Good morning. My name is Yael Wissner-Levy, and I am Lemonade's Vice President of Communications. On behalf of our Co-Founders, Shai Wininger and Daniel Schreiber, I'd like to welcome you all to Lemonade's Investor Day 2024. Before we begin, I would like to remind you that management's remarks in this presentation may contain forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. Actual results may differ materially from those indicated by these forward-looking statements as a result of various important factors, including those discussed in the risk factor section of Form 10-K filed with the SEC on February 28, 2024, and our other filings with the SEC. Any forward-looking statements made during this presentation represent our views only as of today, and we undertake no obligation to update them.
We will be referring to certain non-GAAP financial measures during today's presentation, such as Adjusted EBITDA, Adjusted Free Cash Flow, and Adjusted Gross Profit, which we believe may be important to investors to assess our operating performance. This presentation also includes information about our key performance metrics, including In Force Premium, Premium per Customer , Annual Dollar Retention , Gross Loss Ratio , Net Loss Ratio , and Growth Spend . Definitions of KPIs and non-GAAP measures can be found in the appendix. Reconciliations of non-GAAP financial measures to the most directly comparable GAAP financial measures are included in the appendix to this presentation, which will be posted shortly on our Investor Relations website, investor.lemonade.com. Okay, so thank you again for joining us here in person from our offices in New York, and welcome to all of you joining us online from around the world.
Today, you're going to hear from our leadership how we're going to 10x the company. You're going to get a front-row seat to our plan to grow from $1 billion in In Force Premium, IFP, to $10 billion. On our agenda for today, you're going to hear first from our CEO and Co-Founder, Daniel Schreiber, who will lay out Lemonade's vision and strategy. Next, our Chief Business Officer, Maya Prosor, will talk us through the growth plan from one to 10 billion. She's going to hand it off to Adina Eckstein, our Chief Operating Officer, who will share how AI enables infinite scale. Following that, you're going to hear from Nick Stead, our SVP Finance, who's going to present our model and key metrics. And then Tim Bixby, our CFO, will share the view ahead. Daniel is going to end the day with some parting thoughts.
Just to level set everyone around some of Lemonade's facts and figures before we begin. We sold our first policy, a renters policy here in New York in late 2016. After we went public in 2020, we went from a monoline to a multi-product line insurer, insuring products like Pet, Car, and Life . Today, a Lemonade policy is available for 92% of the U.S. population and nearly 50% of Europe. Our more mature products, Renters, Home, and Pet , show our diverse and balanced book, while Car, as you're going to hear today, is well on its way. In just over eight years, we've grown to nearly $1 billion in In Force Premium with more than 2.3 million customers. Internally, we are a tech company. A quarter of us are in product and tech, another quarter in sales and marketing and G&A, and the rest are in customer operations.
So we are ready and prepared to 10x ourselves. We've done it before, and we're going to do it again. And for that, let's begin with our vision. Daniel Schreiber, thank you.
Good morning. Thanks to everyone present, both in person here in New York and from around the world. We value your time. We value your interest in Lemonade, and we hope these will be amply rewarded during the course of the next couple of hours. I'm going to start by zooming out and try and put Lemonade's ambition and opportunity in its broadest possible context and then zoom in a bit from there. But let's start, as I say, with a big zoom-out picture. The world spends more money on insurance than just about anything else. More money on insurance than on artificial intelligence, than on automotive, cloud computing. Entire defense budgets don't amount to how much we spend on insurance. Oil and gas as a sector is smaller than insurance, semiconductors, software, the list goes on. In short, insurance is a prize worth fighting for.
We'll try and make the case today that it's a prize that is up for grabs. Now, I say that notwithstanding the fact that the incumbents, the reigning insurers, have been on top for centuries. And they have massive heft. They do oftentimes $100 billion, sometimes closer to $200 billion a year in revenue. But longevity should never be confused for immortality. And I actually think that insurance may be the most disruptible industry on the planet. I say that because of structural changes that the world has gone through. Insurers forged in a bygone era are structurally disadvantaged in the age of the machine. Their assets suddenly reveal themselves to be liabilities. 40,000 agents. Is that good in the age of the app-based direct-to-consumer distribution? Their cultures, developed over centuries for legacy preservation, suddenly appear maladapted when what we really need is business transformation.
Their legacy systems behave a lot more like black holes that devour time and cash than what they need, which is black boxes that make everything faster and cheaper. Which brings us to Lemonade's founding thesis. Our thesis is very simple. It is that a new kind of insurance company built from scratch, legacy-free, on AI as a core component, a plank from day one, will enable us to delight consumers, quantify risk, and collapse costs in a way that no other structure allows you to do. Lemonade was founded to do exactly that. By the end of this morning, I hope you will find that proposition investable. Let's begin. I'd like to begin with two things that have changed since we met in this room on this feed two years ago, almost to the day for our very first Investor Day. They're pretty simple.
Today, as compared to two years ago, LMND represents far less downside, far, far more upside. On the downside, a couple of years ago, there were questions hovering in the air. What's the liquidity risk for Lemonade? This past quarter, we contributed $48 million to our cash balance. But two years ago, in the same quarter, we lost $63 million. Now, that $111 million swing is not just a change in degree. It's a change in kind. Since we last met, we have shifted from being a company that burns cash to one that generates cash. If ever there was a concern about our liquidity and our ability to continue as a going concern, there no longer is. There were question marks in the air a couple of years ago when, at the same quarter, we reported a 94% loss ratio. Does Lemonade really get insurance?
Can they master the art and science of loss ratios and insurance? Well, just a couple of weeks ago, we reported a 73, smack bang in the middle of where we guided to where we want to be. 21 points reduced over the course of two years. If it's not unprecedented, it's close to that. If questions hovered in the air a couple of years ago about, do these guys get insurance, I think we can lay those concerns to rest as well. We laid out some pretty ambitious plans. We showed you models for multi-year compounding growth and all different goodness, cash flow positivity. People had questions about, can they pull off what they're telling us? Folks, two years on, we've done everything that we said and then some. Tim will cover some of that later.
But let me just say, we've been a public company for 17 quarters. For 17 out of 17 quarters, we have delivered on the guidance that we've given or handily beaten it. So if there were concerns about the execution machine that is Lemonade, I think these two are materially de-risked or entirely put to bed. And then there's the upside. This time, two years ago, we spoke about a 20% CAGR going forward. Today, we're going to update that to 30%, more growth. This time last year, we spoke about certain efficiency metrics that we thought AI would deliver for us. AI has been moving much faster than we anticipated, and the dividends from AI are much greater and more dramatic, and you'll see that today in spades.
I would put it to you that LMND, on a risk-reward basis, has never been more affordable than it is right now. Some things have changed. Less risk, more reward. Those changes themselves are really byproducts of something incredibly steady, bankable, something that hasn't changed. I'll go through a few of those things because the trajectories are not new. The impact is the trajectories are constant. What you're seeing on the screen right now is our In Force Premium. Those are the tall columns that you see. At the bottom, the small line is our operating expenses. This dates back not just two years ago, but since our IPO, our business has grown sixfold since our IPO. If I look at two years ago, we had just completed the Metromile acquisition, which is the single kind of jump that you see.
Everything else is very steady as she goes. You'll see that two years ago, we reported a roughly $600 million in In Force Premium. Two years later, just shy of $900 million, almost a 50% growth in a two-year period. What happened to the corresponding costs, expenses? Folks, they declined. This time, two years ago, reported $87 million of operating expenses net of growth spend. We'll come back to that later. This quarter, you look at the underlying cost structure of Lemonade. It is smaller than it was when we were far smaller ourselves. There's some magic happening here of us being able to grow dramatically while keeping expenses absolutely flat. This geometric shape, technically, it's called angle rays, I'm told. This shape is the most compact visual and perhaps memorable point I want to make today, and you'll see it recurring throughout our presentations.
I hope you take away many facts and figures and insights from the different presentations and different angles that we deliver, but the principles captured in this image will underlay them all. Grow the business, scale the operation. This is the shape of our expense ratio. Some of you will remember I showed this image, this cartoon a couple of years ago to explain our expense ratio, this image of a young girl struggling under the weight of a backpack and then growing into it, and really, the point I was making then is that our fixed costs are fine. They're not going to grow. We are going to grow into them. The denominator will take care of the expense ratio. That's exactly what happened. Our expenses have not shifted, but our business has grown. Growth for us is the gift that keeps on giving.
So if two years ago, we were uncomfortably burdened by our cost structure, today we're much more comfortable with it. And two years from now, when we're profitable, it will be a breeze to walk around with this backpack. The same geometry, the same shape underpins our gross profit. And this is interesting. I spoke about a 50% growth in our top line, but our gross profit has 3x'd during that time. 3x increase in gross profit without any movement on the expense structure. This is somewhat gravity-defying. As fast as our business has been growing, our profitability has been outpacing it dramatically. And all of that has led to this line, which is our EBITDA margin, our EBITDA over Gross Earned Premium. This, in the most literal sense, is our path to profit.
Our EBITDA margin today is more than twice as good as it was when we last met. It has been improving for about five years at a rate of almost 10% a year, plus minus, and it's been improving pretty much like clockwork. In 2021, we told our investor community that we expected to be EBITDA profitable in 2026. We're still saying that now, and you can see the milestones along the way, the dependability, the bankability of our performance as we near that. I've been showing you two-year horizons. I've shrunk it down to one year because a year ago, we started re-accelerating our business, and this line shows you our acceleration curve as well. Q3 of last year, we grew 18%, then 19%, then 20%, then 22%. This last quarter, 24%. The quarter we're in right now, we've guided to 26% growth. That's pretty rapid acceleration.
How often do you see rapid acceleration and expenses staying flat to declining? In the years ahead, we expect that to continue. 2025 will grow faster than 2024. 2026 will grow faster than 2025 as we grow into that 30% CAGR. We might not hit that entirely in 2025, but from 2026 onwards, we expect to be at that cruising altitude. The upshot is that we're cash flow positive. We had guided to being cash flow positive in mid-2025. I can tell you now that 2024 will be cash flow positive as a year as a whole. And we were cash flow positive in Q3 by any metric: operating cash flow, free cash flow, net cash flow, you name it. It was positive in Q3. What's next?
With cash flow materially behind us, we're going to be focusing, as we continue to do what we're doing, on EBITDA profitability in 2026, net profit in 2027, and massive profitability thereafter. These are all milestones on our way to the 10x. They are all byproducts of maintaining that shape as we 10x. They inexorably follow from us continuing to do what we've been doing, which is why we're going to maintain that shape. We're going to keep doing what we're doing: accelerate top line, hold expense line, wash, rinse, and repeat. So while the strategies, the geometry, the physics of our business is unchanging, notably, our product mix has continuously shape-shifted in order to match the stage of the business as we go from $1 billion to $10 billion, we will see that change manifest pretty powerfully as well.
Most notably, while renters and pets were able to propel us to $1 billion, Car has to start revving up in order to propel us from $1 billion to $10 billion. We bought Metromile, as I say, two years ago, and we've been busy absorbing all of their data, all of their experience. It's been a huge unlock for a lot of what you're going to see in my presentation and ones that follow. But we have not rushed to scale this business. We have been biding our time. We have been building, experimenting, tinkering, and perfecting. While Lemonade Car has played a supporting role in our financials to date, it has been primary in our investments for some years now. Behind the scenes, we have invested far more in Car, more engineers, more insurance specialists, more product and design people, more data scientists than in any other line of our business.
