was great. And now we're bringing in Vivek as well. So I guess maybe just to start first on a big picture question for Clover Health. There are a lot of physician support tools in the market today. So can you maybe talk about why providers utilize or choose to go with Clover Assistant over some of the other options that exist?

Yes, I can talk about that. There's lots of different reasons. But I think that at a very high level, we are building this for physicians. It's not necessarily that we're building this for health systems. So when we say provider, as you know Cal, you could mean like health systems or you could mean physicians, we truly build this as a physician-enablement tool. And we just don't see the physician being involved enough in the decisions of their health system IT department, right? Like it's very rare. They're even brought in, in EHR decisions, even clinical decision thinking. It's often the health system who's doing that, and we're bringing that all together. As I said, we can bring in better outcomes for the health system in terms of alignment to value-based programs and make it easier for them to participate in value-based care by focusing on the physicians' needs so they give better data-driven care. And that results in higher engagement, better satisfaction from the physician, which results in better care, I think, for the members. So that emphasis on the physicians themselves and not just the health system, and the IT department of the health system, I think it's a little bit different and is the core to our thinking.

And when we think about great software, very rarely is great software developed for your own employees. And so when you think about where technology has been generally focused on being built, it's provider controlled organizations, where they control employment and some sort of direct or indirect way. So you have individuals who don't have free will in their decisions as to whether to use software or not. So I think one very important thing to ensure we delineate with Clover is everyone is using, or nearly everyone is using Clover Assistant is not employed directly or indirectly with Clover. And so they have the decision at any point in time or the ability to stop using our platform. So when you've got a counterparty that can make a decision to stop using it, all of a sudden the nature of the value your product is bringing, the usability and the ability to keep improving upon it because that's an expectation that your customer is going to have, all of a sudden, that's a developing software. That's not an understood concept in health care. But when you get out of health care and you think about industries where software is deployed individuals who have the ability to choose not to use it, that's a very logical and already well understood concept.

Got it. So is it possible to license Clover system to others? And have you had any conversations in that regard?

Yes. So we've had a lot of conversations because it's a very natural motion to increase the surface area once we land within a provider to bring Clover Assistant to other areas. So lots of conversations we're having. I think there's interesting models we could develop where we truly want the health system to win. We want the physicians to win and the patients to win. And as well as society, right, to lower cost of health care. So lots of discussions that we're having about where we go with CA.

Should we think of that as like more of a near-term opportunity? Is that a longer-term opportunity? What's the right way to frame that?

I mean I think the thing to think about just even in our historical trajectory, so if we go back to 2020, 100% of our business was Clover Medicare Advantage. Last year, it was about 50%. This year, it might be around 1/3. So 2/3 of our business today is pure software sales. We've chosen as our monetization strategy, taking a significant part of the savings that we generate for our clients. But that's a business decision. So you go back over 2 decades ago, when Google launched their search engine, they made the decision for their monetization strategy to the ad revenue. They didn't have to choose that path. That's how they felt they could drive, one, the most utilization and create the most value on their back end when you think about insight generation. And creating more personalized suggestions on searches, but it also allowed them to attack the biggest part of the TAM in their mind, which was ads. From our perspective, we feel the biggest opportunity is to correlate great decisions with savings that we can generate. So for us, that's our ability to capture a huge part of the -- of what we think is a future TAM that we're essentially creating a new category around. As we go into future years, as Andrew referenced, we're getting really, really exciting inbound interest on extending Clover Assistant to payers that aren't Clover, that aren't fee-for-service. And that's something we're just going to be really thoughtful around before just jumping into it.

Okay. And I wanted to touch a little bit on the slide for the MCRs versus -- for Clover -- PCPs on Clover System versus those who aren't. Because that was a pretty impressive differential there. So I mean, I guess, first, you have about 70% of your total lives right now under CA management. Have you given a target for how many lives do you expect to be under Clover Assistant management long term? And just help us think about the membership progression over time.

Yes, absolutely. So starting there, as a reminder to everyone, so about 100% of our fee-for-service lives are under Clover System management because the direct contracting program we bring it together with CA at the doctors are one and the same. They use CA and they -- and those are the lives that can be attributed to our direct contracting entity. On the MA side, on average, around 60% of our members are going to see a Clover Assistant doctor in any given year. All the remaining, 40, about 15% of them do not have a PCP at all, to be clear. And that tends to be younger, healthier folks who do not have chronic conditions, who don't just don't have a regular PCP yet. So all the remainder, like we are actively working to always be signing up more and more doctors. We're growing a lot every year. So it's painting a bridge, right? You never stop signing up new CA docs, but we're always excited to drive that number up into the right.

