Good afternoon, everyone, and thanks for joining us for a Fireside Chat with Lemonade. Excited to have the company's CFO, Tim Bixby, joining us. Hi, Tim. Jason, how are you? This is our, I think, sixth year we've done this, and kind of maybe our last year as you're formally transitioning away from CFO at some point over the next year. So, yeah. So anyway, so sixth time is the charm, as they say, right? That's right. And I think we did start this live. At one point it was like an in-person maybe the first two years. Okay, so let's jump right in. I think everybody knows if you have a question, you can put it in the chat. I already see some questions already down below. Otherwise, email me at jason.helfstein@opco.com. Okay, so let's start with IFP. Very strong quarter, up 33%, 11th consecutive quarter of accelerating growth. Third quarter and full year implies kind of sustained growth. I guess as you're thinking about the business, what could cause IFP to slow and I guess within your control and without your control, and how are we thinking about next year, despite you not giving formal guidance yet? Sure. For Lemonade, growth is a gift, right? More growth is better. That's not always true in insurance, and historically, it's often been the opposite. For many insurance companies, growth and profit were at odds. You had to choose one. For us, it's typically the opposite. More growth typically leads to faster learning, more improvements, and you said it yourself, something like 11 quarters sequentially in a row of more rapid growth, accelerating growth, and at the same time, significant profitability improvements, loss ratio improvements during that whole period. From a go-forward basis, we've kind of set 30% plus as our marching orders. We first indicated that, I think, at our last investor day, which is almost two years ago at this point. And we've done it and then some, and we've gotten 30% and then grown that a little bit each quarter. I don't know that you can accelerate every quarter forever. There's a limit to that, I think. And I think to the question of what can get in the way or what can offset that or slow that is it's really of our choosing. As long as we can acquire profitable customers, a lifetime value that's forecast by us and by our models that's healthy, we can grow at that rate. And we've indicated 30% plus as far as the eye can see. And the good news is, since we made that declaration two years ago, we've been able to do it at ever-increasing absolute numbers, higher gross spend dollar amounts, though the growth rate is slowing somewhat, again, by our choosing. And maintaining the marketing efficiency and the kind of march toward EBITDA breakeven. We've indicated that's just around the corner. So we feel very comfortable with that 30%+. Our ambition is more. Our ambition is to increase that a little bit each quarter versus what is more typical, which is a flattening or a decline. That day might come, but our ambition is to continue the trend. Okay. We are going to go through a bunch of top-line metrics. Because there are some questions in the chat, we will get to margin. Customer growth most recent quarter was 23%, premium per customer, 8%. I think you expect no material change in that in the near term. I guess, talk about the dynamics of premium per customer, what drives that 8% cross-sell versus mix versus pure rate. Sure. I think premium per customer is a good output. It is not a great input, but it is a good metric to track. There is some noise there, particularly as our Europe business grows quite rapidly. The premium per customer in Europe, even though it is both renters and home, is relatively low compared to the U.S. business. A home policy in the U.S. is quite a bit higher. You have to be a little cautious in just looking at absolute customer numbers. In terms of premium per customer, I think our current theme will continue. We have seen a roughly 8% year-on-year growth rate. I think you will continue to see that, should expect that the remainder of this year, probably through next year. Looking out a little further into 2028, late 2027, 2028, and beyond, I think there will be upward pressure on that premium per customer, where that growth rate might move up a bit versus down from that 8% run rate. That, I think, is really driven by mix shift. I think when you see our current pet growth rate and car growth rate in the 50s versus the overall growth rate in the 30s, you really start to see that play out in the numbers in those out years. I would think of that 8% as edging upwards a couple of years from now. Okay. Segue into car. Grew 60% year-over-year in the quarter. 