Lemonade, Inc. (LMND) Earnings Call Transcript & Summary

Hi. I'm Heath Terry. I cover the Internet sector for Goldman Sachs. Really happy to have with us today the CEO of Lemonade, Daniel Schreiber. Daniel, thanks so much for taking the time to join us. I'm also joined by my colleague, Yaron Kinar.

Daniel, maybe just to start for the audience that we have here today that are maybe less familiar with Lemonade or maybe most familiar with Lemonade as a customer of your product, how should they think about what Lemonade is as a company and what you and the team there are trying to build?

Sure. Good morning, Heath. Great to be with you both. Thanks for having me at the conference. So Lemonade is a young insurance company that's using technology to really rethink some of the fundamentals of the sector. Both myself and my co-founder come from 20 years in tech and are newcomers. I guess by now, less newcomers. But when we started Lemonade, we were newcomers to the world of insurance, animated by the premise that insurance may be the most disruptible industry on the planet. And so we set out to create a new kind of insurance company. Today, we offer renters and homeowners and house insurance. We've announced that we'll soon be adding life insurance. We aspire to offer all the forms of insurance that our customers need. And it's really built from scratch on a digital substrate so that you replace paperwork and bots -- and brokers with bots with the promise of everything being instantaneous and kind of zero hassle. We also took a run and we think we summed up the elements of the business model. So Lemonade is a public benefit corporation and a certified B corp, trying to address some of the animosity and distrust that really plagues the industry. And maybe I'll delay for a minute longer on this business model element. It is self-evident to the point of kind of being by now to say that insurance companies' profitability depends on claims, but that actually has a bunch of perhaps less obvious side effects. One is that customers see this oftentimes as a zero-sum game, some kind of conflicted relationship because if you deny my claim, you have more money. If you pay my claim, I have more money. But it also means that while top line results are predictable, bottom line results are variable. And it also leads to being capital-intensive because that variability requires you to keep a lot of money aside for the rainy day. So we've restructured some of those fundamentals through reinsurance on a charitable giveback program to neutralize some of those conflicts, to bring predictability to bottom as well as top line and to make us capital-light rather than capital intensive. So rethinking some of the plumbing as well.

And so with your most recent results, you've effectively tripled your customer base from where you were just a couple of years ago and reached over $200 in premium per customer for the first time in history. What would you say have been the biggest drivers behind the underlying strength in the business?

Well, we've taken a run at quite a few things, but I think the common denominator really is customer centricity. That may sound self-evident, but in the world of insurance, it's a difficult trick to pull off. Insurance companies are typically regulator-centric or capital-centric or policy-centric, seldom if ever, customer-centric. And that gave an opportunity for us to really rethink the user experience, the policy language, the very business model in a way that we think about what is right for the consumer. And in the course of those same years, we've climbed from being not familiar to anybody to being rated #1 by consumers across the United States in terms of their level of satisfaction. And this is both -- if you look at kind of new fangled ways of assessing companies like [indiscernible] where we score 70 to 80, which is kind of where Apple and Tesla is rather than where insurance is, or if you look at the Apple App Store and see a 4.9-something-something reviews -- star reviews out of 5 for Lemonade. Or indeed, if you go to more traditional measures, J.D. Power and Associates just ranked Lemonade #1 insurance company in America out of thousands reviewed in the sector that we play. So we've really been very focused on this, manically focused on how to bring technology and behavioral economics to bear on insurance, how to create a seamless experience. And at the end of the day, if you can pick up your phone, and in 90 seconds, buy a policy through a playful -- a chat with a bot, which is the median time to buy a policy. If you can get a claim paid in 3 seconds through the same interface and just by holding up the camera and speaking into the video and talking about what happened, you're going to start seeing word-of-mouth spread, the brand catch on, and people start tweeting about it. And they really tweet about how affordable it is, the aligned interest in the charitable dimensions and the experience which is seconds and delightful. So those 3 elements interplay in a way that has led to exactly what you're talking about.

That's great. What would you say the typical profile of a Lemonade customer is? And I think maybe most interestingly, how are you seeing that profile change as you expand from renters into homeowners, condo owners and even pet more recently?

