Root, Inc Investor Event Transcript, Nov 18, 2026
Root, Inc. (ROOT)
Root's partnerships business grew 19% year over year last quarter, and the business is roughly 50-50 between direct and partnerships. Root is in 39 states and expects to be broadly national by the end of 27, excluding Hawaii and Alaska. Its embedded channel now makes up over 50% of new business. AI summary
Conference Transcript - ROOT 2026-11-18
Jeff Lick, Analyst — Stevens
Good afternoon, everybody. This is the 115 session. I'm Jeff Lick. I'm the auto ecosystem and consumer analyst here at Stevens, joined today by Root and CEO and founder, Alex Tim. We don't cover Root, but I'll just give you a quick, you know, backstory of how we've come to develop a relationship. You know, we do cover Carvana, which is one of your partners, and we'll get into that, and we also cover Copart and RBA that are involved in the salvage business, which is a big part of the claims process and profit maximization of insurance, and also LKQ, which is a big buyer of salvage units, and we've gotten into just the Kind of the opaque nature or the, is that of the insurance business, but just how, how it works. And so we had the opportunity, uh, at an event to meet Alex and his team and got to talking and like, you know, you've got an interesting story and we'd love to have you come join us here. And that, so that's how you got here. Uh, so Alex, thanks for joining us. Um, you know, it'd be great to start off with like your. You're a founder of an insurance startup, auto insurance startup. That is not a small industry you're messing around with, and you've gained great traction. You're profitable now, very profitable, actually. So maybe just kind of give us the backstory about what it is that you decided needed to change, because I think that still is the mission, and just how you went about that and what you're trying to do to be different.
Alex Timm, CEO
Absolutely. Thanks for having us. I mean, really, when we started the company, What needed to change was everything, basically, and everything was changing. So when you talk to customers about car insurance, they don't say good things. If you type in car insurance is and let Google sort of autofill for you, it depends whether or not you have SafeSearch on or not, but it's pretty bad what they're saying out there. I've never talked to somebody who said, oh, I get to go shop for car insurance. This is going to be such a good time. And so why is that? And most of our competitors are over 100 years old. And so that was the first problem was just consumers really don't like this product. The second part is the fundamentals of the product were sort of perfectly positioned for what we believe to be like fundamental change in total, based on technology. And so a big part of that has been data, right? As an insurance company, what we do, it's our job to price risk, which means we have to most accurately match each individual with what is the likelihood they actually get into an auto insurance accident. And, you know, you look at the industry, historically, that's been done through linear models, and it's a data prediction problem. And data itself fundamentally changed. Now we're all walking around with a supercomputer in our pocket called a smartphone. And now vehicles are connected, and the structure of that data has changed because now it's, you know, instead of me getting a credit score where it's a number between 1 and 850 or something like that, I'm now plugging in. We're getting 16 data points a second on our customers through a mobile app. Well, how do I make sense of that data? Well, now it turns out there's all these other great things called machine learning algorithms and massive increases in computational power that totally changed my ability to actually leverage and use that data. And so that's where we were 10 years ago. I'm an actuary by training. I was working in the insurance industry. And it's like, okay, consumers fundamentally do not like this product. It doesn't work very well for them. It takes 15 minutes to purchase a product online. By the way, it's a good thing. It's like an ad slogan. Even though nothing takes 15 minutes to buy online, even products you want to buy, let alone car insurance. And then two, we can use data science, modern data science technology to just completely change the way that insurance is priced. And so that's what we. We started. The size of the price, as you mentioned, is massive. It's a $300 billion, over $300 billion industry right now. Auto insurance is not really touched by. And so I think when Andreessen Horowitz wrote, you know, software is eating the world, we say, well, he, you know, choked on insurance because it still hasn't eaten insurance. And so we looked at that and we said, well, there's just a massive opportunity here. And the world's continuing to change. And so that's, you know, our flagship product, the first product we launched was meeting consumers on a mobile phone. That's where consumers moved. We created a product where customers never even have to touch the keyboard to actually get an auto insurance quote, and then we can leverage all of the rich behavioral data off that mobile phone to more accurately price risk. And that was the original concept. That product, by the way, is still around today. We've certainly grown and changed over the years, and it's been quite a journey.