40%-50% of new car sales is from existing customers, so really good cross-sell. It is not widely available yet. I guess the goal is to have it in the majority of the U.S. by the end of next year. I guess just help us understand, where does this go? Is this like, do we assume car accelerates from that 60? Then just I guess how does that also a little bit impact margin, just because there is different maybe requirements per state as you add more states? Sure. More of the same. Generally, we expect car will grow because of the TAM and because of our efforts at quite a bit faster than the overall business. The state rollout is interesting, but it's not indicative, I think, of the trajectory or the growth rate much, meaning more states is good, more population's better. But we're in more than 40% of the U.S. population already, even though the number of states it's a dozen or so, but more than 40%. By next year, that should be more than 50%. While it's 50 is less than 100, it's moving pretty nicely. The growth rate of our autonomous product in terms of coverage is pretty notable. The premium is negligible, but I would look to the pattern of expanding coverage as a theme for Lemonade. We'll expand in traditional car states. We're now up to five car states already in the autonomous Tesla partnership product. Our ambition ultimately for sure is 50 states for all products, but our growth trajectory is not hindered in any way by the pace of expansion. At this point, the driver is really regulatory hurdles. We've become quite efficient at the rate filing process. That was not true three or four years ago. We were kind of new to the game, and we had to build that team, build that skill, build those muscles, and we've done that. One of the places we're seeing the most AI automation benefit is in those filings. A filing has to be done and approved by a human, but a lot of infrastructural work, the logistical work is repetitive once you get to know a state, and so we're getting really good at getting filings in. Then ultimately, we're still subject to the regulator's approval. Maybe talk about the unit economics of the kind of AV versus non-AV car policy, and then kind of the choices you're making around if it is more attractive to the extent it is a better margin product, do you kind of give that back by a lower premium? Go from there. Yeah. There's a couple dynamics there, and again, I would take this as directional, again, because the N is very, very small at this point. It's thousands of customers, not millions of customers. So you've got a couple dynamics that are different. One is obviously the price is notably different. The risk, the expected frequency of a claim is notably less, and the price is intended to kind of capture that impact. Meaning if you're doing it right, you're holding that margin potential intact. You're just paying less claims, and you're kind of returning that to the customer in the form of price. The second piece that's probably more interesting is the marketing piece or the customer acquisition cost. Because autonomous, because it's a very specific subset of driving, it's not just Tesla, but having a partner like Tesla who kind of leads this market, I think creates an opportunity where the customer acquisition cost can be different. And we got to be a little careful about what I can and can't say. You're going to pay to acquire customers one way or the other, but I'd be hopeful that the ability to acquire customers is more efficient when you have a very targeted audience. You've got a strong brand awareness and partnership through Tesla, and you bring another brand that kind of is closely tied with the way they operate and the way they think in the form of Lemonade. I would think you have benefits both on the claims or loss ratio side of the house, but also on the expense ratio, where you'd have a CAC advantage as well. So ultimately, we want to preserve the gross profit, autonomous or non, and we want to preserve the bottom line profit, autonomous or non. Will they be exactly the same? No, but I think because of those two dynamics, the potential they'll both be quite attractive to us. Again, thinking about kind of broadening the company, I think you did 14 state product combination launches in 100 days. I think that goes to your point about getting better at automating the paperwork and the applications. With nationwide kind of now being, I think should we think of now renters can now be a national, more broader product, which then just- Yeah. increases, again, more efficient customer acquisition, new customers in the funnel, et cetera. Yeah, I think that's true. I think our preference on the rollout is to do versus say, so I think looking at the first six months gives you an indication of our appetite. We can roll out faster than historically. A lot of the automation and AI-enabled work we're doing under the covers, under the hood, it's a little tougher to see it. We can tell you about our R&D investments and the ROI we expect from them. This is a great example. Being able to launch a state and a product in a week or a month instead of six months is a new capability. We've reached a new level in the first half of this year. I wouldn't take the first half of this year, which is a new high, and extrapolate that out, but it gives you a feeling for what's possible. Like everything else with Lemonade, we don't see an end in sight, meaning what we did in the first half of this year, we think could be better in the second half and better the year after that. At some point, this rollout question will be behind us, but we have a little ways to go before we're there. Okay. Let's talk about LAE hit 5% versus industry average around 9%. How much further can that go? I guess, this literally is the AI/data structural advantage. Do we see you evolve how you're thinking about pricing versus margin? Yeah. Yeah. Yeah. It's