By 2025, during the upcoming year, we think we will be feature complete. We'll never rest on our laurels. We'll keep developing stuff. But we will get to the point where we feel that this is ready for prime time, that we can start really growing aggressively. And that will happen as the year progresses. What do you need to know? All of that tinkering, all of that building, all of those investments, what's important for you to take away from what I'm saying here? Lemonade Car promises to have unrivaled experience at unbeatable prices. Now, you might say, unrivaled experience. Okay, Lemonade has always had unrivaled experience. Let's focus in a minute on that unbeatable price. And in a sense, it's what we've done already for our other products. We launched renters insurance in 2016 with a promise of killer prices, monthly subscriptions.
It's been our most well-performing campaign pretty much ever since. We haven't always waved the flag of being a price leader, but it has always been part of what you are reasonably entitled to expect from Lemonade. So let's touch on a couple of things around these killer prices for Car. The first is, why are we motivated to have killer prices? And this may be self-evident, but our empirical results have shown that a 10-point change in price can yield a 50% all the way up to 150% change in conversion. This is incredibly price elastic. This is an incredibly price-sensitive market. Being able to get to price leadership has massive dividends for whoever can pull it off. Okay, you say, I get the motivation. That's pretty clear. But Lemonade's Car business is roughly $100 million today. GEICO's and Progressive's are like $50 billion-$60 billion today.
So the guys you're going up against are 500 x your size. How can you possibly hope to outprice them? Fair enough. I want to talk to you about three things that we have that nobody else has in car insurance: telematics, and those of you who think, well, everybody has telematics, I'll address that later on. Telematics. We have it. Nobody else does. Take my word for it for a minute. Suspend disbelief. I'll make my point later. Distribution. We have unique distribution advantages that Maya will talk about, so I'm going to talk about telematics, Maya about distribution, and the way we use AI to automate across our business will yield tremendous cost savings in the car insurance to a place that is oftentimes very, very costly, so I said I'm going to talk about telematics. Let me dive in.
The way most of you have your car insurance priced is based on proxies. Insurers don't really know how you drive. So they look at your gender, your marital status, your education level. In most states, your credit score. These are kind of disappointing ways to price how somebody drives. Under that kind of lens, humanity looks like much of a muchness, very much alike. Under this crude lens, diverse people are combined in this low-resolution image. I kind of end up with this two-dimensional outline of a humanoid or a cardboard cutout. That's the best you can do when you're using things like your gender and your age to price insurance. Telematics changes the game in dramatic fashion, at least the Lemonade version of telematics. It's not a low-resolution 2D cutout. It's the full HD. It's the 4K. It's the 8K.
In a very deep and real sense, it's virtual reality. Because at high fidelity with precision instruments, we capture every mile driven, every foot or inch driven, every tap of the brake, every swerve of the steering wheel, every time somebody texts while driving. Marital status, credit scores, they don't hold a candle to the kind of insights that we glean. What's that got to do with killer prices? Everything, so let's have a look at some illustrative data here. I've charted here 100 people who are being priced as cardboard cutouts, and no wonder they have roughly similar on average proxies, and therefore they end up with roughly similar average prices. The average price is $2,348, a national average price pretty much, and there is some variation, but not much.
Now, the actual variation of these self-same drivers as captured by telematics would make that curve look more like this. People are wildly different in how they drive. The gender doesn't begin to capture that. So Jill should actually be paying $400, no, sorry, $600 more than average. You can see she's got that phone in her hand. She may look good by way of her proxies, but she's a distracted driver. Telematics will pick that up. The cardboard cutout misses it entirely. On the other hand, Jack, who's a young driver and maybe his proxies aren't so great, is actually a cautious person. The true price for Jack should be $400 lower than the average. Both of these entirely missed by the bulk of the industry. Now, to put some proportionality around this, about a third of drivers are worse than average.
They're the ones who ruin it for everyone else. The problem is that traditional methods can't identify them. So we price them as average. When Jill comes to Lemonade, she will not be delighted with the price she gets. She will tweet out that this is not unbeatable prices. This isn't killer prices. It's 20% more expensive than I got quoted at GEICO or Progressive or State Farm. And she's right. We're able to identify that she's a worse than average risk. She's better off being mispriced as average elsewhere. Good riddance. That will manifest itself as negative selection for our competitors. Jack, on the other hand, is one of a huge cohort. In fact, Jack is the majority. 2/3 of people are subsidizing 1/3. Again, insurers just can't identify them as such. Jack comes to Lemonade. We see how he drives.
We give him a 17% discount, a rate commensurate with the risk. He saves $400. He feels like he got killer prices. Welcome on board, Jack. That's exactly what positive selection looks like. All told, our telematics can lower premiums by about 15% for 2/3 of customers. This is not a niche play. This is the big game. This is the game. If 15% has a ring to it, it's because somebody built a $40 billion-$50 billion business promising that they could save you 15%. In 15 minutes, mind you, but we'll leave that one aside. 15% is a game changer. It's how leadership is established in this market. Okay, I said I was going to come back to this. Doesn't everyone do telematics? Hasn't it been around since the 1990s? Didn't Progressive invent it? Fair enough.
Let me take you through this one bit at a time and try to unpack it. Does everyone do telematics? Well, it really depends on how you're using those words. What does do mean and what does telematics mean? I'll show you three or four vantage points onto that. Let's just start with how many companies are actually doing telematics. Data on this is a little bit sparse. The Wall Street Journal did a big cover story on this three years ago, and they claimed that less than 4% of policies in the U.S. use telematics. I've seen 5% and 6%. It's been growing. I've rounded it up to 9%. I don't know what the precise number is. But we are 10 x more in terms of our telematics use than the industry average.
Now, among incumbents, there are some who in some channels do better than 9%, some who do worse. Fine. But taken as a whole, broad strokes, Lemonade is 10x on telematics than the whole industry. Not 10%, 10x. But honestly, this is just the beginning. Because what we mean by telematics changes, what we mean by do changes, but what we mean by telematics changes because those telematics policies of our competitors aren't continuously in use. Those 9% invariably will run telematics for you for two weeks, three weeks, four weeks, and then stop. The Progressive program is called Snapshot because they take a snapshot, they price you, and they're done. So when somebody else says they're doing telematics, best case they mean we're doing it for a couple of weeks, pricing you, and then we discontinue it. That's not what Lemonade means when we say telematics.
For us, for 92% of our business, it is continuously generating and streaming data on how you're driving. This isn't a 10x. This is a 100x for the rest of the industry. Why is this so meaningful? Two things. First is on price. It is true that you get some good data in those first 30 days, and it's better than nothing at all. But it's nowhere near as good as continuous telematics. I can tell you empirically from our data, the telematics score for the median user changes 23% between when those 30 days are up and when the first year is up. 23% is huge in a market as sensitive as ours. 23%. And then there's the claims. If you don't use continuous telematics, then at the moment of impact, when an accident happens, you know nothing. You're back to old-school ways of adjudicating a claim.
In our case, it's radically different. We had in your car streaming in real-time high-fidelity, high-precision instruments measuring everything that you did at that moment of impact. And we get the impact dynamics. We get the location and the timing of the accident. We get the speed and the direction. And we can do a lot with that. First of all, we can help you. Your phone will jump up. Have you had an accident? Do you need us to send help? Secondly, do you want us to start a claim for you? And when you do start a claim, we already have so much of the information that we radically reduce fraud. Think how many lies you can tell when we have that level of X-ray into how you are driving. And we radically reduce the time and the cost of adjudicating your claim.
There's one more layer in which when we say telematics, we mean something different to everybody else. Lemonade's telematics is proprietary. We are a data science company. What I'm doing here is I'm normalizing the no telematics approach, so if you look along the horizontal axis, we're moving from low risk to high risk, and the vertical axis is showing the system's ability to match and to detect the fact that the risk has changed. Now, I'm somewhat artificially flatlining as a baseline those cardboard cutouts, the proxy, so I'm holding that flat in order to show you the relative strength of different systems. Traditional telematics, the industry-standard telematics, I'm not talking about now whether it's continuous or not. I'm talking about the data science behind it. If you have an insurance company and you now want to know, somebody tapped the brake, what risk does that represent?
How should I translate that into rates? What does that even mean? Mapping behavior onto risk is a data science, machine learning, deep learning challenge. To the best of my knowledge, there is only one incumbent in the country out of tens, if not hundreds, who are using telematics, who have done this hard work themselves. Everybody else uses the same off-the-shelf lookup table that we started with as well. So we have a pretty good sense of how good or limited it is. And it does show you some lift. So it is better at detecting low risks, better at detecting high risks than my baseline. But let me layer on top of that a real drawn-to-scale version of Lemonade's own telematics. That's dramatically different. We use the same word, telematics, but we mean something quite different.
This is born of a structure, the benefit of having a single integrated system, having our own data science in-house, not outsourcing our core competencies as the rest of the industry has wanted to do. Note, for many customers here, the delta between no telematics and industry-standard telematics is smaller than the delta between telematics and Lemonade telematics. That's how big the difference is. Combine those, 10 x the usage of telematics, 100 x the usage of continuous telematics, and a qualitative jump in every data that comes through because we can analyze it, have analyzed it using all of a trillion data points dating back to 2013 when Metromile pioneered this thing. Now, the same guiding principles, we don't always have telematics, but the same philosophy and principle underpin everything that we do at Lemonade.
Adina will pepper in an example from our Pet insurance product a bit later in the morning. I delayed on Car for obvious reasons, but nothing about Car is distinct in how we approach the challenges or in our determination to offer unbeatable prices. A concern I often hear when I start talking this way is, "Okay, what about loss ratios? Don't killer prices sound daunting, scary? Don't they translate into elevated loss ratios? Anybody can grow by selling dollars for $0.90 . Is that what's going on here?" Not by a mile. Let me say two things about this question, and they center around two concepts that I'm going to return to a couple of times this morning: precision and automation. Let's start with precision.
Precision is what I was just talking about with telematics, the ability to detect with great precision and differentiate a homogeneous group into its component parts, to see every cell and judge it differently, price it distinctly. Well, if you have that and you use that to lower prices, you do not adversely affect loss ratio. If I take Jack, see him for the risk he is, which means he's a lower risk, and I price him commensurately lower, the loss ratios are maintained. Nothing at all changes other than my conversion goes through the roof, my retention goes through the roof. The loss ratio will be a constant. A perhaps more interesting insight comes to what happens when there's automation in play. What happens when you build your company on AI rather than on humans, and you can scale infinitely, as we'll be talking this morning?
Because high loss ratios can, of course, be a signal of bad underwriting. More often than not, that's what they are, but they can also be the telltale sign of extraordinary efficiency. Let me show you what I mean. I'm taking here two illustrative companies. One is Lemonade, and the other one I'm calling Acme Insurance, and the company on the right, Acme Insurance, is pricing a policy at $2,000. Doesn't matter if it's Car or H ome or whatever policy this is, and we're charging 15% less, $1,700. And in my illustrative example, we have a 30% advantage when it comes to automation. So our cost structures are that much more efficient. As I say, illustrative, but I think quite reasonable set of assumptions. I have held the losses constant. I'm making a distinct point to the one I made earlier about ability to detect risks.