Jeff Lick, Analyst — Stevens
So, when you... Think about, you said the product has changed. We talked about at ICR, that conference, I was like, hey, I've got five drivers in my family, and we talked about some of the data. And I was like, well, if you describe some of the data you'd get, I'm like, maybe I'll just stick with what I've got. But how much of your policies and force are actually driven by that data? Because I know you do kind of, you'll give people the option to under, hey, you're just a normal piff like everybody else, as opposed to, hey, you're someone that's getting ripped off because you actually are, you live in the wrong zip code, but you're a great driver. What's the breakdown there?
Alex Timm, CEO
Well, on our direct business, it's the vast majority of our customers have telematics data. And so if you come to us directly, almost everybody goes through a test drive period. We also have an embedded platform. And that's been a really nice growth area for our company, where we're beginning to embed directly with vehicles. And that's both during the point of vehicle purchase, like with Carvana, as well as points of vehicle ownership. And as we've expanded that platform, a lot of that actually is not telematics rated. We find we don't need telematics in that channel to be profitable. There's other advantages that we're relying on, and that's probably – and right now, the business is roughly 50-50, 50% direct, 50% partnerships.
Jeff Lick, Analyst — Stevens
And in terms of the advantages on the partnership side, are you still offering a better value, or is there some other way that you're competing?
Alex Timm, CEO
Well, the biggest way we're – so this was interesting. Interesting. So when we created this three-click purchase experience on a mobile phone, we realized, and really Carvana was our first partner in this, we realized that that three-click purchase experience was really powerful in order to sort of meet consumers in that moment of need during that vehicle purchase process, and we could remove a ton of friction. So it used to be when you were on Carvana, you'd go online and you'd buy the car, you'd be excited, you found the Mustang you like, and then it'd be, oh, now you need proof of insurance. And it was, okay, I have to remember my Geico login or my Progressive login, get a PDF download, upload it. It was a pain. It was the number one friction point in their funnel. Uh, we, uh, they, they also experimented with ads in that spot, uh, which was even worse because now I'm clicking out and going through a 15, 20 minute flow and then trying to get back into the car. So it was a mess. Through that platform, because we have a modern technical platform that we've built in terms of like an API on our rating engine, we're actually able to ingest all of the data that a consumer already puts into the process through purchasing a vehicle on Carvana to offer basically a bindable quote within three clicks. So now, instead of having to go through that, it's here's the car, here's what it costs, you want insurance included, click this button, and that's it. And that like more than four or five X attached rates of insurance on Carvana. And so that's been the competitive advantage we've used within that channel.
Jeff Lick, Analyst — Stevens
And are you rating that consumer through DMV data and state points in the same way?
Alex Timm, CEO
We pull all of it. So we've always pulled all data, DMV points, credit score, other proprietary scores and data that we've actually developed that we don't disclose. We just shipped our newest and most predictive pricing algorithm a couple months ago. And so, yeah, we're using all that traditional data plus more.
Jeff Lick, Analyst — Stevens
And so, at the end of the day, then, like, what would keep? Carvana, could Progressive come in and cut a deal with Carvana as well and do the same thing, or what's the competitive advantage there?
Alex Timm, CEO
Well, there's a couple. The first has been, and we know this very well why the big guys can't do this and haven't done this. The first is that most of these competitors, most of the technology that they've created, they created online about 30 years ago, and they don't have, I mean, mainframes are still in place. They're still working COBOL programmers. It's very, very difficult for those companies to actually technically seamlessly embed inside of another environment. And that's the first hurdle. So it would take a lot of technical investment to actually get to the point where you have an API in a rating engine to actually create three-click bindable quotes. The second piece is actually an incentive problem and a risk-reward problem. And we know this very well because we've tried to work with larger insurance companies too as like an agency because we're not national yet. So we've said, oh, wouldn't it be great if we could just act as an agency and some of these states are not national yet. Not only would the product, the three-click purchase experience not be – it wouldn't be a thing because of technical reasons, but the chief underwriting officers don't like any of this at their companies because they have had the exact same. We're in the Q&A process. So over the last couple of years, I don't think we know that the case is coming from. But I think that's a very recent message, some of the comments will be asked if you and I have no idea what they did, because I don't know what this is. And so the chief underwriting officers oftentimes have said, we're not doing that because we don't know what happens if we start to collect data in a significantly different way from the thing that our business has been built on.