a great indicator. I think when we were getting to about 7%, we indicated we thought we might be able to cut that in half again. We cut it in half from 14 to seven. We thought, aspirationally, maybe we can cut it in half again. We are already at four. That wasn't too long ago, right? That was a few quarters ago, and we are already at five. The good news is it's not just one product that's fueling this, one type of claim. It's all the products are showing that improvement. Each product has a different LAE. There's a range of low to high, but they're all showing improvement, which is another good sign that this is a fundamental advantage. It's not like a one-off because one product is better than the others. We like that, and there's no sort of a floor in sight. There's obviously a floor of, you can't go to zero, or you can't go below zero, but I think the cutting in half from seven is a reasonable aspiration. At the time, I didn't say it. One of our founders, I think, tweeted it. But it's an ambition that's within reach. Talk about the synthetic agents program has become more efficient for you. I think you cut the cost of capital by six points. Where does that six-point savings go to? Does that change your LTV/CAC hurdle? Do you reinvest that in more growth? How do you think about where those six points of savings went? It phases in over time, so it's not an overnight. The renewal starts in January, and we got two great partners. General Catalyst has been a fantastic partner for many years in this area. Hannover Re is a unique player in the market. They've been with us as a reinsurance partner for 10 years. These are companies that know us really, really well, and that's why we're able to put together this structure and improve the economics a little bit. It'll phase in, though, over time. So January will be the switchover. We'll continue to repay all the cohorts that we've borrowed via General Catalyst until they're all repaid, and that will continue over time. There's no sort of balloon repayment or anything. That plays out over the next couple of years. In January, for new sales, new borrowings, new growth spend, that will come from Hannover Re at the lower rate. From a P&L perspective, you'll see that the benefit of that lower expense rate phase in over the course of those couple of years, and then ultimately, it'll all be at the new rate. But it comes in over time, not overnight. It runs through our G&A cost, rather than below the line, this interest expense, and we chose to do that. We think because it's an operating expense, it's tied to growth and customer operations, it makes sense to put it there versus not. I don't think it won't fundamentally change our LTV model, because again, it's an expense, it's not free. But it helps at the margin, to give us a little more freedom to lean in or to lean towards growth on cases that are maybe closer to the edge. Just talking about, I think you recently de-emphasized gross margin as a percent in favor of gross profit dollars by arguing the structural cost advantage will show up in pricing. I guess, we've seen some nice improvement. As you're looking out, I don't know if we want to say what long term is, but maybe several years out, where gross margins should be, I guess, given the whole mix of products? A significant part of the shift in gross margin over time, it's a combination of growth and loss ratio improvement for sure. Loss ratio will now ebb and flow more than it will just decline or improve over time. By definition, when you're in the 60s, I think we had a 59, but you were in the 60s, loss ratio shouldn't be in the 40s or 50s. That means something's unhealthy about the business. You'll see an ebb and flow probably in this range of the 60s. The change or the potential improvement in gross margin will be primarily efficiency driven versus loss ratio driven. LAE factors into it. Our LAE is embedded in the loss ratio, and for some companies, it's not, and so you have to sort of think about that dynamic. Then we've got some other components of gross margin. We distribute products that are not underwritten by us. It's relatively small, but that will grow over time. I think you'll see gross margins sort of normalize, not too far from where they're headed, maybe by the end of this year, I think you'll start to see sort of a more normalized rate. But again, you said the most important part, which is it's all about growing gross profit, maximizing that growth rate versus what the gross margin percentage is. I would think of the gross margin as an output and the gross profit dollars as the goal. We didn't ask one about gross loss ratio. 