Here, we're underwriting the identical risk. Both companies are underwriting a risk which is going to cost $1,050 to pay out. Look what happens to the loss ratios. Lemonade's loss ratio here is a 70% in my example. Acme Insurance is 63%. If this is the prism through which you look at companies like ours, you'll come off quite disappointed. Our unit economics by this measure look worse than Acme Insurance. We have a higher loss ratio, and even if you look at the combined ratio, they're identical. Best case you'll say is that they're on par with the rest of the industry, but folks, that is a very misleading conclusion to draw. Unit economics and loss ratios in particular only take you so far.
If we're able to price 15% less than the rest of the industry and deliver the same net margins, this is a company on a rocket ship. That's those 15% that can make you into a $50 billion business. The company on the right will have growth dynamics that the rest of the industry don't know, conversion dynamics, retention dynamics that are unmatched. And for you as investors, if you get to choose which of these two companies to invest in, 10 x out of 10, you want to invest in the one with the worst loss ratio on my example here. All told, beware of averages, particularly as we lower prices. Follow the dollars, follow the gross profit. See what the machine is generating in terms of gross profit growth rather than fixating exclusively or even primarily on loss ratios. But it goes deeper.
In fact, it goes to the very core of our strategy since our founding. For the very first time, I want to share with you a snapshot of a deck we shared with the team at the very first management meeting back in 2015, just as the company was being founded. It hasn't been changed, hasn't been doctored. This was the strategy slide. You'll notice a few things. One is we hadn't discovered the color pink quite yet. The other one is that we hadn't realized how dumb it is to try and call ourselves the world's first peer-to-peer technology insurance company. What a mouthful. You'll also notice some other things that have not changed. Technology, artificial intelligence, guys, not 2023, 2015, who was talking about artificial intelligence as the foundation of their business in 2015? And untouchable rates. Those are the points that we laid out from day one.
Why? Let me take those one at a time. I'll talk about technology, and then I'll explain how I get to untouchable rates. Technology and AI give us two profound advantages. It's a structural advantage, and that structural advantage begets you a competitive advantage. And I'll fly through this part. I hope it's familiar. The structural advantage is that we were built as a tech company. We are vertically integrated. You saw the example now with telematics. We get to do things that others don't get to do, and the whole morning will illustrate the advantages of our system. We've got a single proprietary system. It garners and deploys data at every touchpoint with the customers and grows smarter and smarter as it goes. Contrast that with the incumbents. This is a quote from Ajit Jain, the Vice Chairman of Berkshire Hathaway, who owns GEICO.
Last year, he said that GEICO has more than 600 legacy systems that don't really talk to one another. That's a structural advantage. Build a company from scratch on your own technology as opposed to decades of layering archaeology, one on top of the other systems that don't even correspond to each other. How much can you contend with the AI era if that's how you built your company? And then GEICO is considered one of the better companies in the sector. But this is a distinction with a difference. It matters because I'm going back to those two key words. Two things matter in insurance: precision and automation.
Precision is the ability to use big data, machine learning, deep learning to quantify risk better than anyone else, to gain dominion over the loss ratio, whether it's determined to be optimal at this level or that level to own the loss ratio. That's what precision is about, and technology does that better than anything else, and automation, of course, is about technology. That allows you to delight consumers with three-second payment of claims, an instant everything, and no paperwork. That boosts your brand, and it lowers your cost of acquisition, and it collapses the cost to serve. That's your loss ratio. That's your expense ratio. That's the whole kit and caboodle in insurance. Technology is a structural advantage that manifests as a profound competitive advantage in the two things that matter in insurance. Okay.
It does not follow, though, that we have to cash in that advantage as untouchable rates. That's a choice, and I want to explain it. I gave you an example before: Acme Insurance charging $2,000 and us because we're more efficient charging $1,700. We don't have to do that. We could charge the same $2,000, get similar growth rates, and just have a real margin-rich business. It's a choice to cash in our technological advantage as margins or growth, and we bias growth, and I want to explain why. You can map out a profit frontier: margins versus growth, smaller margins make it up on volume, and you can map out what kind of profit potential anywhere along this line will generate the same amount of profit. So at price X, this is the profit pool. If I give those now famous 15% discount, this is the profit pool, so at price X, this is the profit pool.
And lo and behold, they are identical. That's the nature of a profit frontier. So at any given scale, we are truly indifferent to whether we make it up on volume or make it up on margins. Our results will look very, very much alike. But while that's true at any point of scale, we are not indifferent to scale. As we scale the business, the profit frontier itself moves. If that was the potential profit at scale X, this is the potential profit at 2x. And this 2x is born of us being a technological company, as having that as our substrate in our DNA. Traditional businesses, traditional architectures enjoy benefits of scale, but they diminish. They taper off. At some scale, you barely feel the advantages of the marginal scaling. Tech companies like ours have the opposite dynamic.
For us, the efficiency returns, the savings that we get at scale accelerate. They don't diminish. The more we grow, the more marginally profitable we will be. And that is why, since we're planning for the long term and for maximum profit over time, maximizing the enterprise value, we're going to bias growth in order to get to levels where we can continuously increase the total profit of the business. Let me try and bring it all together by connecting the dots. And I'll try and create a single image that hopefully coheres in your mind and captures everything that I've been saying. Lemonade is a tech leader in insurance. I think that is inarguably true. Technology begets you two things: precision and automation. And we believe we're well on the way to being leaders in both as a direct result of us being technological leaders.
We choose to cash that in as price leadership, which will beget us growth leadership, which will ultimately yield the largest pool of profit we ever could. That, in a nutshell, is our strategy. There's another dynamic at play here, that frontier moving. Because the more scale we get, that compounds our automation advantage. And the more scale we get, that compounds our precision advantage as we get more and more nuanced in our ability to manage our business. So you saw our founding strategy. It is that, and staying true to that, that has brought us today to being cash flow positive. It is the same strategy that will deliver EBITDA profitability, net profit, and in due course, massive profit. It is that that is propelling us to 10x our business. And when we 10x it, it's that that will propel us to 10x it again.
By the end of this morning, I hope you will find that proposition investable. Thank you. Thank you. Let me just introduce the speakers that follow. I want to do it by way of going back to my geometry. I said this is going to follow us throughout the day: growing the business, scaling the operation. If you like, you can map those onto two leaders in our organization. Growing the business is Maya Prosor, our Chief Business Officer, and scaling the operation is Adina Eckstein, our Chief Operations Officer. You'll hear from them in succession, one after the other.
We will then follow that with our SVP Finance, Nick, who will try and model everything that we saw and try and make sense of all the plethora of examples and data points that he'll here present in the next couple of presentations so that it all comes together for you in a simple form that you know what to do with.
Good morning, everyone. It's great to see you all. I'm going to focus my presentation today on why we believe Lemonade is an outstanding growth company and our growth strategy in the upcoming years. Lemonade is about to cross a very special and exciting milestone. In early 2025, we will reach $1 billion of In Force Premiums after eight and a half years in market. This is about the same time it took Tesla, Spotify, and Salesforce to reach $1 billion of premiums. So we're already in great company.
We've been accelerating our growth. As Daniel mentioned, in 2025, we should expect to see faster growth than this year, and in 2026, we're going to grow faster than we plan to grow in 2025. Overall, we plan to grow at a CAGR in the upcoming years of above 30%, so if you stretch that straight line, that means that it will take us about the same time it took us to reach $1 billion to 10x ourselves and reach $10 billion. In order to 10x ourselves, we're going to rely on three core pillars that are unique to Lemonade. The first is having a leading brand and best-in-class product experience that have redefined customers' expectations in our industry and serve as a platform for our growth. The second is having a proprietary go-to-market machine built on learning models that constantly optimize for profitability.
The third is a differentiated and defensible go-to-market strategy for our next growth engine, Car, which is unique to Lemonade and will enable us to win and scale effectively in the upcoming years: winning brand and product experience, profit-optimizing marketing machine, and a differentiated go-to-market strategy for Car. I'll now dive into each one of those and expand. So what does it take to build a huge brand in the insurance industry? Let's take GEICO, for example. We don't mean to keep picking on GEICO. We think they're great, and we have a lot of appreciation for their brand. So GEICO has been around for 90 years. So first, you can say that it takes a real long time to build a big brand in the industry. They also have an attractive selling proposition that centers around value and savings.
And they've spent billions of dollars on marketing over decades, getting awareness and attraction from customers. Let's not forget they also have a very cute mascot. But at a fraction of the cost and in record time, Lemonade has established itself as the brand for the next generation insurance buyers. We set a new standard for insurance product experience, proving that a customer-centric approach can win. We made it easy, real easy, to quote, bind, service all the way up to getting your claims paid. And we have one of the highest customer rankings in the industry. We also control the experience end-to-end, constantly looking for more ways to optimize the experience for our customers without having to depend on any external vendors. Let me give you an example of how this works. Every week, we do about 50 code releases, optimizing the experience and changing it for our customers.
Let me repeat this: every week, we change our product 50x . All other insurance companies rely on off-the-shelf systems in order to make the changes to the products that they want. Think of the advantage that we have and see how our customers appreciate it. This is what redefining the product experience looks like. Our product experience has attracted an audience and a customer base that is representative of the next generation insurance buyers. 70% of Lemonade's customers are younger than 35. This is a great indicator of where the market is going. This is also on par with other category leaders at a similar stage to Lemonade, who built on this audience in order to become the iconic brand of their domain. If you had to bet today who will be the next-gen insurance brand, we would argue that Lemonade is the one to bet on.
We believe that we are well on our way to become that chosen brand for the next generation insurance buyers. In fact, we are already the most loved brand in our category for our target demographic. This is now driving much of our acquisition, where Lemonade Renters Insurance is ranked number two in searches. These are customers not doing generic search, searching for renters insurance, but Lemonade Renters Insurance, ranking above Progressive, GEICO, Allstate, and many others. In Pet insurance, in record time, we're now the fourth most searched brand, with the highest year-over-year growth. We did this efficiently, punching above our weight, spending 5% of what the largest incumbent spends a year on marketing. Even if you add up all of Lemonade's marketing spend throughout the year, it doesn't add up to what some of our competitors spend in one year alone on marketing.
Mindshare translates to market share, and we expect this trend to continue. Next year, we'll bring in more than a million new renters and hundreds of thousands of new pet customers, continuing to outpace the growth of these two respective markets and thus capturing more and more market share. We love these products. We plan to invest in them so that we can continue to drive this momentum. It's important to note that these two products represent a high contribution of profit to Lemonade's bottom line. But they also play a very important role in our Car go-to-market strategy, which I'll get to later. To sum it up, investing in our brand and product experience has created a strong platform for growth and expansion.