Yes. So pet is very recent. It was launched in July, but condos and home and -- well, condos and home have been with us since day 1. It's just the mix has been changing. Maybe we'll talk about that in a second. The profile of our customer is either obvious or surprising, depending on your perspective. So we do tend to skew younger than the industry at large by some measure, something like 75% of our customers are under the age of 35. And that may not be surprising because it's a digital first experience, although we do think that instant and affordable and accountable is something that has appeal at all age groups. The more surprising aspect is about 90% of our customers still tell us that they're not switching from another insurance carrier. And that is very unusual in the landscape of the insurance industry because if you watch more than 5 minutes of TV, you'll hear a TV commercial hitting you with the I switched and I saved value proposition. And that's really how the whole industry operates for obvious reasons. And it turns out that at Lemonade, we're not -- by and large, we're not playing that game. We're not trying to take customers from other insurance companies. We're really competing with nonconsumption. We're getting the new cohorts of buyers where we have a very dominant market share, and we're then growing with them. So we are seeing younger consumers and first-time buyers as really the anchor of our book of business. And then it changes largely, Heath, because those people mature. So if you just look at the Federal Reserve data, you'll see that in America, the median net worth of somebody under the age of 35 is $11,000, and that over the next 10 to 15 years, the median net worth will grow 10 to 15x. So people do accumulate wealth. And insurance really does track to that, right, because what you have that comes with responsibilities and things you want to protect, so as people move from being a renter to being a homeowner, for example, you'll see multifold, oftentimes 6 or more fold increases in premiums. And what we're seeing more than anything else, I think, is just our cohorts beginning that maturation process. And as people hit the age of about 33, you start seeing a lot of things happening whereby they oftentimes start getting married, having kids, that can trigger a life insurance, moving into a home that can trigger home insurance. And we're really seeing that. So it's not so much that we're attracting different customers as our customers are going through predictable life cycle events and we're seeing the benefit of that.

And so as we step back into the core renters business, you've seen this continued improvement in the efficiency of customer acquisition costs and premium paid for retained renter over time. What are some of the common policy upgrades renters make after their first year with Lemonade? And what are some of the things that you're doing to improve retention?

Sure. And it's already the first year that's a trigger from the consumers' perspective, although the underlying policy is an annual policy. The experience is very much a monthly subscription service, just like your Spotify or your Netflix. So there's no big event at the end of the year per se. And we do indeed see these kind of upsells and upgrades happening throughout the year in a fairly evenly distributed pattern. So a consumer will buy whatever they need, and then we're coming up on Christmas, they'll get something nice for Christmas. Somebody buys them a laptop or a bike or a new camera or a piece of jewelry. And they'll go to the app, and it's such a simple thing to just add a bit of coverage, so they'll add whatever they need for that. And we have seen that over the first 3 years, if you look at our cohort -- 3-year-old cohorts who remained renters, some become homeowners. But if you just look at renters, to your question, the average premium that they'll pay us after 3 years will have grown by about 50%, actually more than 50% over 3 years. So these natural progressions of accumulation of assets, which then reflects itself as accumulation of insurance needs, is something that is very strongly manifested in our data. You add on top of that basic kind of conveyor belts that people are climbing, the fact that sometimes the step function changes, like they'll have bought their first home, they move into a condo, they have got a pet. Pet insurance is a pretty pricey policy. And so you can see a big jump. So we see something like 6x increase, 600% increase for 1 graduation, 4x increase when people add a pet. We're about to launch life, and that we're estimating, I think, about a 5x increase. And these are unusual. Usually, when you think about a subscription business like Spotify or Netflix, where people are paying $5, $10, $15 a month, an upgrade will look -- an upsell will look like another 10%, another 20%. To be able to talk about hundreds and even thousands of percentages perhaps unique to insurance.

Yaron?

So I guess one question I had, as we're looking into growth into the homeowners market is I think the company today has a concentration in, I think, the Texas region, New York region, California region, which are regions that have fairly significant capacity exposure, right? So as long as you're in the renters and condo market, I would think that exposure isn't that much at play, if you will. And I think we saw it in the third quarter results as well where cat losses were quite manageable. But as you look to convert those customers into homeowners policies, how do you address the catastrophe risk that comes with that?