Jeff Lick, Analyst — Stevens
I should mention, if anyone has any questions, just feel free to raise your hand and jump in. In terms of, you know, so the Carvana relationship's been pretty successful. I think you've over 250,000 policies. You know, what has that experience proven about, you know, embedded insurance? And, like, could you go elsewhere, find other partners? And, you know, where have you been trying, and where do you think that might be some stickiness?
Alex Timm, CEO
Yeah, I think, you know, it's interesting. The biggest thing in the U.S., for a long time, it was thought you can't sell insurance at the point of vehicle sale. Dealers don't like it. People don't like it. Consumers won't do it. Carvana sort of disproved that. It was really the first time that you saw a significant attach rate for insurance at the point of vehicle purchase. So we've now been replicating that. It turns out OEMs really want this. Insurance is roughly 18% of the total cost of ownership of a vehicle. They control every other piece. They control financing. They have captive financing. They control fuel efficiency. They control cost, all of that. They don't control insurance costs. And that's a huge issue for them when insurance, for instance, went up 20% year over year in 2022. A lot of people blamed their OEM. They called their OEM. They were angry. So we've actually, we just launched a pilot recently, last couple of months with Hyundai, and now it's Hyundai Secure Plus, powered by Root. You can see their ads. It says, you know, 10-year, 100,000-mile warranty and six months free insurance, you know, covered by Hyundai. And so we've now started to embed directly with OEMs. We love that because it's not only embedding at the point of vehicle purchase in dealers, but we're also getting closer to the vehicle. So that's then allowing us to actually pull telematics data directly off of connected cars as well, because the OEMs control that technology. And so it gets us even closer to, you know, understanding risk even more granularly.
Jeff Lick, Analyst — Stevens
And that's on the direct side, correct? You mentioned you're not in every state, and you also, you know, you use agents or, you know, independent agents. You know, walk us through a little bit of that growth process and where things stand.
Alex Timm, CEO
Yeah, so, I mean, fundamentally, when we created this, like, three-click purchase experience that's super easy to embed in other software, we started to ask, okay. Where else would this apply? And a third of the markets, over $100 billion of revenue, is actually still going through independent agents today. A lot of people thought the internet was going to kill independent agents. It hasn't. They've evolved because it was also a third 50 years ago. And so what we've discovered is that, hey, if we just embed inside of these independent agent experiences too, they want a three-click purchase experience. Right now, the average time if you call an agent on the phone to just get car insurance, the average time of that phone call is about 25 minutes. With Ruth, it's about two and a half minutes. So we've learned that was really important for agents because it makes them far more productive. So we've been appointing these agents and entering into this channel now, and it's like hitting singles. Every time we appoint more agents every single week, we grow. We appoint more agents every week, we grow. And so we've got that flywheel going, too. So our distribution has totally changed. We have Direct, we have our embedded insurance platform, and then now we're increasingly having an independent agency presence as well.
Jeff Lick, Analyst — Stevens
And, you know, obviously getting the customer and agreeing on a premium is really just the first step, right? So then all of a sudden, there's going to be claims. I believe last time we talked, you guys pretty much handle everything yourself. How do you compete with... Companies that, you know, have millions of PIFs, whereas I think you guys have around 480, you know, and how do you compete on a cost side and also just from the customer service perspective? Are there things that? About that part of the model that were also broken?
Alex Timm, CEO
Yeah, claims is really interesting because a lot of it feels like hand-to-hand combat, honestly, because a claim comes in. It's an independent body shop. You have to negotiate those rates. You don't know if you should be paying that claim, shouldn't be paying that claim, and it's incredibly human-driven. So depending on what adjuster you happen to get assigned at your insurance company could change whether or not the amount of claim payment you're going to get, plus or minus 30%, 40%. With AI, that's totally different. So what we've done is we've built our own in-house claims platform, and it allows us to effectively create agents that are always looking at all of the data associated with a claim and saying, okay, what's the next best action? And what we found is that not only do you get a lot of benefit from automation there because you move faster, so for instance, our turnaround time on claims is much better than the industry average. But also you reduce that variance in your losses. And so you eliminate loss leakage. And so Claims is like perfectly situated for an AI application. It's just data and then trying to figure out what should I do next with that data to optimize for an outcome. So it's a giant optimization problem.