This is now driving much of our sales, with over a third of them coming in organically and getting us to the point where we have hundreds of thousands of customers signing up for our future products. These are 700,000 customers signing up for car insurance in places we don't offer car insurance. These are customers on a waitlist for car insurance, not the latest drop of Air Jordans. This is the power of our brand and product experience. The second pillar of our growth strategy is our proprietary LTV, lifetime value model, and go-to-market machine that are both designed for profit optimization. Two of the biggest challenges in acquiring customers directly in insurance are risk selection, getting the right risk with the relevant risk profile, and managing your acquisition costs in a highly sophisticated and competitive market where all companies are going after the same customers.
How do you manage your acquisition costs so that you don't lose even when you win over the customers you want? Now, imagine if you had a crystal ball that told you about every single lead, what's going to be its lifetime value, and how much you should actually pay for it. We have it. It's called our LTV model. Now, in its 11th generation, our unique model is comprised out of 50 different learning models and uses more than 3.6 billion data points. This model was built specifically for the task of optimizing our spend and enabling us to bring in the best customers. We then use this model to drive all of our marketing decisions: risk targeting, channel optimization, our spend decisions. This model also gives us key insights when it comes to multi-line customers and cross-sell capability.
As this is a learning model, the bigger we get, the stronger the flywheel and the more accurate it's going to become. So this is only going to get bigger, smarter, and better. To the best of our knowledge, we are the only insurance company that is able to use LTV modeling at the lead level on a customer-specific basis, which gives us an unfair advantage when it comes to customer acquisition. To give you a sense of how good this model is, take a look at this. This is our first model prediction for the first-year new cohort losses across all of Lemonade's products. It ends in Q3 as we need at least one year to verify the model. The dotted line represents the predicted losses of that cohort, while the solid line represents the actual losses.
It's looking exactly as you would hope, on top of each other and going in the same direction. We're not just growing faster and accelerating our growth, but we're doing it much smarter. We've been accelerating our growth for the past four quarters, and we don't plan on slowing down. Traditionally, as you ramp up your marketing spend, the trajectory of the spend curve is such that as you grow your spend, you lose your spend efficiency. The last dollar you spend is always less efficient than the dollar before. With a combination of our brand, product experience, and smart acquisition, over the past four years, we've managed to defy gravity. We've tripled our new sales per quarter at the same and an even better level of efficiency.
What I can tell you today that I wasn't able to tell you in the past is that we expect that trend to continue. We plan to continue to accelerate our marketing spend at the same level, if not even higher level of efficiency. We discussed our brand, product experience, and serve as a platform for growth and our go-to-market machine to scale effectively. Now let's talk about our third pillar, reaching 10x, which is our differentiated go-to-market strategy for winning in Car. Looking into our future growth path, as I said, a 30% CAGR world over would enable us to 10x and reach $10 billion at roughly the same amount of time it took us to reach $1 billion. You all can draw many different product paths for us to get there.
Here's one where we will continue our predictable growth with our existing product lines: Renters, Pet, Life, Home, and Car . Naturally, Car, as it represents the largest market of all of them, will need to increase its proportional share in Lemonade's overall portfolio. While this might seem outsized relative to the other product lines, at the end of this, this is less than 1.5% of the Car market. Building on our performance in Pet and Renters , where we grew to 5% and 8% market share respectively in record time, we have the confidence in our ability to achieve this. Our belief in this stems not only from past performance, but also because we are now a mature company, not starting from scratch, and Car aligns very well to our core strengths. We excel when there's high frequency of interactions with our customers.
It enables us to deliver a better experience and allows us to optimize the Car pricing and deliver more value to our customers. Our technology advantage is even more enhanced in Car, and not just in how we sell and service the product, but at the core of the product itself, with our unique approach to telematics that is unlike anything in the industry. With our market share among young consumers, we know how to target these customers, and many of them are already our customers. This is why we believe that we are well positioned to win in Car. Our go-to-market strategy relies on two groups in particular. The first channel is our customers. In fact, we can build a large Car business from our existing customers today.
The second largest group, the second target audience, is what we call next-generation drivers, who drive better than what they're being charged on average. Our competitors are Car first in a high cost of acquisition market. If you want to beat Patrick Mahomes, don't challenge him on the field, but challenge him to a game of chess. We're going to lean into the structural things that make Lemonade better and unique. Let's start with our existing customers. I have some good news. We already have a lot of them. Lemonade has more than 2.2 million customers today that love us and want more of our future products that don't have our Car product yet. Let's assume that slightly half of them have cars and pay for car insurance, and they pay around $2,500 a year. That means that they're already paying $3 billion today elsewhere on car insurance.
These are zero cost of acquisition Car customers. We are not a Car first company. We're a Car as a cross-sell, which represents a huge untapped potential for us. In fact, we can build a multi-billion-dollar business just from our existing customers today. Most important to note is this is not a static number. We plan to continue to utilize our go-to-market machine to attract millions more of customers at a low cost of acquisition. And this only gets better. While we continue to optimize our Car pricing and our offering, we have to increase awareness to our Car offering to our existing customers. This graph represents the percentage of Lemonade new renters in states where we offer car insurance, searching and looking for our Car product. We've gone from 3% to 15%, 5x in two years, and that trajectory will continue to climb.
But perhaps the best news is that Lemonade customers buying car insurance are better in every way. Our initial indicators are showing that they have better retention, 70% higher than new-to-Lemonade customers, and a far better loss ratio. And another interesting point we're seeing is that traditionally, new car drivers have what we call new business penalty, meaning that their loss ratio in their first term is much higher than at renewal. With our customers, there's no new business penalty, no additional acquisition cost, better retention, better loss ratio. This means that this business is very good business for us. And as Daniel mentioned, our strategy is to take all that goodness, implement it into our filings, and pass along all of those savings to our customers, winning them with unbeatable prices. And that's when you're going to see conversion and growth in car really take off.
Shifting gears to the second group we're targeting, how do we reach the next-generation drivers? If you're searching for another sports analogy, think of the movie Moneyball. How do we do that? In short, using telematics and being a telematics-first and out-of-the-gate insurance company where our entire experience and product is built around it. Good young drivers are paying more on average for insurance because they belong to a group that is on average riskier. We need to target the ones that will benefit the most out of the Lemonade offering and price them specifically for their risk profile. How do we do that? We use our existing book and telematics to identify who these drivers are, price for them as the good drivers that they are, rinse, and repeat.
As Daniel presented before, our own proprietary model is able to outperform for our book of business in finding higher degrees of variations. But more importantly, it identifies better-performing groups of drivers than originally perceived. As this is a learning model, with every claim, it gets even more accurate. And all of these signals and actual claims data are then fed real-time into our LTV model that I discussed before. Our LTV model, in turn, drives our marketing machine to look for similar groups across all of our markets and channels. These groups traditionally also have a lower cost of acquisition. We then acquire drivers in this group. We price for them and retain the better drivers with unbeatable prices, which grows our telematics group and then generates more data that is again fed into our LTV model. And this is how it looks like in one of our ads.
When you get car insurance, you get a price based on all these drivers. But when you get car insurance with Lemonade, you get a price based on you. If you drive less, you save. If you drive safely, you save, whatever your age. We see you, new drivers. You can even save by customizing your policy to suit you. And when you need help, we've got you. Sign up through the Lemonade app or at lemonade.com/car. Just don't do it while driving.
As you can see, we have a very different approach to winning in this high cost of acquisition market. We don't plan to outspend the incumbents. Instead, we will play to our strengths. When they're going Car first, we're going to go Car, which is, by the way, the most expensive way to acquire customers.
We're going to go Car as a cross-sell, selling to customers that we've already acquired, building on their data to acquire new ones more efficiently. They market low prices. Our prices for our target customers will be lower thanks to our leadership in telematics and lifetime value models. All that with unmatched experience on par with tech-leading brands that our customers use today. We believe that we are well on our way to 10x our business, and there are multiple paths for us to get there. This slide shows one possible path to $10 billion that doesn't even include us launching any new products, which we will.
And we have all that we need already built to get there: a brand and product experience that resonates with next-generation customers, that drives our growth, an LTV model powering our acquisition machine that continues to learn and improve as we grow, and a differentiated go-to-market strategy for Car that leans on our unique core strength. We feel the wind in our backs, and we're excited for the next chapter in our book that would no doubt also require us, as we grow, to keep our expenses and costs as low and flat as we can. To talk more about how we plan to do that, I'm going to hand it over to Adina, our COO.
Hi. It's so great to be here today. Today, I'm going to talk to you about that flat expense line that Daniel showed you beforehand. You may be wondering, what's the catch?
What's the compromise? I'm going to demonstrate the answer to you. I'm going to show you behind the scenes of our technology that we've never shown before. And I'm going to show you things that no other insurance company has done before. That's because I'm quite confident that the back of our house is just as beautiful as the front. Let's get started. Traditional insurance companies, with their decades-old org structures and fragmented technologies, see diminishing returns over time at scale. Without technology as the backbone of their culture and their foundations, they find it near impossible to reinvent themselves for the AI age. Lemonade has had the good sense of being founded in the digital era. And we have vertically integrated technology at the core of our business.
We believe that we're now at an inflection point, where the decisions and investments that we've made in the past are paying exponential dividends. These payouts have brought us to cash flow profitability and will continue to deliver accelerated returns as we outpace the burden of our fixed cost with size. The bigger we are, the more effective we'll become. We'll grow our business and become twice as efficient. I ask that you remember that dynamic throughout our presentation about the gift that we call growth and how it fuels operational excellence. Let's take another look at our IFP numbers compared to our headcount numbers. Our IFP has grown 25% year over year in the last three years. This represents more than doubling our business. At the same time frame, our headcount has only grown 2%. That's basically flat.
I've put the number of employees at Lemonade here in absolute terms to make that very real for you. I'll focus today on our people cost as the single biggest component of our fixed costs, and it's a great proxy or leading indicator for how our expenses are going to scale as our business grows. When I think about the different components of our headcount that determine our human resources needs, I can categorize them into three key categories: our customer interactions, which we need to support, the claims that we need to handle, and our corporate and technology teams that build products to better serve our customers. Today, I'm going to take you through each and every one of these categories, and I'm going to demonstrate to you how they're optimized with technology.
One metric that's helpful when you think about the efficiency of a company is the ratio between a company's top line and its headcount. Today, Lemonade is just shy of $1 billion in IFP, and that ratio of IFP over headcount for Lemonade is $730,000. As we double our business, which will be more or less when we reach net profitability, we will already rise past most traditional insurers, including the likes of Progressive and GEICO, companies who at this point in time will be 20 x our size but less efficient. These companies tend to get confined and stuck in this efficiency zone as their ability to scale effectively diminishes over time. But you know what? This isn't our peer group. We're aiming higher.
As we double our business again with really conservative assumptions regarding any future efficiency gains, we expect to be still smaller but more efficient than some of the best companies in technology. And we continue down that $10 billion path that Maya just described to you. At this point, our headcount numbers are perhaps more speculative and vague, but the principle still holds. We expect every employee at Lemonade to make $4 million in IFP. Our foundations are well set up for infinite scale. Wait a minute. These charts seem too good to be true. At what cost is this happening? Is Lemonade simply the budget airline that's always delayed and doesn't serve peanuts? The answer is a whopping no. We must be the most customer-obsessed insurance company in the world. None of these technologies come at the detriment of our customer experience.