60%, there were seven points of favorable property development, three points of CAT. What's the right target level, I guess, as people are thinking about over the next few years of what the loss ratio should be? I would expect it not too far from the mid-60s. I think some of these patterns replicate and some don't. We've had fairly favorable CAT experience. That doesn't last forever, and so we have to be somewhat thoughtful about that. That's part of the reason we've been cautious about growing our home book of business. We've renewed with our reinsurance partner with greater protection against named storms. That doesn't show up in the P&L, but from a risk perspective, it mitigates some of that risk. So we're being thoughtful about that. There's makeshift components to it. The renter's book's super healthy. Pet is edging up a little bit. Car is showing nice trajectory. So, I would expect not too much diversions from that sort of mid-50s, or sorry, mid-60s range for gross loss ratio. To the extent we can, the gross loss ratio goes up, premiums come down, and conversion can improve. That can drive a faster growth rate, and that would be a good thing. On EBITDA, your guidance implies positive EBITDA in the fourth quarter. You said EBITDA should be positive for next year, but not necessarily every quarter. I guess investors, because there was one or two questions in the queue on this, does that mean if it plays out like that, then we're talking about positive margins in every quarter for 2028 and off to the races or- Yeah. Just more to do that could limit that linearity. Yeah, I think that's fair. While the Q4 guide for this year is important and notable because it's a change from negative to positive. Once you've made that shift, we didn't want to get too focused on Q1 versus Q2 versus Q3 next year. There's nothing fundamentally different about Q1, Q2, Q3 next year versus Q4. So the potential for positive every quarter is for sure there because it'll be pretty thin and there's uncertainty about weather and things like that. It could be in the first three quarters, it could be tight, and that would be fine and in line with our expectations. Q4 will be solidly positive, and the year will be solidly positive. I would read it that way. Then, yeah, for sure the following year, you're past that trough period where it's pretty tight. And we're seeing that pattern this year. If you look at the actual EBITDA Q1 to this year, the guide for Q3 and Q4, you see a similar dynamic, a fairly consistent Q1 through Q3, a bit of a step up in Q4, and that's the pattern we'll most likely see next year. All that, of course, is subject a little bit to weather and CAT, but everything else equal, that pattern is fairly predictable. Then again, translating that to earnings is another question in the queue. Look, we've got you kind of positive EPS in 2028. I guess, how do you think about your ability to control the earnings number once you're in the positive EBITDA range, several quarters in a row? The difference for us between EBITDA and earnings is pretty straightforward. It's stock comp and interest expense. Both are very predictable. We had a bit of a step up in stock comp in Q2, and we kind of talked about that with some unique multi-year founder grants. That's a step change, but that's not a repetitive thing. So that will be at a new normal, and then very predictable. So our ability to the extent EBITDA is positive and we have comfort and a track record with that, our net earnings will follow. Stock comp and interest expense are actually, in many ways, much easier to predict absent wild stock swings than customer growth is. So our confidence level is quite high. We've not indicated an exact date yet, but we have said within a year, roughly a year after EBITDA positive, we would expect net income positive. Your model's in the right range, so I think we're on track with that. To that point, once you have consistently positive free cash, the balance sheet is healthy. There is leverage, but the thought would be this business should always have leverage, and you have a healthy cash balance. Is basically the idea of if your choices are M&A or buybacks, and again, it's probably premature to like, it's not now, but as we're thinking about 2028 and folks are kind of going, "Okay, well, what's their ability to offset stock comp dilution?" Things like that. How do you think about that? Do you think once it's like, okay, we've clearly making money, EBITDA, GAAP, where does the money go? Yeah. I think our first bias will always be growth. We're at the edge of growth in a good way, I think. We've shown that we can grow a little more, a little faster each quarter, not only a faster growth rate, but if you do the numbers, the absolute number's growing at a pretty healthy clip of added business. But we want to see a bottom line that is positive, and we want to see it that it is number one, then predictably positive number two, and then growing over time. But I think our bias will be towards growth, before you get to things like buybacks or M&A, which are the two things you noted. Again, because if we can grow 40%-45% annually, at least, before you get real pressure from capital surplus requirement. Which is really the next thing that's not prohibitive, but it's a thing that you've got to finance that one way or the other. I think our bias would be in that range of 30%-45% to lean towards growth. Someday, if that's not the case and growth rates moderate or we're so large that 20% growth looks amazing, then we might consider things like a buyback or that kind of thing, dividends, that kind of thing. Okay. Let's now shift to AI, which can be wide-ranging topics. One, if you were to think about your organization's ability to deploy AI to improve efficiency, where do we think we are on a scale of one to 10? Ten is like it's doing everything and there's