Jeff Lick, Analyst — Stevens
And so are you using AI to build almost an internal CCC or Mitchell? Mitchell?
Alex Timm, CEO
So we don't really use a CCC or Mitchell in terms of the software. That is all root. What CCC and Mitchell have that is interesting is that they have data from the industry. That we're not trying to rebuild. So we do use CCC, for example, to say, hey, in the zip code, what's the average cost of chiropractics? They have that data set, and that's great. I think they have a network effect there because it's so contributory and everybody shares it. So we're not rebuilding that element to it, but we're not really interested in the software elements to it because the software elements we've just found we're better at building it ourselves.
Jeff Lick, Analyst — Stevens
And then when you get to the point of, hey, we're either going to fix this or we're going to total the car, how do you guys think about that? Is there room for improvement in terms of how the industry works there?
Alex Timm, CEO
Well, it's actually rapidly changing. The way you usually do that is it's an economic analysis. It's a machine learning algorithm that basically says, okay, how much damage is done to the car? Is it drivable? Isn't it drivable? And it says, okay, this is what I think I'm going to spend if I repair it. If I total it, here's what I think the average. Value of this vehicle is, and so therefore I can, you know, I'll just pay it out. And that's it. It's a pretty simple. What's changing now is there's actually now companies that are AI companies that consumers are hiring to basically negotiate on their behalf as to what the value of that vehicle is. And by the way, it's also happening on legal claims where they're starting to use AI to like generate legal demands. And so what you're seeing now is like this war of algorithms of like, well, okay, your algorithms try, you know, the consumers using ChatGPT or another service to try to get a higher payment, we then have to have, you know, data on our end. And that's where the negotiation begins.
Jeff Lick, Analyst — Stevens
It's interesting. For example, there was something on J.D. Power. They just did a survey about customer satisfaction about, hey, if I file a claim and it's a fix, customer satisfaction is remarkably higher than the total loss. And so I'm wondering, do you put any factoring into, hey, if we total this car, The person's going to be disappointed versus if we fix it. So is there kind of a soft economic analysis there?
Alex Timm, CEO
Absolutely. And you can look at retention. And so, you know, obviously there's an economic analysis, particularly on, hey, if I repair this, I'm more likely to retain the customer. So we do favor repairs all in. You know, see, some of our partners are like Carvana. If you get a total loss and you have Carvana insurance, we actually tell you, hey, here's five different cars that you can replace it with at no cost to you because the insurance payment would cover it. And Carvana can deliver that car to you.
Jeff Lick, Analyst — Stevens
And in that regard, you're actually almost, you're partnering with Carvana, and Carvana's effectively a marketplace.
Alex Timm, CEO
Yeah, and Carvana will offer a $500 discount for that, and so it's nice. It keeps the customer in the Carvana ecosystem. So Carvana likes it. The customer likes it, too, because it's, oh, I can get this car that looks like my other car and can be here right away.
Jeff Lick, Analyst — Stevens
And is that for only customers that have arranged insurance through Carbonef with you, or is that your entire platform?
Alex Timm, CEO
We are actually experimenting that on the... With a direct business as well. So right now, it's definitely everything that comes through Carvana where you do that. But even on our direct business now, if you have a total loss, we will push you to Carvana to say, hey, here's some vehicles in your area that you can get delivered to you.
Jeff Lick, Analyst — Stevens
Not that I needed another reason to be bullish Carvana. We'll just kind of put that one down. Okay. And so. And then, so on the total loss process, where do you think total loss could go? Is that something that you could eventually see getting into the 30s? It could.
Alex Timm, CEO
Yeah, it could. I think it's a big opportunity. I definitely think it's also a really big opportunity for anybody trying to sell cars. Total losses are a huge. Point that they're a huge percentage of the total vehicle sales in the United States. So if you look at it, getting that loop that we talked about to work is actually really attractive for a lot of OEMs, Carvana, other folks as well.