Maybe in an airline, there's a compromise that you need to make between legroom and cost, but not with technology. We're able to serve our customers first-class champagne. And I'm allergic to peanuts anyway. It's true. Our human intelligence has stayed flat, but our total intelligence continues to overflow as we use AI to amplify human performance. Since last Investor Day, we estimate to have already saved $120 million by using AI and automation across our businesses. We expect to save an additional $1 billion in the next four years. Our IFP over employee ratio is high. That's true. But we've not compromised on our quality. And our customer satisfaction scores and employee satisfaction scores are the best in the tech industry and miles ahead of our insurance competitors. Let's go deep inside. The first component that determines our headcount is our customer interactions, servicing. Insurance is a high-interaction business.
Customers need to talk to us all the time: before, during, after a quote. Our competitors employ tens of thousands, tens of thousands of people to support their customers, and you know what? This is the nature of the business. Engagement isn't bad. It's an opportunity for us to increase loyalty and ambassadorship towards the brand. Just this year, we're on track to solve almost 2.5 million tickets. If we improve the automation rate of our ticket handling, this gives us the opportunity to delight our customers at a lower cost. In this section, I'm going to demonstrate the backend that we've built that increases AI and automation in a scalable and compliant way, but I'm also going to show you real dollar results. Before I take you behind the scenes, let me orient you a little bit about how customer operations works at Lemonade.
Since our inception, we've been using AI in servicing. So the basic principle holds: a customer reaches out, and the ticket gets routed and classified to either be handled by a human or handled by AI, in which case it'll be handled by our AI agent, Maya, distinct from the Maya who you just met but definitely inspired by her personality. With recent developments in AI, we've been given the opportunity to handle more complex tickets with AI, which means our AI component goes up and our human component goes down. In order to onboard these new AIs onto our system, we've developed a cutting-edge AI platform within our Blender system that I'm just about to introduce to you. In doing so, we've also developed new roles and responsibilities within our organization.
Our once customer support specialists that know everything about supporting other humans have now become specialists in training and composing AI models. Our AI supervisors use our Supervisor tool, which you're just about to see. They act like coaches, and they grade every single interaction that Maya, our bot, generates for customer tickets. If anything falls short of magnificent, they fail this interaction and provide feedback for our AI for future learnings. Our AI trainers use our Playground system, which you're also just about to see, to compose and refine the prompt logic that Maya uses in order to answer customer tickets. They then test out this logic on a sample set of customer tickets. As the precision rate improves, these tools graduate into production use and then are used on real customer tickets.
This part may be a little bit easier to understand when you see it for yourself, so let's take a look. We're transitioning into a live system now, so bear with me as it's not as clean as a presentation, but we really wanted to show you the real system. OK, this is Blender, or a subset of Blender, our proprietary back-office platform that we've engineered in-house since day one, built by our engineers but used by almost every single function at Lemonade. I'm showing you a subset of Blender now that is used for our customer support tool generation, whereas, in fact, this is the same tool that's used by almost every team at Lemonade: quotes, sales, claims, telematics, LTV, payouts, internal operations. If GEICO has those 600 systems that Daniel mentioned beforehand, we have just the one, and it's really easy to integrate new technologies onto it.
This is Supervisor. Here is where our AI supervisors, once customer support specialists that have no engineering background, take a look at tickets that AI Maya has generated for customer queries and grades them. We're going to put on our AI supervisor hat and take a look at a few, shall we? Help me out, please. OK, Mia is moving to a new place. She'd like to update her policy. Maya responds, "Hey, congrats on the move. We can definitely help you with that. You can use this link to create a quote for your address. Then once you've got your new coverage sorted, schedule a cancellation." Great. "Don't worry. We'll issue a pro-rated refund for any coverage you've already paid for." Fantastic. So that was a really good comprehensive response. Maya even provided more information than Mia requested, so I'm happy with that.
I'm going to go ahead and pass it. Let's take a look at another. Sarah would like to know if Maya is a real person. So we actually get this question quite a lot, and Maya should self-identify as AI. Let's see what she's done here. "Depends on what you consider real. I am indeed AI, but I'm powered by superhumans that ensure I answer everything precision, speed of light." Very cute, Maya. "Best of both worlds." Let's go ahead and pass that. OK, Aaron would like to insure his parrot. I know we do not insure parrot, so let's make sure that this answer is accurate. "Big fans of all animals here at Lemonade, including your parrot. At the moment, we only provide policies for cats and dogs." A bit of a bummer, agreed.
But we're always looking to expand our offerings." A little bit of optimism for the future. "Great tone of voice, Maya." I love that. Let's pass that. OK, Carly would like to insure her engagement ring. Yay, she said yes. "It's a gold ring from Tiffany's. What's the fast way to get coverage?" Maya says, "You can easily add your engagement ring by using this link. We'll keep you posted and do an update." I'm actually going to take this opportunity to pause for a second and explain to you something about this link. This link is not just a general how-to FAQ kind of page. By clicking on this link, this is a DIY link. This is a do-it-yourself link.
What will happen is that it will take Carly deep into her account within our application and enable her to add this piece of jewelry to her policy by herself, meaning the AI doesn't take this action on her behalf. What I want to do now is actually click us through this link and have experience this from a customer standpoint. What would happen from Carly's perspective? Internet a bit slow. We'll give it a sec. OK, "Add extra protection for your valuable items, more coverage, higher limits, deductible fee." Sounds good. OK, sorry, keep zooming in and out here. We'd like to cover our engagement ring, right? So we said we had a new engagement ring from Tiffany. I know the name of the model: Eternity Ring. What a beautiful ring. Purchased from Tiffany, I wish. Let's price it as $980.
OK, "When do you want to upload a receipt for your item?" Oh, I didn't realize I needed one, but luckily I can do so later, and Maya says that she'll remind me if I need to submit a claim, so that's fine. I guess I can continue with the flow. OK, so it looks like I can submit this item for under a dollar and then add extra coverage. I can see that my total price will now be $24.92, so I've got all the information at hand. I'm going to go ahead and submit it, and that all seems to be in order. OK, so this process was fine. I managed to add extra coverage to my ring within a matter of seconds, but to be honest, I wasn't 100% happy with that interaction.
First of all, it's not like Maya's personality not to congratulate Carly on her engagement. She's usually, you know, that's just plain rude. Second, it would have been nice to get a heads-up that I would need to have an invoice at hand. That way, I wouldn't have to have that little bit of inconvenience that I didn't have it. So what I'm going to do is I'm going to go ahead and fail that interaction. Let's see. "Poor turn." "Where are your manners, Maya?" You should have congratulated Carly on her engagement, right? Maybe next time she'll improve. Now, at this point, if I'm a real AI supervisor, I'm moving on to my next ticket. This is a production line, and I have quotas for today.
But amongst this very intimate and trustworthy audience, what I'm going to do is I'm going to take us actually deeper into our Playground system so we can go ahead and modify the tool itself. Let's go ahead and do that, shall we? All right, taking off my AI supervisor hat and putting on my AI trainer hat. This is our Playground system, still within the same Blender platform. Here is where we actually build our tools that enable our bots to generate responses. What we're going to do is we're going to take a look at our extra coverage ticket, which is the one that we were just using. Here we can see a description of the tool. This is where customers, active renters, can use this to add extra coverage for high-value items.
Here is a set of questions that we've delivered to our tool to sort of get a sense of what customers may ask when they want to use this tool. And then we have some classifiers. These are certain rules about how Maya should respond. Should she answer it? Should she escalate it to a human? Should she perhaps use a different tool such as this based on various different sets of criteria? And we also have a set of instructions that tell her how she should respond in terms of composing the answer. Here on the right is the playground. This is why this system is called the P layground. Here we'll actually simulate it with real tickets and see how Maya will respond before we graduate these into production. What we're going to do now is we're going to add some rules so that we can improve Maya's response.
The first rule that we're going to add is when a customer adds a valuable item, remind them to have a proof of purchase and value at hand, like a receipt, right? It was OK. We did the flow, but it would be nice to have it at hand. The second rule we want to have is more like a condition because it doesn't always occur. If the customer mentions a life-changing event, say an engagement, then react with empathy or excitement. Whoops, help me with spelling here. Whatever applicable. Great. OK, so now what we're going to do is we're going to take the same ticket that we just reviewed and see if Maya can answer it better now. Add extra coverage ticket. OK, here it is. I'd like to add my engagement ring. Yay, I said yes. Remember, ring from Tiffany. Hey, Carly, congratulations on your engagement. Woohoo!
Such exciting news. Let's talk about getting that beautiful ring insured. I already love this tone of voice much better. This feels much more like her. We'll need some proof of ownership and proof of value. Fantastic. Great, works. But we're not going to publish just yet. I just want to simulate a couple more. Add extra coverage. OK, let's simulate this next one. I was gifted a new bracelet from my grandmother who passed away. OK, this is the exact opposite. Not happy. I'm sorry to hear about your grandmother. It sounds like the bracelet she gifted you holds a lot of sentimental value. I shouldn't be smiling now. It's a simple process. You can add your new bracelet to your policy, heads-up, proof of ownership. Exactly perfect. I love how she listened to all of our instructions. We're going to go ahead and publish those changes.
And from now on, Maya has a new and improved way of answering extra coverage tickets. Let's go back to where we were, shall we? This isn't just cool AI tech. It is cool AI tech. But it's not just cool AI tech. It has driven real business impact that no other insurance company is doing. Our overall customer service automation rate has increased to over 50%. I want you to notice the scale on this chart. It's a single year with a drastic queue-by-queue improvement. We launched this AI platform within Blender less than a year ago, and already it's solving a third of our tickets. These numbers are already stale. In Q4, we've already demonstrated a lot of improvement. Side by side, our average handling time for handling tickets in minutes has decreased steadily, queue-by-queue as well.
All of these result in a steady decrease in our cost to serve, for which we can see true dollar savings. Our services costs continue to go down as our growing business improves our models. The bigger we are, the more efficient we become, remember? Since last Investor Day, we estimated to have saved $35 million in driving down the unit cost of our service channels through AI and automation. Extrapolating that to 2028, that number will be more like $250 million. And this extrapolation is only for servicing, only for existing channels, and only for existing technology. And we all know that technology and AI are going to get compoundingly better. AI delivers the full package: accuracy, efficiency, and experience. I'm going to demonstrate that through talking about the second component that drives our headcount, our claims handling.
In essence, we're in the business of paying claims. People buy insurance so that if something devastating and unexpected happens, we're there for them. Paying what we owe, but also doing so in an efficient way, are both huge drivers of our long-term profitability. There's a certain hierarchy between these three components. Claims handling is a bigger opportunity than servicing, but a much less obvious implementation. I spoke to you about the high-frequency interaction, high-interaction nature of the insurance business. With claims, it's not just about the frequency of these interactions. It's also about the intelligence of them. In claims, you have claims adjusters and legal counsels that need to review medical records and surveyor reports. There's a lot of regulatory and compliance considerations, and we need to make sure that our AI implementation is a gazillion%.