Yes. So it's something we're quite mindful of. And we have been growing fast, but we're trying to grow responsibly. And you see that, as you intimated, our loss ratio has declined precipitously, even as our growth rates have been pretty unusual. So we are walking that rather unorthodox combination of rapid growth, while rapid -- combined with rapidly improving underwriting results. Specifically with regard to homeowners, really, since the get-go, we've been fairly conservative in our underwriting. So we have refused to quote or declined to quote more homeowners than we actually quote. We could have grown our business dramatically faster if we weren't also being quite cautious about the kinds of risk that we load up on. And you're absolutely right that Q3 was the worst wildfire season in California history, and there were more hurricanes in that quarter than there have been ever before. So that was a good litmus test, or a test case, if you like, for how are we doing in that regard. Because even though we have more renters than homeowners, about 1/3 of our sales are in dollar terms come from homeowners. So it's a considerable part of our business. And if we were not being careful, you'd see that hit us hard. And we got hit disproportionately lightly compared to the rest of the industry. And indeed, we had profitable results in terms of our losses in Q3, which was this perfect storm, no pun intended, it's a perfect storm kind of test. So just to give you a couple of examples. In -- California has wildfire exposure. And you can write a lot of homeowners in California, if you're not careful about that. The zone, the fire zones are ranked typically from 1 to 30, 30 being the most hazardous. You're kind of building a house in a fire. We will write homes in zones 1, 2 and 3 and then we stop. So we do that. We do the same with flood zones and hurricane-exposed areas. So even though we have a concentration in areas that are prone, you see that we're being careful there. And it's -- these are large states. And if you're just careful about where within the state you're writing, you can avoid the bulk of the exposure, not all of it.

Got it. Heath, Back to you.

Great. Just stay on homeowners for a bit, how would you characterize the relative growth contributions that you're seeing from elevated demand versus limiting to expanding the criteria of homes the company is willing to underwrite. How do you think about the trajectory of this business in terms of customers and premium contribution over time relative to renters?

We have been broadening our aperture a little bit over time in terms of what we will write in homeowners. So it's really a question of getting the data in place and then gaining confidence that we're getting better, we're not where we want to be quite yet, but continuously improving and learning about which kinds of homeowners we feel comfortable underwriting. So over time, our declination rates have dropped. We've been more willing to assume risk as our ability to focus in and determine good risk from bad has been improving. And I think it will continue to improve for years to come. I don't think this is -- it's not like we've arrived at some destination. We just -- as we get better, we gain confidence, we're willing to undertake more. But the bulk of our growth has not come from that. It's really come from acquiring customers, both within homeowners and within renters, and now increasingly pet, which has been a positive and pleasant surprise to see how fast that's been growing as well. Reaching out to customers have a word-of-mouth spread, a brand has gained a lot of traction among our target users. So we're seeing our cost of acquisition continuously decline. Our marketing efficiencies continuously improve. Absolute marketing spend in Q3 this year was lower than it was during the same quarter of last year in absolute dollar terms, but we grew our business 100% year-on-year. So you're just seeing these efficiencies continuously to kick in. And it's not really because we're taking on board more risk. It's not a fundamentally different mix of products. It's just the machine learning how to do what it's doing better.

Yes. You touched on that, the machine learning, the data side of what you're doing at Lemonade. How has the technology at Lemonade played a role in improving your customer experience and lowering your loss ratio. And I think when we hear every company in financial services, talk about the investments that they're making in technology, we're seeing technology budgets increase, how does Lemonade's technology really differentiate itself from other companies that are also trying to throw a lot of money at technology problems?