Jeff Lick, Analyst — Stevens
You know, and so now we've reached a point where you're, you know, you're profitable generating cash. I think you've been buying back shares. It seems like you're... You're not like a lot of tech companies that, hey, we're just going to grow at all costs because we're going to get valued based on growth. But how do you think about growth, and what do you see in the confines of what your economics seem to be settling out, like where you could grow to?
Alex Timm, CEO
Well, I think we're just getting started on ultimate growth and where we are. Our partnerships business has grown 19% year over year last quarter, and so it's continuing to grow. So our direct business is more – there's something called an underwriting cycle here where you basically get these periods in time where it's very attractive to grow an insurance company and then a period of time where it's not. A lot of that is because of the lagged effect of inflation in insurance. So if you see a lot of inflation, it takes a while to increase my rates, and so then I usually overreact. So you see the cycle. And so on Direct, we've actually been pulling back, right? Because what we saw was just record levels of cost per click and CAC, driven mostly by just ad spend that we don't think is rational from a lot of players in the market. On that channel, we always are looking to just basically only deploy growth capital, provided it on a unit basis will return our cost of capital. So we're super disciplined on that. We don't just push to hit a growth number. We don't try to optimize for calendar period targets or growth. We really make sure that we're always returning the cost of capital in that channel, and I think that's why you've seen us have such a good profit record over the last some years.
Jeff Lick, Analyst — Stevens
And in terms of getting into, how many, I think you're, what, 30-ish?
Alex Timm, CEO
We're in 38 states. We just launched a couple, so I think we just launched another one. It's 39 now. And we think by the end of 27, we can be broadly national, not Hawaii, not Alaska, and then New York may or may not come in by the end of next year.
Jeff Lick, Analyst — Stevens
And so let's use New York as a case study. I mean, what's the rub here in these different states where you want to enter in as a new entrant? Why don't they just say, oh, sure, come on in?
Alex Timm, CEO
Well, I think increasingly they should be just saying, oh, sure, come on in. You know, we're making money, we're growing, we're a great option for customers. But insurance is regulated on a state-by-state basis, and there's different political environments in all of these different states. And depending on that and sort of who's in charge, they either have strong mandates to sort of increase competition in the market, provide more options for consumers, or they don't. And in some of those states where they don't, it takes longer. There's not a real incentive. For them to move quickly.
Jeff Lick, Analyst — Stevens
And what is kind of the perverse incentive at the government level that wants to keep competition out?
Alex Timm, CEO
Well, I think a big part of it, they're certainly lobbyists by all the big guys. And so we've definitely seen that where, hey, you know, the big guys want to say, don't let any of the small guys, don't let those upstarts in, you know, they're not going to follow the rules, they're, they're not going to be profitable, and they're not going to be sustainable. And so there's a lot of skepticism. That's one. And I think that's been sort of part of the perverse incentive. And then in general, I just don't think there's been an incentive. It's a lot of work. You know, so more competitors, more filings, more things to do. A lot of the departments just don't even have the capacity to review the work that's coming their way.
Jeff Lick, Analyst — Stevens
And now, how do you think about, like, so much of insurance now is based on the bundle? Because, again, it's another way to address affordability. My guess is there's economies of scale of pricing. How do you guys think about that? Have you partnered with others? And, you know, where are you at in terms of whether it's renter's insurance, homeowner's insurance?
Alex Timm, CEO
Yeah, we have a renter's product. Renters is super simple. It's basically, you know, it's a small market. You're insuring really trivial claims. Claims are very small. It's usually like 40 bucks a month. So renters is a simple product to add. In insurance, it's an easy product. You know, other products like home, much harder. Now I have to actually understand weather risk, and I have to understand some catastrophe exposures, things like that. So we don't do home ourselves. You know, it's interesting. The bundle discount is like 10%, 20%. It's more of a marketing play than it is actual economics. And, you know, if you go and get five different insurance quotes for auto, you're going to be plus or minus 40%, 50% easily, just amongst that alone. Probably the same thing on home, by the way. And so if you de-bundle, you actually end up saving a lot more money. And this is particularly interesting now in the age of AI where Muse and these companies can actually just go and get 35 different auto insurance quotes for you And look at all of the home quotes that you want and actually pair you with the best individual product.