So again, on the MCR differential there. I mean, you have a big portion of your MA members there. You had a 13-point differential, but the consolidated MLR is still -- I am thinking for 2022, you're talking about the upper 90% range. So how do you get that down over time? And with the gap that large, at least for MA, I would think those numbers are probably closer to profitable year 1. Can you just give us a sense for what membership profitability in your MA book looks like year 1? And then how we should think about that tracking year 2, 3, 4, 5 and beyond?

Yes. I think there's 1 thing also worth going over. Derrick, may be just put up Slide 17 for a second.

Sure. Give me second.

Yes. So just to give a sense as to where we're at this year versus the prior year and the year before. So you'll see on the left, what we were experiencing last year was a reversion back to the mean on the medical expense side. But -- and that's a pretty significant thing as we go into this year. Obviously, there's going to be continued effect from COVID that we're all aware of. But on the right side, is straight out of our MLR file. So when we had a significant drop in office visit volume across the country in our markets included, that had a corresponding dramatic drop in risk scores. And so there is a bounce back given office visit volume coming back last year. And we see that in our first MLR report on our risk score for 2022. As when we think about MLR, it's 2 sides of the equation. So it's the MedEx and the premium side. So I think that's something that we think will come through, obviously, this year. As we think about into 2023, there's -- we have a lot of discretion in how we think about pricing and plan design. So as an example, COVID had a big impact last year. It will have a meaningful impact this year. At some point, CMS will have to make a decision. Do they create some sort of accommodation for COVID on a permanent basis. And if not, ourselves and other plans will have to make adjustments around that, and that's a decision that we control. At the same time that it's probably not well known, but New Jersey, which -- where today, we have about 80% of our members or so. New Jersey ranks #1 in the United States for most COVID hospitalizations on a per capita basis. And that's based upon a data as recent as January 6 of this year, going all the way back to last year. So there's -- it's not really a secret that our kind of core market has been impacted the most. So at the same time, our view is making sure we focus on the normalized view of our business but not making major reactions or assumptions in the short term.

Right. So if I think about some other companies have given data on how profitability has tracked by members cohort, right? And you guys have this 20-point difference in MCR for members who have been with you 3 years. Do you have any data on how profitability tracks by member cohort? Is it fair to say that the members you've had for over 3 years on CA are EBITDA positive? Just like any context you could sort of provide to give a sense for how profitability is tracking by cohort.

Yes. So we actually track it. And the way we track it, we're very physician-centric, just to be really clear. So we actually track it by clinicians, right, and the panel that the clinician is managing and how long they've been on CA. So the reason for that is with our pure Medicare population, as they get older, it's natural for to them to be sicker, et cetera, et cetera. So there's a lot of bias in there. So what we've seen is in our core markets, and we shared this a few -- a couple of earnings ago, I can't remember which one. But in our core markets, if we take away the effects of COVID because COVID has raised MedEx across the board, like Vivek's talked about. But if we normalize for COVID, we feel very good about where the MCR is for those who have been on with -- seeing a Clover Assistant doctor, seeing it even for -- with even year 1, but certainly years 2 and 3 for CA doctors. Where we have -- in those markets where we have very high Clover Assistant coverage. We're very pleased about where we are from a profitability and sort of like margin perspective. Now that's clouded, I think, by the overall rise in MedEx from COVID, right? That's on a normalized basis. Like on a GAAP basis, that includes because it's fair, right, all of the hospitalizations from the pandemic. So once -- we all want the pandemic to be over for many reasons. But I think once that clears up, people will see that flow through all the way into the actuals, and I think that will be very helpful. But in our data, when I look at Clover Assistant, when I look at the control for the pandemic, I feel good about those markets, and I feel good about those doctors who have been on CA.

And the other important thing to remember is because we've been growing really, really rapidly over the last many years, a lot of the share that we get isn't just from fee for service over to Clover, but from competitor MA plans to Clover. So about half of our market share is coming from taking share from competitor plans. So just kind of one thing -- Derrick, if you can just pull up the slide again, the New Jersey slide. But one -- the thing that I think is key when you're asking about cohorts. So if you look at kind of this slide, this is our growth in New Jersey and comparing us to United over the years. United stayed fairly static in the grand scheme of things that we've taken significant share here, and we have a 2022 estimate in here. What's really important about it, from a cohort data perspective, that delta that you see on Clover Assistant value, half of those lives have migrated from competitors. And so that's off of their baseline. So whether they've been with a competitor plan for 1, 3 or 5 years, they're now entering Clover MA. And we're creating that incremental value from an MLR perspective via Clover Assistant. And so that's all net found value driven by Clover Assistant, and not having been accomplished by local competitors. And so when you think about then cohorting on an MLR basis, MLR in terms of the revenue side is very much a function of plan design. So we want to lower our MLR down to low 90s or 80s as an example. We have the ability to do that via plan design. But importantly, said another way, if competitors in our markets attempted to compete against Clover on the same plan design, their apples-to-apples MLR would be 1,000 basis point plus higher than ours. And that's something that's very, very important to understand when we think about the power of the Clover Assistant.