nothing more. One is we're still learning- Yeah. how to use AI. Where do you think you are right now? I think relative to where we ultimately can be, I think we're probably a one. Okay. I think relative to- Why is that? I think we all acknowledge the capabilities of what these models can do. yeah. Is it just like, "Hey, look, we haven't unleashed them because, I don't know, we're worried about hallucinations, we're worried about leaking customer data, we're worried about token maxing and uncontrolled spending." Why only a one? Because I feel like, I don't know. I think that would probably surprise some folks that you say only a one. Well, the second half of my statement was going to be, I think if we're a one, then all of our competition and potential competition is at 0.1, just to put a finer point on it. We're pretty far along. But yeah, one out of the goalpost keeps moving, so that 10 today looks very different than it looked three months or six months ago. The reason I say that is we see dramatic day-to-day impact in subsets of our workflows. But it's hard yet to see it in 100% of a given workflow. I'll give you an example with that. Now what does that mean? Adding features or launching a new product in a state might have taken us six months in an old structure, and that's just our part of it. Designing, reviewing, QA-ing, testing, all the different steps could have taken, or in some cases, can still take months, and that's before you get to regulators. We'll ignore regulators for now. We can now do certain types of products or certain type of features in certain type of locations. That set of work that might have taken months in hours or days. We've taken subsets of work from very long to very short, but we can't take the regulator out of it, and we haven't been able to apply that to 100% of all workflows, and so you're kind of stuck with your bottlenecks tend to move around. In finance, similar example. There's a process where we reallocate our P&L, either actual or forecast, through a reinsurance waterfall, either based on a new reinsurance set of agreements or a scenario or a pro forma that we want to consider as a new structure. We want to roll all the numbers through and see what the impact is. To get something quite accurate, either for the actuals has to be perfectly accurate, or for a scenario, you want it to be very accurate. That kind of work could have taken days or weeks for a bunch of humans to do a bunch of work. Now we've automated that to such a point where I don't see all the nitty-gritty, but I can do something like that in a couple of hours or a day or two, something that used to take weeks. We are applying that across the board in places that we either do not talk about publicly because it is competitive or things that are a little bit more arcane. What we do not see yet is when you do that across all of a set of workflows. We are in finance, we are ultimately shooting for a four-day close from a 20-day close. For us, that means a four-day close doing every step. Some big giant companies close in 4 days because they skip a bunch of steps, and there are lots of estimates, because they have to do that. We want to do a perfect 100% lockdown close in four days instead of 20 days. We are getting pretty close. Part of the reason we can have 50 people in finance today and 50 people four years ago, when we were a third the size and far less complex, is because of this kind of work. What we have not yet nailed is sort of the piece where AI can enable us or point us in the right direction to move some needles that have been tougher for us, like cross-sell and retention. Those are sort of the Holy Grail, sort of the magic goal of all of insurance is if you can move retention a little bit, it has huge value. If you can move cross-selling costs to existing customers a little bit, it has huge value. We are nowhere near sort of cracking the code, and whether that is using AI or human, it is usually a combination. Do you think that is because? It is early. There's just not enough liquidity in the system? Thinking, it would want to find a pattern, right, where you're like, "We did this, and that was better." But if you don't have enough of the data points, you can't know that, right? Once you're doing renters nationwide, you're going to have a lot more data points around how you move somebody from just a renter to something else. Would that be the reason why? Because I think most people would be like, "Oh, really? You throw the data in, and it won't tell you that answer today. Well, there's knowing, and then there's implementing, right? So you've got a lot of steps to flow that through, and you're deploying capital, and you're doing it in a way that's SOX compliant, and you've got to satisfy the regulators. The AI enables the capability, but the doing still has a pretty rigorous set of structures that we have to satisfy. I don't think it's the scope of where we are live in terms of territory. I do think it is the pace of learning, which is accelerating, and our comfort with it. A lot of our investment has gone towards making all of these steps more efficient. But I do think it's still pretty early. It's interesting, right? Given I think you've clearly been at the forefront of using AI and big