Jeff Lick, Analyst — Stevens
Well, you've mentioned AI a bunch of times. At the end of the day, is NetNet is AI a positive for you guys?
Alex Timm, CEO
Yeah, absolutely. I mean, one, and sometimes we say it's better to be lucky than good. But when we founded the company based on modern data science and machine learning techniques, we didn't think 10 years after founding the company that, oh, there'd be this other massive step change in predictive sciences. But that's now happened with transformer models. And the amazing thing is, so we've got this, you know. Brand new technology stack that allows us to effectively deploy algorithms very, very quickly. We deploy new pricing algorithms far faster than any of our competitors. And so we have the technical wherewithal to use these algorithms. And then two, because we're a regulated insurance company, we also have all of the claims data, over a billion dollars of auto insurance claims data to train these models on. And the only way you get that data is as an insurance company, and it's really hard to start one. And so that has given us, really made us like this in of one, where we both have the technical capabilities to deploy AI in pricing and in claims, as well as the full stack infrastructure that you have to have as an insurance carrier to create these algorithms to actually produce predictive results.
Jeff Lick, Analyst — Stevens
Are there elements to your size or nimbleness or newness that give you... Well, don't necessarily give you advantage directly, but prohibits the bigger players. There's something about the bigger players that they can't.
Alex Timm, CEO
We talked to a very large carrier, for example, and they said our AI strategy has three components. The first component was software and system modernization. The second component was data modernization. And then the third component was applications of AI. And you have to do one and two before you can do three. And they said, and the first two bubbles have been the same two bubbles for our technical strategy for 20 years. And so the problem is when you have these systems that are based on mainframes, and you've got multiple, I think at one point Ajit Jain at Berkshire said that Geico had 60 different back-end systems. When you have that kind of infrastructure debt, the ability... To actually deploy these models is almost, it's basically impossible. And so that infrastructure, that like modern infrastructure and being run on a tech chassis, a modern tech chassis, is a huge advantage that we have in our ability to move quickly. I mean, for example, like over 80% of all of our code right now is written by AI. So we are starting to move very, very quickly. And that just allows us to continue to run very, very fast compared to our competitors.
Jeff Lick, Analyst — Stevens
I mean, I suspect you, some of the things that you, we talked about, what you were founded on, which was, you know, use the, you know, the zip code. The issue where, you know, things were written underwritten by zip code, but you might have a safe driver. You know, these were like, kind of blunt force examples of little idiosyncrasies. But, you know, now with, you know, and you, it seems like your business model is built on kind of. And taking advantage of these, you know, commercializing them. And now all of a sudden with AI, you know, we're all finding that there's a lot of things that seem counterintuitive. But could you give us some examples as it relates to risk, where, you know, what we've discovered is this is more predictive of risk, or, you know, we always thought this predicted risk, and it actually really doesn't. There's so many.
Alex Timm, CEO
I mean, it's, and by the way, it's... Maybe it's intuitive, maybe it's not. It's kind of crazy, though, when you think about it. If you don't have, like, tickets or a driving record, the number one thing predicting your auto insurance price is your credit score. So, like, if you just take a step back for a second and ask, okay, do I really think credit score is going to predict whether or not I'm going to get into an auto insurance accident over the next six months, turns out it's a pretty weak predictor. Then you think of things like your cell phone data, your behavioral data, your social media data. And increasingly, the data coming directly off of your vehicle, whether or not you're driving home from a bar at 3 in the morning or whether or not you're driving home from work at 3 in the morning, those things matter much more. And yet, when you look at the industry, still less than 20% of insurance policies underwritten today actually use that data. And so it's remarkable that that just is not actually, so the gap there is gigantic. And now with AI, what we're doing is we're actually like automating this research. We, you know, you have an agent go out, collect data, automatically test data. These things can actually do what we call like auto research on data science, where they will test new. Whether predictive methods, whether it's a gradient boosting model or random forest, if it finds signal, it will immediately move that into a filing and say, hey, let's go file this with the state. There's even an agent that mimics a state's objections. So it acts like a state regulator and says, this is what the regulator is going to say to you, and this is how you should already adjust this. And they're like adversarial. I mean, you can watch the agents talk. They're like adversarial with one another. And so that all becomes an automated, closed risk loop where you can start to collect data, train data, look at new models, and go very, very quickly. And that's just totally different than the way it currently works.