Well, this is why I think it gets really interesting. So we improved our loss ratio over a 3-year period from kind of the 300 right down to the high 60s, low 70s. And that was a steady, stable decline at a rate of 5, 10, 15 points quarter-on-quarter, which is, I think, unprecedented in the history of the industry. And we did that, frankly, without changing price. Usually, when you see a massive improvement in loss ratio, you think, oh, they raised prices, and that wasn't the case. I think maybe in those 3-year period, maybe 1% of improvement could be attributable to price changes. Almost everything else was about this machine learning, both in the literal sense of machine learning as an AI and just a company as a machine learning. And I think what we're doing is something that is genuinely difficult for incumbents who have built up that business over the eons, over the decades, in some cases, centuries. It's very, very difficult to retrofit 21st century capabilities on a company that was founded in the 19th. So we built every line of code at Lemonade just built from scratch, the core, the very founding of the company is by technologists. The core capabilities of the company are technological. Every facet of our business, from finance to compliance and underwriting are being automated and run with a very strong technological infrastructure. We like to think of our teams as having kind of a technological exoskeleton. We do see levels of efficiency where each customer is serving about -- sorry, each employee is servicing about 10x more customers than incumbents can. So the technology is really giving tremendous, tremendous leverage. And what we're finding is that the least of it, although this is the most visible part of it, is the consumer experience. Whenever you use the tip of the iceberg that points out, kind of protrudes from the water line, and there, you're seeing the 3-second claims and the instantaneous cart you're seeing and the delightful experience and all of that, and that's fabulous. But that is the tip of the iceberg. What is perhaps less visible is that those delightful user experiences are generating order of magnitude 100x more data than traditional broker-based interface with customers would generate. And it's not merely a question of being able to generate a whole lot more data. The tonnage of data doesn't matter in a business like insurance, which is all about using data to price risk. But it goes much deeper than that because not only are we able to collect more data parenthetically. If I was offered to trade data set with any of the incumbent companies who have been around for 100 years more than we have, today, we wouldn't do it. We actually think that our data sets today have probably reached supremacy over the traditional ways of handling data. But not only -- as I said, not only on the collection. We're able to deploy data and to use data to our advantage in ways that if your business wasn't built for that, it's very difficult to do. So when you were chatting to Maya, a bot, she's not only collecting a lot of data, she is shapeshifting and changing her interface with you based on the very data that she collects. And humans, of course, can't collect data like machines can, but they certainly can't do that. So the whole experience of what she's offering you, what questions she asks you, what kind of coverages she's recommending, all of these things shapeshift depending on the data that she's collecting in real-time automatically. And then it's not only during one part of that process. But if you think about the claims bot or the customer support bots, every place that we go, from the Facebook marketing campaign that you clicked on, through to questions that you were asked prepurchase through the purchase experience, the post purchase, the claiming file, all of those things are really using a single repository at once, informing that repository and being informed by it. I think that companies that won't architecture that way from scratch, retrofitting that onto existing infrastructure, is near on impossible. So if you've been building technology since the 1980s on COBOL or whatever you're doing, your technology today needs billions of dollars of investment. Sure, you're putting a lot of resource into it, but it's much more of a black hole than it is a black box. It will swallow up all the dollars that you'll put into it, but it's really unlikely to deliver the kind of capabilities that I just described.

So that leads us into what some of your key investment priorities for 2021 might be. You've started down the path of launching life with a partner, you're building out your pet initiative. What other priority should we be thinking about as we look into the year ahead? I like that laugh. That suggests that there's a lot there.

There's a lot there. Not -- some of it is material non-public information. So I'll be careful with it.

Well, we're on a webcast, so the public is listening.

We launched -- you know we launched pet insurance just a few months ago, and we are on the cusp of launching life insurance. And yesterday, we launched France. A quarter before that, we launched the Netherlands. All of that during lockdown with everybody working remotely. So we launched multiple countries, multiple products. I think you can expect more of the same. So we are looking to continue to flesh out the product offering. I think we've been reasonably transparent about the fact that we want to cater to all of our customers' needs, and we're not there yet. There are insurance products that they need, and they turn to us to and we're not addressing. And the process that we began to talk about a bit earlier of our customers, the cohorts maturing and the needs growing, that also means that we have to continuously evolve not only to sell them a product, but to manage their lifetime value with us as customers. And that's fairly new to us. Until July, we really had either renters or homeowners, and you never had them both concurrently. You had 1 and then you graduated to the other, but it was really a single policy that you had with us. And now we've got multiple points of entry. People are coming into pet and then cross-selling to renters and coming in for homeowners and then adding pet. And pretty soon, we'll have life. And hopefully, after that, other products as well. So one of the things that, in addition to adding products, one of the things that we're continuously focusing on is dollar retention or lifetime value. How do we make sure that once we have these customers and we delight them, we also understand the needs and can service them more fully. And I think in the results that we published in Q3, we began to see elements of that. I still think we've got a lot of upside there before we've exhausted that. I think there's a lot that we can do there as well.

Daniel, I wanted to jump back to a prior question around technology and the use of AI and machine learning, which clearly has been successful and is one of the pillars of the company. And I think I understand its use in a very frequency-driven line of business like renters where you really see a lot of data. Can you use that technology and those capabilities and transfer them to lines of business that are more severity-driven like homeowners? And where I think both the industry, I mean like property catastrophe modeling agencies have, at times, struggled a little bit more.