data insurance at scale. If you're telling me that there's still these roadblocks, I can only imagine what the legacy companies are like. I guess for them, it might be like, "Hey, if we could bring AI into marketing and get some efficiency there," like, "Oh, okay, that could be meaningful." Whereas you guys have always been super efficient at marketing- Yeah. since day one, right? Some of this is like we may hear from incumbents, "We did AI with this," and they're just fixing something that was just highly inefficient, whereas you're starting much higher. The bar is already higher. Yeah. Yeah. There's another dynamic. That's exactly right. We have one single system that we've built internally to push everything through, which is hard, but wildly, dramatically easier than if you're a typical incumbent. There's another aspect, which is the tools are changing so fast, and the release of last week versus the release of three weeks ago or three months ago has dramatic impacts. You have to have the skillset to adapt to that, decide what to use, what not to use, how fast to adapt, and how to use that new release, and when to ignore the new release. These are really hard decisions that come very rapidly. This is just business as usual for us, and we're quite adept at it. That's a challenge before you even get to implementing it, understanding what it is and how to use it is super complex. Did you ever get to a point where you were getting nervous about the cost of deploying AI and not having to rein that in, or it's generally stayed pretty much within the bands? Not really. By definition, I'm always nervous about something that's new from a forecasting and budgeting perspective. From an absolute dollar perspective, it's been nominal and manageable, and we haven't had any I think it was the half a billion-dollar surprise at Amazon or whatever it was. We haven't had those kinds of issues. What do you think is the most exciting thing right now from a consumer standpoint? If a consumer tried Lemonade three years ago, and now they go and download the app and try it again, what are they going to notice and be pleasantly surprised that's different? I think the difference in the user experience today is still as striking or more so than it was five years ago. Forget AI for a moment, if you can. The most undervalued asset we have is how extraordinarily seamless and facile the user experience is for Lemonade. We have three million customers. We've got three million people who know that and then a bunch more who've tried it who know that. But in a market of 150 million in the U.S., and another couple of hundred million in Europe, these are just tiny numbers. The customer doesn't care about how fast we deploy. All they care about is a seamless, incredible experience, and there's no one who's remotely close. Everything we've done is out in the open. All the tools are available to everybody, and we've seen nobody who's coming anywhere near to closing that gap. All of the things we're talking about are really on the back end, which is great. Great for us, great for profit, great for investors. But in many ways, the hardest part we did first, which was that user experience, and now being able to deliver that at half the cost or a lower CAC or an LAE of 5 instead of 14, that's where the magic comes, is the product itself is just unparalleled. Last question. Obviously, you're seeing really good growth or a lot of positives going on in the business, and yet the housing market is really not particularly strong unless you're looking at the affluent side of housing. One could argue that by definition, you probably skew more average, not affluent, maybe in some cases below, just because of the renter mix. If we ever got into a rate-lowering cycle, and we actually started seeing housing loosening up, which is still not guaranteed because of the structural issues in housing, but when's the last time you felt like you saw that as a tailwind to your back? Because again, to the point of you think you got this great cool product, but most people don't think about their insurance, right? Yep. If you're not moving or you're not buying a new car, you're probably not thinking about switching, right? Yep. There's been a lot of people who haven't had the opportunity to even think about trying the product. Yeah, I think one of our benefits is, given our size, even at these high growth rates, we're going to be small for a while, and so the macro trends tend not to buffet us too much. That said, a radical shift in housing starts or lower interest rates or whatever could definitely benefit us. More first-time buyers is better. People switching from renting to buying is certainly a potential tailwind for us. But that's why the multi-product strategy has always been a core benefit for us, is that's a subset of the market, and there'll be ebbs and flows in each of the product lines, and our ultimate goal is just more humans. No matter where AI goes, we want three going to six going to nine million humans using Lemonade and using all the products. So all those things are good. Great. Well, we're going to stop there. Thanks everybody for joining us. Thank you, Tim, for your time, and we look forward to seeing you all at our next session. Thanks, Jason.