Jeff Lick, Analyst — Stevens
Do we run the risk? It seems like in the old way that kind of call it the analog way, there was a cohort of people that were paying too much, and there was a cohort of people that probably shouldn't pay more that were benefiting. But if we were all rated perfectly, do we run a risk that there's just a significant portion of society that just is not going to be able to afford insurance, and there's going to be another portion that finds out that, hey, you know what, it's 400 bucks.
Alex Timm, CEO
Yeah, I mean, but think about that. Right. So that's absolutely the case where it's like the worst drivers, like the worst 10 to 15 percent of people should be paying like five times the amount that they're currently paying. But think about that, what that actually means. It means they're really likely to go cause an auto insurance accident in a car accident and like kill somebody, hurt somebody. Like, I'm not sure it's good for us. It should be punitively expensive for those people to get on the road. I think that's good for all of us.
Jeff Lick, Analyst — Stevens
And then kind of as we start to close up here, you know, in a lot of your most recent transcripts and presentations, you've used kind of the phrase, the next phase of profitable growth. Can you explain what you mean by that?
Alex Timm, CEO
Yeah, so... We think we're at a pretty strong inflection point right now, and the reason for that is because we've been working very diligently and very hard on expanding our embedded platform, getting our partnerships channel to where it is. You can imagine working with these large OEMs, how long that takes to get products into market, and we've consistently, and now it's now over 50% of our new business, And we think that that product actually creates differentiated access to customers because we can uniquely meet them in those periods of time. And so we think that that's part of the foundation. The other part of the foundation is our loss ratio and our command over pricing and really being able to come in as one of the price leaders. And then expanding out nationally. We think that those three components, as we become better at pricing, create this differentiated access to customers through our embedded platform, and then expand nationally, that sets this foundation for what we believe is like that inflection point and that next phase of driving significant profitable growth in the company. Any other questions before we wrap up? Go ahead.
Analyst
Yeah, just a quick one. So Alex, can you compare the pricing and loss experience of the embedded channel versus the direct channel?
Alex Timm, CEO
In embedded, it's actually better. So one of the nice things of the embedded channel is that you're not just winning because you're the cheapest. So in the direct channel, you're talking to people who are shopping for car insurance, so they usually get three, four quotes, and you win when you're the cheapest. In embedded, they're buying it because it's easy. And so you get this favorable selection effect from the channel itself.
Jeff Lick, Analyst — Stevens
And one last question that popped up. We've analyzed the insurance industry from the point of view of analyzing Copart and IAA. And one of the first things that jumps out at you is half the PIF is mutual. And as listening to you, you're a capitalist, you're a for-a-profit organization. It's always struck me as counterintuitive. It's such a big industry that is so complicated with such thin margins and outcomes is actually run as a not-for-profit. Is that just in itself a huge structural advantage? I mean, it just seems like as we get, AI is going to, you know, like how hard is State Farm sitting around thinking about these issues if it's not for profit?
Alex Timm, CEO
Right. I mean, by the way, mutuals should be studied because it's just fascinating they exist in a capitalistic system. Like, you know, if you said you wanted to go start an insurance company and get billions of dollars of capital to do so, but oh, by the way, you know. You're not going to own it if you put the billions of dollars. Who's going to fund that? But people at one point did. And so now you have these 100-year-old companies that have significant market share that are not for profit. You'd assume, okay, I don't have to make money. Maybe I should be leading on price. But that's not the case. In this case, competition in capitalism has clearly won because the mutuals are dying off. They've been shrinking market share considerably.
Jeff Lick, Analyst — Stevens
Yeah, I mean, Progressive has been growing. PIF has been kind of, it's around 210, 220. It's not like it's something that's growing by 5% a year, yet Progressive has been growing PIF by 20% a year for a couple of years now.
Alex Timm, CEO
Yeah, State Farm is no longer the number one auto insurance company, and it had been for decades. It's fascinating. It's weird that these things exist. But also, to your point, because they're not competitive, we're seeing them sort of die off.
Jeff Lick, Analyst — Stevens
Awesome. Well, thanks very much, and everybody have a great rest of the day.