It's really complex, difficult to nail, and has a huge upside to grab. Our Pet business is powered by technology in every facet. So it's a great product to take a look at when we're thinking about claims. Because of the frequent claim nature, pet has an accelerated feedback loop, working in dog years, if you will, that showcase rapid iterations of our technology and give us a glimpse into the future on how other high-frequency products such as C ar will operate as they're at a higher scale. More claims give us more data that feeds the machine, and so on. Our Pet business grew 20x in less than five years. This is due to our tech stack that enables us to deliver products and coverages unbelievably fast. Side by side with this rapid growth, our loss ratios in pet have decreased every single year from 110 to 69.
This is due to our ability to select the right customers predicted by improvements in our LTV model and coupled with smart data science-driven pricing techniques. I'm going to double-click on that for a second. Here you can see how our LTV models predicted cohort losses compared to how those cohorts delivered in practice. Now, Maya showed you a very similar chart earlier, but she showed you the chart for a business as a whole. I'm referring to our Pet business in isolation. The next chart is simply incredible, in my opinion. In May this year, we launched a new data science-driven pricing technique. Since we launched it, our conversion rate, which is represented by the pink line, increased. At the same time, our loss ratio relativity, represented by the gray line, decreased.
This means that we're becoming more attractive for profitable customers and less so for those that are more risky. This is exactly where we want to be: rapid growth, smart growth, and efficient growth. At the same time, we managed to decrease our variable cost of handling claims from $65 to $19. That's a 70% reduction. I'm going to pause for a second and take a deep breath so we can take this in, because this is a really big deal. 20xing the business, 40 points knocking off the loss ratio, and 70% more efficient. How did we do this? I'm going to demonstrate the answer to you. I'm going to show you a single customer filing a pet claim through our app, coupled with a visual abstraction of our technology working in the background. Beforehand, I showed you our Blender system, the back office, operating the brain.
What I'm going to do now is attempt to show you the brain activity itself. Let's take a look. This is our app. Here's where our customer, Emily, is going to submit a claim on behalf of her dog, Rufus. So she's going to go ahead and do that. At this point, our AI, Jim, starts having a conversation with Emily to gather basic data about the claim. And Emily tells us that Rufus has been sick. But luckily, he's feeling a bit better now, so we're all very relieved. Emily tells us that Rufus has been coughing. Now we need to assess whether or not we can cover the condition called coughing. We need to make sure that there isn't any pre-existing condition in Rufus's history that would mean prior to the purchase of the policy that would prevent us from covering for this condition under this existing policy.
In order to do this, our AI, Jim, takes a look at all the medical records we have at hand, up to 12 months long of medical records, which can be extremely long, 100 pages long of documents. In doing this, our AI needs to decipher vet summaries, take a look at medical prescriptions, test results. This is an extremely tedious process that Jim can do quite easily. For a human, this could take hours, if not days. Other insurance companies charge for this service. But here, our AI is able to do so in a matter of seconds. Jim hasn't found any pre-existing conditions in Rufus's medical history, so we're able to move on to the next phase of the workflow. Now Jim wants to make sure that all the dates check out. So he asks Emily for some additional information and starts placing it on the calendar.
When did Rufus get sick? When did Emily take him to the vet? When did we file a claim that we know? We're also able to place the policy purchase date because we have that in our system. Still no humans in the loop. All the dates check out, so we're able to move on to the next phase. We ask Emily to upload a copy of her invoice so that we're able to see what she'd like to claim for. This is classic doctor's handwriting. Imagine trying to decipher this yourself, but our AI can decipher all of it. First of all, our AI determines that it is indeed an invoice. Then it extracts only the relevant information from the invoice and matches it one by one with coverage that we have in our Blender system.
Only once every single invoice item finds a match of a coverage within our Blender platform, and only within the limits of payout that are defined, only then can Jim go ahead and approve this claim. It checks out. Fantastic. Claim approved. At this point, Jim can communicate to Emily that the claim has been approved and that the money is already on the way. This is obviously a great experience for Emily, and she's over the moon as she's managed to receive payout for her claim just within a matter of seconds. But sometimes it doesn't all happen 100% perfectly. Let's rewind for a second. Let's imagine that, for example, one of the things AI couldn't achieve automatically. For example, one of the invoice items didn't find a coverage as matched within our Blender system.
At this point, our AI, Jim, solicits help from Mike, who's a pet claims adjuster who specializes in pet illnesses. He's not a random match. Mike uses the same Blender platform that you just saw, and he's able to manually match the CT scan to a coverage that we have in our system. After he does that, he hands the baton back. Mike is able to manually approve the CT scan and match it with a coverage that we have in our Blender platform before handing the baton back off to Jim to then communicate the claims decision and payout to Emily. This took a little bit longer than the instant claim that we saw beforehand.
But still here, our AI, Jim, and Maya were able to collaborate to provide a near-instant experience for our customer. I just showed you one claim. At scale, this has incredible impact. Our Blender system enables autonomous claims to flow seamlessly in and out of manual and automated processes, completing tasks and removing periods of uncertainty for our customers. This means that even the most complicated claims can be significantly automated. Do you remember those low-cost airline peanuts? I'm still allergic. But here you can almost taste that first-class champagne, the instantaneous, delightful, and impeccable experience in the front, coupled with the depth and sophistication of the brain in the back end. With technology, you really do have the full package.
Amazing efficiency gains cost a third of our original cost five years ago, and our customer satisfaction scores of 4.7 out of 5 and our Net Promoter Scores of above 75 are unparalleled in the insurance industry, and I want you to note that these stats are being collected for both paid and denied claims. I haven't forgotten the dollars. Zooming out across our businesses, we've saved over $60 million in claim handling expenses since 2022 through AI and automation in our claims handling. We expect to save nearly 7x that much by 2028 to $410 million in loss adjustment expenses avoided. This also directly contributes to our consistently improving loss ratio performance. The last component of our headcount. This component is the least trivial of them all. This is the fixed component of our headcount covering our insurance core product teams.
The same component that stays fixed as our growing business outpaces it on the path to deliver unparalleled expense ratios. Infinite scalability means touching every single part of the organization. You're only as efficient as your weakest link. In order to explain how we've kept this flat, I'm going to talk about something that isn't typically talked about in forums such as this: the people, the human element. Culture matters. Culture can be a competitive advantage. Bad culture can be a strategic disadvantage. At Lemonade, our culture is technology, and it's ingrained in our DNA. The fast exponential growth of technology makes it really difficult for companies to maintain their competitive edge. As advancements accelerate, the time between innovation collapses, meaning even the most cutting-edge companies need to constantly evolve or they'll risk falling behind.
It's like a ginormous wave that keeps coming at you, and you're either going to surf it or you're going to drown. At Lemonade, our culture isn't one where we hire people to do specific jobs with rigid job descriptions. We hire an attitude. We hire resilience, growth mindset, curiosity. Today, you're a customer support specialist, but tomorrow, you could be an AI trainer. Your roles and responsibilities need to constantly evolve as we onboard new technologies onto our platform. Employees at Lemonade aren't threatened by the wave. They're exhilarated by it. Take our head of procurement, for example. She has a team of two humans, but she has countless bots working for her, amplifying her performance, transitioning her into some sort of iron woman with superpowers that enable her to do more with less. At Lemonade, we don't use AI only to support customer-facing business functions.
We use it literally everywhere. Employees use AI to replace every single mundane, repetitive, error-prone business process, amplifying their performance, but at the same time, cultivating a deep understanding of these technologies and ingraining it within their DNA. Some of you have met Cooper, our internal automation bot, and he's the third and final bot that I'm going to introduce to you today. With Cooper, we managed to automate 95% of our IT tool requests, resulting in $1 million in savings last year. Cooper helped us achieve 100% automation in our procurement purchase renewal request, resulting in $3 million in savings last year. Cooper helped us reduce the invoice approval time in finance to a single day, even in performance management.
Cooper selects 360 feedback from a select number of employees and then takes those inputs and synthesizes it into a summary that our manager reviews before delivering the feedback. It also issues a monthly pulse report to every manager at Lemonade and asks them, "How are your employees doing?" and includes that information as well into this review summary. Why am I telling you this? Part of maintaining expenditure discipline is making sure that every person is performing at their best. This may have felt a little bit overwhelming for you. I get it. We've been at this since 2015. It feels like a lot of AI because it is. I have hundreds of examples like this. I just showed you the tip of the iceberg of what our technology can do. We're just scratching the surface.
And when you're constantly surrounded by innovation and technology and inspired by how it can boost your performance, you can't help but becoming an innovator yourself. From day one, we sought out to build a different type of company, one that has less people and more code. Our competitors have hundreds of different systems bolted onto one another through series of acquisitions and mergers. If they want to make use of any new technology, they have to seek help from third parties and then become dependent on them and their roadmaps to have lengthy, dependent-heavy implementations. Our singular Blender platform knows everything there is to know about a customer. It knows what campaign led a customer to us. It has a single system of record of every single interaction the customer has had with us.
It's able to flag risk factors or opportunities for upsell into our models for future learnings and predictions. Today, I showed you a glimpse into the most inner parts of our technology, and you got to get an appreciation of the interconnectedness of its components. Let's zoom out. Today, I demonstrated to you how the biggest part of our fixed cost, our headcount, stays flat with technology while we deliver a champagne-like product to our customers. Zooming out even further. Lemonade's skill is demonstrated in the flywheel between our growth and our operations, our top and our bottom line. The more efficient we are, the more savings we can pass on to growth in the form of killer prices, which in turn boost the conversion, adds on more customers that better feed our machines that make us more efficient. The bigger we are, the more effective we'll become.
In a second, I'm going to hand over to Nick, who'll talk you through the financial metrics that measure and exemplify this business dynamic, and he'll simplify how you can think about Lemonade in financial terms. To sum it up, we've built an organization that has a bring-it-on culture and a technology stack that delivers unparalleled efficiencies at scale. Lemonade being the sum of its exceptional parts, we're well set up to deliver infinite scale for our next 10x. Th ank you so much.
Hi, everyone. It's great to be here. Time to talk about your favorite topic, financial metrics. We've heard lots of great material today from all of our presenters. As if that wasn't enough information, we typically disclose over 100 metrics each quarter.
Each of these adds texture as our business is multi-product, multi-regional, and unique in our application of technology to the insurance business that has its own important and nuanced metrics. But I'd like to help you digest all of that information. I'll provide a framework to monitor our business that I hope you'll find simple and practical. As you deepen your understanding of our model and perhaps look to build your own, I'd like to recommend that you prioritize your focus on just two metrics: Adjusted Gross Profit and Adjusted Free Cash Flow. Tracking those two rather conventional metrics will give you a great sense of how we're doing as a business. Now, we're not taking any disclosures away. In fact, we just shared new, helpful, granular info in a supplement to the Q3 shareholder letter and will continue doing that.
But today, I'm focused on synthesizing all of the information and orienting you to what we find most important. Why those two metrics? Well, it's essentially how we manage the business and captures all of the beautiful work Maya and Adina just presented. Adjusted Gross Profit is the money we're left with after growing the business and paying variable cost, and Adjusted Free Cash Flow is what we're left with after paying fixed costs. That's the full picture. Those two metrics capture growth, retention, loss ratio, automation, efficiency. Adjusted Gross Profit and Adjusted Free Cash Flow are the output metrics by which to measure the execution of our strategy to grow the business and scale the operation. Our model can be represented as a cycle that builds to those two critical metrics. The flow begins with capital injection, which comes from our own cash flow generation Synthetic Agents.