They're not identical. I think you're right, there are some businesses that, like home, which may be influenced by things other than your behavior and you need other data sets in order to ascertain them. But I think the short answer to the question is yes, we can. And I'll give you 2 or 3 ways of thinking about this. I'm just making a couple of notes, so I think I'll come back to a couple of these points. But the one is the basic belief, it's more than a belief, that we have, which is that we're in the business of insuring people and not property. So if you and I have condos or homes that are identical and we live door to door, it doesn't mean that an insurance company should be charging us the same because you're a wonderful responsible human being and you clean your gutters and you lock your door and you turn on your alarm and you never leave the stove unattended. And I am different, and I'm going to forget to lock the door and I never turn on the alarm and the gutters are going to spill over any moment now and I regularly leave candles unsupervised. So I do think that understanding the human in the equation is a huge competitive advantage. And the more data we can get on that, the more competitively advantaged we will be. And in broad strokes, the industry knows that this is fundamentally true. In the '80s and the '90s, the industry learned that credit scores are highly predictive of insurance risks for drivers for homeowners for everybody. And we think what's the connection? Paying bills, what they've got to do with whether or not you'll lock your door. And it turns out that being a responsible human being is a kind of a generalized condition. People who pay their bills on time had a very high level of correlation to also lock their doors and leave their stoves supervised and all those other things. These things go together. It's a personality trait. So the more we get data on who our customers are and how they behave and whether they pay their bills on time and everything else, the better we get. And that will carry over in our thesis. And I think that the data supports this, from pet insurance, to renters insurance to homeowners insurance, et cetera. And indeed, that aspect of it cross pollinates across products, right? That's transferable because you may go from being a renter to a homeowner, but that fundamental, the content of your character does not change with that transition. So I think that becomes very important. Beyond that, we find that there's a lot of data sources that we can use and the ways that we can use data that is applicable to homeowners, for example, as well. So we've publicized this, we've got an internal bot called Cooper. Cooper twice a day ingests NASA satellite imagery. NASA overflies the continental U.S.A. twice a day with spectrometer imaging photography, which is heat sensitive imaging. And Cooper will ingest that and use machine vision to analyze it and will detect wildfires. By the way, now it's also learned other things, hurricanes and other things as well that it can detect with something we call watchtower. And what happens as soon as Cooper detects that, Cooper computes some kind of safe perimeter around the emerging disaster, wildfire, for argument's sake. And without waiting for human intervention, he will stop marketing campaigns in that area and oftentimes stop selling policies in that area. And so the ability to automate processes, to detect things early, to warn customers that that's coming and to ensure they're not writing policies in a fire zone while the fire is encroaching is something that you can do again if you're digitally infrastructured, but much harder to do when there's brokers in the field and you've got to operate through at the speed of humans. So I think there's a lot of advantages there. And we see that even with the underwriting of homeowners. We use satellite imaging to look at properties, to detect defects, to flag them oftentimes for human inspection. But we can get away with inspecting far fewer properties by humans because the first level of triage is done by machines, and they keep getting better and better at that. So I think the arc of history, if you like, is that the kind of infrastructure, the kind of building blocks that we have used as the foundations for our business get smarter and smarter and smarter, giving us both a cost advantage and a performance advantage over time.

Got it. That's helpful. And then if we switch gears a bit to reinsurance. So you changed the reinsurance structure in July. Purchase, I think, about 75%, or you ceded about 75% of your premiums, so reinsurers. And I think it's an interesting structure, right, because you're ceding over half of that on a 4-year contract and then the remaining just under half on an annual contract. So I guess my question is on that under half that's ceded, that's renewed annually. On the one hand, you've shown a lot of improvement on the loss ratio, which should not go unnoticed by the reinsurers, right? On the other hand, we're seeing just a general lift in reinsurance pricing for the industry as a whole. So as you look into your next renewal, what's the type of conversation that you're having with reinsurers today? What's the tone? What are your expectations of reinsurance costs next year?

And just one small question. It's a 3-year rather than a 4-year term.

Sorry.