Okay. Great. So maybe switching gears a bit. You guys put out a very positive update on MA membership guidance yesterday. Can you maybe talk a little bit about your revised expectations for the full year? And maybe perhaps some of the competitive dynamics that you're seeing in your markets? There were some concerns from last week about pricing, margins. So just curious what you guys are seeing in the business.

Truly, yes. So we have our earnings coming up in a month. So we'll save some content for there as well. So I won't give any revision, but we weren't obviously very happy with where we are sitting. We won't get the full numbers from CMS. As you know, kind of like -- that's kind of like a cadence to what CMS shares in terms of the absolute number. But we do believe that on the fee-for-service side as well on the MA side that we are very comfortable with the rate at which we are bringing more lives and medics under our management and particularly to Clover Assistant management, right? So I think that looks good, and we'll have more to share when CMS finalizes their numbers, which is typically near the end of January, and we'll have it for February earnings. The other thing I'll say to your point, though, is that what we're seeing very much about the market dynamics is we believe that the wide network and the PPO is where the growth is in Medicare. That's on the fee-for-service side, wide network by definition. And also the PPO side, wide network is our specialty because we managed care on the wide network with Clover Assistant, right, because we can bring that model to any provider. That is what people want. People want choice in their physician. Like I don't know about yourself, like for me, like I always -- for my own health care, I take the PPO. I rarely pick the HMO, right? It's just built in that people like of choice. As Americans, they like choice. And we're seeing, I believe that the HMO side, the narrow network side cap out as a model. And that's resulting in a lot of sort of zero-sum exchange between those in any given geography who have the narrow network. It's either competition for the most lucrative value-based contracts for certain big providers or it's competition for the same lives who are in the market for that narrow network, sort of HMO. The PPO is appealing to a much broader set of folks. I think that's what drove a lot of our growth. It's core to who we are. And we see that on the fee-for-service side and the MA side. The other thing I'll quickly touch on because it's in the [indiscernible], is that our exposure to the electronic brokers like the eHealth, et cetera, is significantly been limited. Like we talked about this last year a bit. Like we have -- we are available on those brokerage channels, but we have never gone heavily into those channels. We've always been sort of like word-of-mouth feet of the ground broker oriented. And because of the lack of exposure to those channels, I think it's been better because those channels did not do as well, we believe, in the last season, right? They've resulted in higher churn, lower LTV, higher TAC at this point. And that's been difficult for plans to have reliant on them for growth to prop up those narrow network plans. And so I think that's been challenging.

Okay. And we have a couple of minutes left here. I wanted to ask 1 or 2 on Direct Contracting. So your direct contracting membership projected more than double next year, and Direct Contracting is going to comprise the majority of your 2022 revenue. So the program is operating at a loss today, but your guidance implies a pretty significant MCR improvement next year. So just wondering if you can talk about your expectations for Direct Contracting performance next year and what some of the key MCR drivers are going to be?

Yes, absolutely. So in Direct Contracting, a lot of mix under management, 100% CA coverage, which -- by definition, which is really exciting to us, right? So on one dimension, it is continually driving utilization, very good engagement already by physicians, even just on Direct Contracting, not DC NMA, but just DCE. So day to day, driving utilization of CA and we're getting tons of data points and feedback from physicians on improving the product, that program via that. So just as a technologist, I love the engagement that it drives with our software and our software service from hype from real-world primary care physicians. Second of all, we'll have more to share. Because right now, the benchmarks on the performance standpoint, the benchmarks do not include COVID effects, right? They were all -- the benchmarks were measured during non-COVID years. CMS and CMMI has guided that they will provide an adjuster for COVID, which seems reasonable, right, because costs obviously have increased during the COVID years. But those adjusters haven't really been finalized yet, which I think has resulted in some of the confusion in the marketplace. So I think that will clean up soon, certainly well ahead of the first reconciliation, which comes middle of this year. And then we'll have more clarity to share with the audience. Before that, what I can say is, is that because it's driving a very high-quality engagement on the CA side, I do believe we'll be able to adapt to whatever we need to in terms of making sure we have great care delivered by PCPs. We are empowering those PCPs, very good engagement. That gives me high confidence that we will have a great program on the fee-for-service side.

Great. And so that brings us right at the time. Andrew, Vivek, thank you guys very much for taking the time this evening. Really appreciate it, and we look forward to hearing from you soon.

Thanks, Calvin.