Synthetic Agents program is an innovative growth financing structure that has unlocked capital-light growth. It's rather unique and enables us to turn $1 into $5. I think that magic deserves the spotlight, so I'm going to come back to it a bit later in my presentation and do a deep dive. Capital flows through to new business acquisition. Example: we write a check to Google. Cohorts of customers then stack on top of each other, yielding a growing book of business. We pay to cover our customers' claims and other variable costs to get to Adjusted Gross Profit. We then pay expenses and are left with Adjusted Free Cash Flow, which flows back to the beginning of the cycle and allows this flow to continue. I'll go over each of these steps in a bit more detail. I'll start with our investment in growth.
We invest in new customer acquisition with attractive unit economics to maximize total new business. As you can see, and as Maya covered, we notably accelerated the investment pace in 2024, and we were able to do so while preserving attractive and stable unit economics as measured by a 3x LTV to CAC ratio. Growth spend generates a cohort of new business. Once that cohort is on the Lemonade book, we retain and cross-sell that business to maximize the size of the total book. This chart shows a new versus existing breakdown within Gross Earned Premium. As you can see, the existing book has scaled really nicely in the past few years. This cohort stacking concept is critical to our path to profitability, as each cohort added drives its own discrete stream of gross profit that flows to the bottom line.
After growing the top line, we pay claims and variable cost, and the gross loss ratio is the key metric in the variable cost structure. You can think of it as our COGS. Our trajectory has been stunning since we last met two years ago, as we have now realized our target range in the low 70s. This is particularly impressive due to the mix shift to newer product lines that has taken place during the same period. In the past couple of years, there was a notable tailwind as regulatory approvals for rate caught up to inflation. In the next phase, technology will continue to play a pivotal role. Automation will enable us to deliver best-in-class LAE ratios (that's the cost of managing claims) and we'll leverage our ever-smarter AI models to become increasingly granular and precise in underwriting.
An important note in our rationale for showing a range in the forecast, as you can see here, the loss ratio is simply an input. We are not pursuing the lowest loss ratio. Rather, we're pursuing the highest gross profit. We constantly optimize for pricing, given the trade-off between conversion rates and the loss ratio. Premium less variable cost yields Adjusted Gross Profit. We focus on Adjusted Gross Profit because it's the best measure of profitable growth. We think it reflects everything that matters: customers, premium per customer, the total premium base, the loss ratio, other variable cost. Unlike premium or revenue, which can paint a partial picture, gross profit shows you if we're growing sustainably. If we were to grow unprofitably, premium or revenue would rise, but gross profit might be flat or decline, and we don't think that's attractive.
This metric holds us to a higher standard: not growing for the sake of growth, but growing to create true profitable value. The growth rate of Adjusted Gross Profit in the past three years has been over 50%, and we're expecting over 70% growth this year. You might hear the phrase "shrinking to excellence" in insurance circles, as it is a notable challenge in this industry to pair sustained growth with profitability. But we have been able to grow the premium base rapidly alongside an improving loss ratio, and as a result, gross profit growth materially outpaces premium growth. Next, we incur expenses. We've seen fantastic progress on all of our expense efficiency metrics in the past couple of years, and the driver is consistent: automation, which has been a powerful lever for the business.
Because Synthetic Agents finance our growth spend, and I'm building to a cash flow view, I'll review operating expense excluding our growth spend. I'm expressing OpEx here as a percentage of earned premium, and as you can see, since we last met in 2022, we've seen dramatic improvement in this ratio, and we're expecting that trend to continue. How have we done it? Well, look at the track record. This chart shows OpEx over time next to Gross Earned Premium. As you can see, for nine straight quarters since before our last investor day, OpEx has not grown, and that's alongside a consistently scaling book of business. After incurring expenses, we're left with Adjusted Free Cash Flow, the best representation of our business activity in aggregate in a period, the net impact of all of the inflows and outflows.
This is a new metric we're introducing today that we'll continue to disclose moving forward. It is the cash flow metric that's the tightest fit with how we manage the business. As Daniel mentioned, in Q3, all types of cash flow were positive, whether you're focused on net cash flow, free cash flow, or cash flow from operations. In our definition of Adjusted Free Cash Flow, we include cash flows from our Synthetic Agents program to capture the true cash impact of our growth spend in a period. Since those inflows and outflows are directly linked to core business activity in a period, this approach provides a clear and comprehensive view of cash performance and operations. Adjusted Free Cash Flow breakeven is here. As you can see, we're expecting a positive year in 2024.
This materially precedes P&L profitability due to the working capital dynamic that is built into our model, where premium collections precede claim payouts. Cash flow performance has powered strength in the balance sheet. Due to Synthetic Agents and faster-than-expected growth, we now expect to end 2025 with nearly double the cash and investments balance as we expected when we last met in 2022. We have $1 billion of cash and investments on the balance sheet today, expected to grow with positive cash flow generation. So our balance sheet is very nicely poised to power the next phase of the business. I've touched on Synthetic Agents a number of times now, and I wanted to spend a couple of minutes on what it is, how it works, and why you should care about it.
In short, it's a novel and creative financing structure that was custom-designed for our growth strategy and has proven powerful for the business. The mechanics are that our partners advance us 80% of growth spend upfront, and in exchange, we pay them a fractional stream, 16%, of premiums that they helped finance for a finite period until they're made whole, typically two to three years. Incumbent insurers, including Liberty Mutual, Allstate, and State Farm, all leverage agents to drive their growth, as agents bear the cost of customer acquisition in exchange for commissions. But with traditional agents, there are a couple of notable downsides. First, there is disintermediation between the customer and the carrier, and second, obligations to pay commissions to agents are perpetual in nature. Well, neither of those facts apply here.
We simply call the program Synthetic Agents because we have mirrored the benefit of the traditional agent model in efficiently funding growth without any of the downsides. Let's work through an example to paint a clearer picture. I'll start with a case where we invest $100 in growth through our direct channels without Synthetic Agents. We're $100 out of pocket, and it takes two years to realize payback. That's quite a capital-intensive strategy. By the end of the customer's lifetime, we've generated $200 of net positive cash flow. That $200 equates to LTV minus CAC. So in this case, we have $300 of LTV generated per $100 of investment, and there's your three-to-one ratio. But I'd like to suggest a different metric to measure marketing efficiency.
IRR should be a metric that is very familiar to many of you, but for those who aren't, it estimates the annual rate of return of an investment. We find IRR to be a useful metric, as it captures nuances related to cash flow timing that are very important in this context. Let's analyze this investment through an IRR lens or a CAC IRR. Without Synthetic Agents in place, our growth spend typically yielded a 56% IRR, already a strong return, I'm sure a few investors in the room today would pass up. As I mentioned, Synthetic Agents finance 80% of our investment, allowing us to buy $500 of growth while still deploying just $100 of our dollars upfront. In other words, we're able to turn a dollar into five. This unlocks a multiplier effect for our capital efficiency and a whole new level of growth potential for the business.
Have a look at CAC IRR. It has doubled to the triple digits. We think that's incredible, and I hope you agree. This program gives us a menu of options in growth capital planning. We can ramp up growth, as I've shown here, or the other extreme is that we can leave investment levels unchanged. Now, with the same $100 investment, we are only out of pocket $20, significantly reducing the upfront cash flow gap. By the end of the lifetime, you can see we have slightly less net cash flow generation, and that difference represents the cost of capital transferred to the synthetic agent over the life of the customer. Again, you see the same IRR unlock, as this business is now generated with a fraction of the upfront capital outlay. Those two examples are the ends of the spectrum.
Invest 5x as much to maximize growth or leave growth investment unchanged and maximize cash. In reality, we choose to play somewhere in the middle. In 2024, we more than doubled the pace of growth investment and will continue to put our foot on the gas next year and beyond. Let's have a look at the impact of new business on our two critical metrics before and after launching this program. In the 12 months prior to deal launch, we generated $20 million of new gross profit, but have a look at cash flow. New business was a significant drag. In 2024, our first full year with this program in place, we have significant growth in new gross profit, and the cash flow drag is gone. The impact to our business since launching this deal is significant.
Have a look at this chart that compares Adjusted Free Cash Flow to Adjusted EBITDA and net income. I'm highlighting when we launched the deal with the dotted vertical line. As you head to the right of this chart, post-deal launch, we have an accelerating pace of growth investment. Remember, that's an investment with a triple-digit IRR, and we should do as much of that as possible. But have a look at EBITDA and net income, which become increasingly negative as higher growth spend burdens the P&L. Adjusted Free Cash Flow, on the other hand, which captures the fact that 80% of our growth investment is now financed, shoots ahead of EBITDA and into the positive. In most businesses, the P&L leads cash flow, and investors are rightly focused on the P&L. Our business is the other way around.
EBITDA and net income will follow Adjusted Free Cash Flow into the positive, and the driver of lag is simply the accounting treatment of our growth spend. It's our perspective that with this divergence, cash is king. We focus on Adjusted Free Cash Flow for the same reason EBITDA became popular in the first place: to capture core business cash flows and to adjust for discretionary investments like capital expenditures. For our business, growth spend is the new CapEx. 56%-112%? That sounds too good to be true. So what's the catch? Well, there isn't one. We've gained this efficiency while adding exactly zero risk to the business. The Synthetic Agents program is designed so that our partners' commissions are solely linked to premium streams that they financed. If those premiums aren't realized, the Synthetic Agents bear the downside risk.
My fellow model junkies in the room and online will be pleased to hear that right now, Shai will tweet an Excel file that captures the math behind this unlock. The analysis will use a simplified approach and only relies on information we've already disclosed. I hope you'll find it interesting or helpful or both. Coming back to the flywheel, let me now work through an example that resembles 2024 using rough directional numbers, and I'll build from growth investment all the way to cash flow, covering each of the steps we just reviewed. I'll highlight in pink as we build. Synthetic Agents finance 80% of our growth spend, seen here as an inflow of $96 million. We then invested $120 million in growth, which generated 3x the premium, and as we earn that premium, it translates to $180 million of new business impact.
As cohorts stack, the renewal business contributes $820 million, and the total premium base here is $1 billion. We pay claims and variable costs to yield $200 million of Adjusted Gross Profit at a 20% margin. We pay expenses, and to get to Adjusted Free Cash Flow, we need to include impacts related to working capital, Synthetic Agents, and CapEx. So to conclude this illustrative case, $1 billion of premium yields $200 million of Adjusted Gross Profit at a 20% margin, and with the expense structure shown here, positive cash flow. What happens when we double the business to $2 billion of premium and maintain 20% margins and stability in expenses? Would cash flow double? No. Cash flow would multiply by 20x . We will be generating hundreds of millions of dollars in positive Adjusted Free Cash Flow. Let me say that again.