It doesn't change the basic question. There are a couple of things. We do make extensive use of reinsurance, unconventional use of reinsurance. And we do that to achieve certain goals. We spoke about the predictability of bottom line results, our chargeable give back and being capital-light. All these things are into play in an interesting way in the way that we've done our reinsurance. But one thing to make clear is we're not -- we don't have to have reinsurance. We got discrete treaty in place that came into effect July 1 after another 3-year treaty expired. And we're thrilled with it, and we think it's a huge boon to our business. But we were out on the road getting that treaty done as COVID was exploding around the world, and it wasn't sure at all, everybody was panicking, a lot of businesses had difficulties, a lot of reinsurers had difficulties. Whoever was reinsuring the Olympics had a very bad time, et cetera. And it wasn't at all clear that we would be able to do what we ultimately did. And therefore, we were thinking through plan A, plan B, plan C and what would this all look like. And one of the realizations we came to is we don't actually need reinsurance. And we are a licensed insurance carrier. The great State of New York and other states that follow them have given us licenses and -- as have the Europeans in a way that is independent and not beholden to reinsurers. So we own our own underwriting entity. And if July 1 had ticked over and we hadn't got these fabulous terms for reinsurance, we'd have continued to write policies, we'd have continued on our business, we'd be absolutely fine, which means that when we go out looking for reinsurance, we're looking for reinsurance that makes sense for us that will lower our cost of capital, that will lower volatility, but that will do it at a price that works for us. I think as we go out for renewal, we'll find it there again. This is a commodity. Reinsurers are looking to make money. And so long as our loss ratio remains, on a multiyear basis, attractive, as it has been, then there's absolutely no reason why they wouldn't sign up for another year and another 2 years and another 3 years. But I have to tell you that as we get better and better at this and as our results become more stable, as our volatility naturally declines, we might look at this differently. There's an element of margin stacking that happens with reinsurance, right? Somebody else is making money out of our premiums, so we will regularly review this and decide whether we want quite as much reinsurance as we have previously. We'll assess that annually, and that is exactly why we opted to have half of this renewing annually, because not so much we're worried about the reinsurers not wanting to renew, but us questioning whether we would want to renew on the same terms again or not, and wanting to be able to have these windows where we can periodically review this and rethink this. So that's the way we're hoping to do it. If reinsurance is available on terms that are compelling, we'll take it. And if not, or the margin stacking is too aggressive or we think we can do better on our own, we might dial it back some. We're not committed to this structure.

Got it.

Daniel, we touched a little bit before on the move that you're making into life. I think maybe more as interesting or maybe even more interesting is the decision to do that as an agent. This is the first time that you've made that choice to do something as an agent rather than principal. Can you talk to us about the decision-making process there? And how you might use that same sort of decision-making progress around -- or process around other verticals in the insurance category?

Yes, absolutely. And this is a major product launch that we're doing using this distributional business model. It's not actually the first. We've been selling, for example, earthquake insurance in California, which is underwritten by Palomar. So -- and we decided not to write our own homeowners earthquake insurance for various reasons. So we've had this capability from the first day, and we've been taking this kind of make versus buy decision on an ongoing basis. But I take your point, what we're doing with life is certainly an extension of what we did previously and that takes it to a new level. There's 2 or 3 sets of considerations. Earthquake and even more so life are very low-frequency products. Most customers will never make a claim. Conversely, something like renters insurance is a much more high-frequency product. We made this decision really by being customer-centric and saying, we want to make sure that we give a seamless #1 rated Lemonade experience to our customers throughout their experience with us. And on a product like life, we could do that while using somebody else's paper. On a product like renters insurance, we could not. So life insurance, most people fortunately make it to end of their term life insurance alive and well and no claims are made. And in the rare instance that a claim is made, it's an uncontroversial claim, there's a death certificate, the policy pays out. So these are very straightforward policies that pay out rarely and uncontroversially. So we felt that we could give the Lemonade experience even if we weren't the underlying insurance company in that, in the instance called life insurance. And indeed, the user experience will be entirely Lemonade. It will be Maya the bot, she'll be chatting. She'll take you through streamlined experience and give you an instantaneous quote, which will be instantaneously bindable. We're doing a lot of stuff that is really cool and unusual in the context of life insurance, albeit doing it on somebody else's paper. The other thing to bear in mind here is that what would the lift be in order to do it on our paper. So in the case of life insurances, it's pretty dramatic because insurance companies are not allowed to write life insurance. In the regulatory environment in which we operate, you have insurance companies that are either P&C, property and casualty, licensed all life, but it's a never the twain kind of thing. So we would have had to have stood up a separate insurance company and have capitalized that. And that was another thing that made us just wonder whether this is worth doing or not. So we really take that kind of rational decision. There's nothing dogmatic about this. It's not principled that the only overarching principle is can we deliver the Lemonade level experience. And once that's answered, then how the planning is done, whether it's on our paper and reinsured or somebody else's paper, those are decisions that we take from time to time, and indeed, change from time to time. We're not committed to any one model of how to do the plumbing.

Great. Daniel, thank you so much for taking the time to join us today. This has been great. And we certainly look forward to doing it again soon.

Thank you. Thanks, Heath. Thanks, Yaron.