The next doubling of the business will unlock massive cash flow generation. So in 2024, we more than doubled the pace of growth investment, which helped drive over 70% growth in Adjusted Gross Profit and our first Adjusted Free Cash Flow positive year. The model is humming really nicely and is materially de-risked relative to where we were just a couple of years ago. Our CFO, Tim, will walk you through what it looks like when you simply draw a straight line and extrapolate some of these trends to the future. He will share some thoughts on the path to profit, our view of the next phase of the business, and some thoughts on valuation. Thank you.
Lemonade's remaining path to profit is getting awfully short. Squint a bit from wherever you're sitting, and we're essentially there. This shows our remaining path from an EBITDA perspective as compared to our Gross Earned Premium, with just a few quarters remaining. If you layer in the cash flow benefits that Nick just walked us all through, you'll see why we are confidently planning for both growth acceleration and no change in our pace of progress to break even and beyond. If you look here, you'll see our actual progression since 2021, which is quite something. In broad strokes, in 2021, this ratio was in the negative 60s. A year later, 50s. Then below 30. This year, likely below 20. And next year, we expect the teens, and by 2026, in the single digits, on its way to positive. Steady as she goes.
Now, all that said, we have been listening to our investors and our potential investors, and candidly, there remains a bit of skepticism about this path to profit, or so we've heard. So I'd like to hit that question head-on with a few different takes on what drives our confidence, and perhaps we can make our confidence your confidence. First, this view, you've seen this in a few different ways and a few times today, one proof point of our consistent scaling over eight-plus quarters. Flat expenses with every marginal dollar of gross profit closing the gap to profitability. Wash, rinse, and repeat. Essentially flat operating expenses outside of our growth spend in a period where we've accelerated top-line growth from the teens to the high 20s. And going forward, as you've heard, we expect our growth rate to head towards 30% with modest expense growth.
Second, we've highlighted the benefit that visibility into our Adjusted Free Cash Flow can provide as a leading indicator of how the P&L is developing. Adjusted Free Cash Flow leads, followed by EBITDA, followed by net income in a fairly orderly fashion. And it's mostly just math. It's hard to begin to generate cash at an increasing rate without EBITDA and profit to follow. Two years ago, with a fair distance to cash flow positive, it was a little easier and more reasonable to ask the question, when are these lines going to cross zero? When does any of these lines cross zero? Today, the cash flow view is now changed, and with it, our confidence in this path.
We can see here as free cash flow becomes positive for the full year in 2024, followed by EBITDA by year-end 2026, and then followed by net income in the following year. Third, let's look under the hood a little bit, and we can see that we're essentially a profitable business today, absent our increasing growth spend that's increased our growth rate. This investment into growth is wholly discretionary, and if we halted that spend entirely, we'd be EBITDA positive now in Q4, and even just spending at a nominal pace of growth rate to just maintain our top line, we'd result in a break-even business next year in 2025. The underlying business, where cohorts have been added one on the other, gives us additional confidence in this trajectory, and finally, a fourth, consistent and reliable execution.
Two years ago, we stood on this stage and told you a few things about our expectations. We said we expected to grow about 20%. We'll hit 26% before the year is out, we expect. We set a long-term gross loss ratio target of about 70%. Last quarter, most recently, we just reported 73%. Right on track with that target. Each quarter since going public, we have spoken about our expectations for all of our key metrics, and we found ourselves in or above our own guidance range every quarter, 17 out of 17 quarters. Two years ago, we did a long-term model, and we looked out, and we expected a trough, a low in our total balance of cash and investments of about $600 million. Today, that number looks more like about $1 billion, right around the level we're at today.
We told you we planned and expected to drive continuous efficiency. Lo and behold, a 70% increase in our top line, our In Force Premium per employee in just those two years, with no plateau in sight. Finally, perhaps most importantly, we told you we expected cash flow positive right around 2025, and in fact, we've achieved that a year ahead of schedule in 2024. Hopefully, this gives you a bit better feel for what drives our confidence in that path. So here we are, just days away from 2025. How are we thinking about the coming year? On the growth side, our consistent growth acceleration has been notable. Beginning in Q3 a year ago, 2023, we were growing about 18%. A quarter later, it was 20%. Then it was 21%.
Then it was 22% and 24% in just the most recent quarter, and we expect to be on track for 26% growth in Q4. We expect this acceleration to continue into 2025 and expect to reach our cruise growth rate of about 30% in 2026 and for the foreseeable future thereafter. All of this would take us to an IFP of about $1.2 billion in 2025, with gross profit growing significantly faster still. And on the cash flow side, solid free cash flow, continued nominal expense growth as we planned and on track as we planned since 2021 for EBITDA break-even before 2026 is out. Let's look out a little bit further. Two years ago, we laid out our expectations for the next several years. We did a multi-year model that we shared with folks.
The difference, if we look at our updated version of those plans, is pretty dramatic. A 50% increase in our top-line growth rate from 20% to 30%. A 25% increase in our expected 2027, looking out a few years, In Force Premium to about $2 billion. Despite a generational inflation shock that rocked the insurance sector and other sectors over the last few years, essentially no change in our expected timing to EBITDA break-even by the end of 2026, with a bigger business growing at a faster pace at that time. Studies have shown that the vast majority of CFOs and CEOs believe that their shares are undervalued. Big surprise. To be fair, it's not my job to estimate the value of our shares. It's really yours.
That said, I'm going to share a few thoughts on valuation that we've derived from a couple of our more thoughtful investors, and maybe just look at a couple of different ways about how you might think about a framework for valuation. So let's start out heading towards $1 billion in Gross Earned Premium, likely in early 2025. Let's grow that at, say, 30%. Does that number sound familiar? For seven years, optimistic and doable. Let's apply a, let's say, a 12% EBITDA ratio to that top line seven years out, and then maybe a 14 x EBITDA multiple. You've probably got a multiple you like better or a methodology you like better, but let's just go with it for the moment. Maybe we'll discount that back to today using a rate that you like. We come up with an answer of about $90 a share.
Now, I get it. In your head, you've got your own valuation metrics with your own multiples and your own discount rates and your own approach. We did the same thing. We did it a number of different ways and came back to a similar range each time of between about $80 and $100 a share. Now, we don't say this so that you'll write it down and take it home. We say this so that you'll appreciate that we think about running our business day-to-day and not so much about valuation, but once in a while, it helps to throw out some of our thoughts. So where does all this leave us? Everything that you've heard today. From my view, a couple of things. A notably de-risked plan, an unchanged strategy, and a resilient investment thesis.
The industry, as we all know, is ripe for disruption, and I believe no one is ahead of Lemonade in disruptive potential. One thing I say a lot is we've done the hard part first. A market-leading product and an unbelievable user experience. High growth, reliable operating leverage, a tip-top balance sheet on a clear path to profit, and a market leader to boot. In closing, I've been doing this a long time. I've been a CFO for 25 or so years. I think I'm on earnings call number 75, so it's been a while. I can honestly say I have never been more confident of the next phase of the business I'm in the middle of than now. If I can maybe reprise a couple of thoughts that Daniel shared, it goes something like this in my head. We are structurally advantaged.
We are a new kind of insurance company, and it is, without question, a prize worth fighting for. Thank you.
Thank you, Tim. By way of closing, I'd just like to share a few thoughts. I'd like to kick it off on behalf of Shai myself with a big thank you. Thank you to my fellow speakers. Thank you to the team who helped us format, edit, gather, fact-check, and prepare the presentations that you saw today. Thank you to the broader Lemonade team whose amazing work it's been our privilege to showcase today. And I want to thank all of you, our investors and our soon-to-be investors. As I said at the outset, we value your time, we value your attention, and we hope these have been amply rewarded. I said that I hope you find the proposition investable.
I'd like to try and encapsulate it, distill it down to just three propositions really simply in order to try and put some structure around what you've heard today. Three propositions. The first one is, in insurance, nothing is more important than precision and automation. They map onto expense ratio and loss ratio, and that's the whole nine yards. Proposition number one. Proposition number two, nothing does automation and precision better than technology. I find that kind of inarguably true. Proposition number three, Lemonade is the best in technology in insurance. It's as simple as that. That is the proposition that I hope you find investable. We find these truths to be self-evident. What that does, that technology difference that you saw manifested in so many different ways today, that manifests as a competitive advantage, a moat, something that is deep and defensible. It is structural. It is cultural.
And that gulf between us and what others can do is growing, not shrinking. I'd issue a small warning, a kind of a buyer beware around this. All insurers use some of the words that we use today. All of them will talk about AI and apps and telematics. You might even hear machine learning thrown in here and there. When I hear everyone is doing AI, my answer is simply, no, they are not. No, they are not. Some of you will have been to presentations by other insurance companies. I would wager that nobody has seen anything remotely like what you saw today. Nobody else is doing AI. George Bernard Shaw said that the United Kingdom and the United States are two nations separated by a common language. The mere fact that we use the same words doesn't make us the same. So do look beyond the slogans.
Lemonade didn't discover AI in 2023. I showed you our decks from 2015 when we were founded. Artificial intelligence and technology are not for us some pixie dust to be sprinkled on our quarterly earnings. They aren't an afterthought. They aren't a graft that we're desperately hoping the body won't reject. You saw today AI Maya. You saw AI Jim. You saw Cooper. You saw how our people train them. Wranglers of AI. You saw our 11th generation of LTV itself, an amalgam of 50 other AIs, and its incredible prescience as we showed you historical results mapping onto actual predictions, and I hope you understand the profound difference, multi-layered profound difference between what we mean when we say doing telematics and what that term usually means to people. All of this amounts to a change in degree that truly amounts to a change in kind.
A tech company doing insurance is a different species than an insurance company doing tech. Take note also of the pace of change. We met in November 2022. I showed you at the time foundation models and GPTs, but ChatGPT launched only two weeks later and burst into everybody's awareness. GPT-3 was ranked at about the level of a fourth grader. GPT-4 is a high schooler. OpenAI's latest model, o1, which is still in preview mode, is already performing at PhD levels in many, many areas. In the meantime, during those same two years, the cost per million tokens has dropped 99%. This is just staggering and world-changing stuff that's happening. We see no empirical or theoretical limit to that or to its impact on our business, and we were absolutely born for this moment, as I hope you've seen exemplified time and again throughout the day.
It's not merely that we're running faster than our competitors. It's that we are running on a conveyor belt that is itself accelerating to dangerous speeds. It's hard to jump onto that if you weren't there in the first place. Jack Welch cautioned incumbents at large with the words, "If the rate of change on the outside exceeds the rate of change on the inside, the end is near." If ever those words were true, they're true of insurance today. You are investors. You are always looking for investment opportunities, and all of this AI and its deep transformation has attracted a lot of investment. For good reason, NVIDIA has done what it's done, and so have other AI-related stocks. But I put it to you that if you're looking for alpha today, you can't look for it in the same places that were discovered two years ago.
In fact, I think a lot of the value creation is going to shift from the foundation models to the application layer that harnesses all of that goodness in order to do business-transforming things in the real world. In that context, no one is wielding AI in insurance the way Lemonade is. And if you're going to harness AI to slay any dragon, insurance is a big dragon as you're going to find. I hope you find that image investable. And as that image sets up shop in your mind, I'd like to thank you once more for your interest in Lemonade and to wish you all a wonderful day. Thank you so much.