Carvana Co. (CVNA) Earnings Call Transcript & Summary

Hey, everybody. Just a reminder, today's discussion may include forward-looking statements within the meaning of the federal securities laws, which are subject to risks and uncertainties that may cause our actual results to differ materially from such statements. A detailed discussion of the material factors may cause actual results to differ. Forward-looking statements can be found in the Risk Factors section of Carvana's most recent 10-K filed with the SEC. Forward-looking statements and risks in this presentation are based on current expectations as of today, and Carvana assumes no obligation to update or revise them, whether as a result of new developments or other. Thanks so much for joining.

All right. Mike, thank you for that. So welcome, everybody. I'm Ron Josey, I cover the Internet sector here at JMP, as you know, and thrilled to have with us Ernie Garcia, Co-Founder, CEO of Carvana, which has been an amazing ride. So Ernie, welcome. Thank you for coming today. We have you for a full hour, which is exciting. [indiscernible] all sorts of things.

Thank you. Glad to be here.

[Operator Instructions] So with that said, let's start off.

So Ernie. Let's see, I wanted to sort of try to break the ice a little bit with some word associations and sort of first thing that comes to mind. And let's see. So first thing that comes to mind is just what was your first car?

Well, so first of all, I'm very anxious about this section of the programming as I feel like any psychologists in the audience are going to learn all about me. But my first car was a Toyota 4Runner. And it was awesome. It had dice in the mirror. It had dice in the mirror.

What was the year?

Oh, man, I don't know. Give me a second. It was probably -- I think it was like a '93, '92, something like that.

It's a great -- I know it's a good model year. It had dice in the mirror? It's probably going to last...

It did have dice in the mirror. Yes. Yes.

That's awesome. Okay. Let's see. What would be the first thing you do once COVID sort of lifts and everyone is vaccinated?

I think concerts. I'd like to go back to concerts.

Yes. It's a good answer. That's fair. Let's see. If you could name a CEO that you most look up to, that is not named Garcia. In other words, not your dad.

Oh, yes. Okay. Well, that would not have been the answer, just so we're clear. Deep, deep in the list. I like -- I'm going to try to answer as a way that's not so obvious because there's obviously like a short list of people that you could pick that I think are very straightforward, where there's a ton to learn from and a ton that we're trying to emulate right now. But someone that I'll give credit to is the CEO of DriveTime. When we launched out of DriveTime, I feel like I learned a tremendous amount from him just because he was like so unbelievably different in the way that he approached the world than I am. It's -- he's very kind of bottoms up and very kind of like views the world through individuals instead of kind of like tops down kind of like macro strategic view. And I think I just learned a ton from him about how to manage an organization. So not anyone that anyone know. His name is Ray Fadel, but I think I learned a lot from it.

That's really interesting. Knowing how to manage, it's hard. Managing people is hard. Let's see. If you had a choice, maybe it's an and versus an or. But when I say what do you optimize for, convenience or price?

I'm going to say -- I'm going to take neither the and nor the or. I think comfort. Like I think the customer being comfortable in the transaction and feeling at ease all the way through, I think, is the most important thing.

Okay. Let's see. Personally, a year in the pandemic, you've changed what about either your life, your home? Have you bought a Peloton? Anything changed dramatically a year into the pandemic?

I think we've all learned -- I'm going to try to keep this in like the Carvana realm, but I think we've all learned how to try to manage like a business with people in hundreds of locations without being able to go to those locations or without going -- so it's like -- and I think that's been a really steep learning curve. But I think it's been super interesting. I think we've learned a lot of things that will stick with us. Something that's interesting is like once you start communicating digitally, like digital communications scale very, very well. And so I think we're doing a lot more of that internally than we were doing before. And I think that we should have done -- with like the benefit of hindsight, we should have done more of that ahead of time, like before pandemic. And I look forward to being able to kind of supplement that with being in personal locations again, but I think that -- I think there's been some real like fundamental gains.

Yes. It's been pretty fascinating. Maybe the last one, you get this one a lot, but Amazon's marketplace strategy is blank.

I got to be careful with that. There's a lot [indiscernible] No. I think...

I have 1 more [indiscernible].

Unbelievably interesting. There's so much in there. Like I don't know how to answer that concisely. I think there's so much in there in terms of the business model itself, the business model applied to their products, what it's meant for them and the feedback it's created. So I think there's a lot in there. But interesting is my total cop-out answer.

That's good. And then last one, average year of -- average years of car ownership, going higher or lower? Where do you think that goes?

I think lower. I think my simple model for that is -- and I'm having to pick a direction. I don't think -- I don't know if that's the most important input to like the Carvana equation, but I think that people would like to swap out cars. I think that, that desire is what sits underneath the popularity of the idea at least of different subscription models over the last 5 or 6 years. I think the economics of subscription models have been tough because of the switching costs. But I think you can kind of think of switching costs as being like the cost structure that dealers traditionally have to transact in a car. And I think that to the extent that can be lowered and like the experience frictions can be reduced, I think people would like to switch cars more often, whether that's in the form of a subscription-like structure or just trading in their car and buying a new 1 more often, I think that's their underlying preference set.

Yes. That's great. That's interesting. I think all that is coming up. I agree. So let's get in the meat on the bones, and I'm sure a lot of people want to know background and everything else, and maybe we'll try to smoosh in this first question. In prepping for our conversation, I went back to our initiation report, which was in May of 2017. Stock was at $11. Our price target was $17. But who cares, right? That's that. What struck me odd is that -- or interesting was that Carvana had 8,000 vehicles in inventory at the time. And the company sold 11,000 vehicles in that second quarter in 25 markets. And you fast forward to today, in January, I think the number in the letter was 9,500, like 9,500 vehicles, and the company sold 72,000 vehicles in the quarter across 266 markets. So just putting in the realm. So 11,000 vehicles in 2Q '17, 72,000 in 4Q '20, 25 markets, 266 markets. That speaks to the demand that you have been able to grow, that Carvana has been able to grow in just 4 years. A lot of hard work's got into it. There's no doubt about that. So maybe can you talk about bigger picture with that because it's hard to get our arms around it. I'll never forget talking to Mark -- with Mark Jenkins, CFO, maybe a few years into it and just a conversation around managing 100% growth is very hard because you need to continue going at it. So anyway, for those not familiar with Carvana, can you just -- or that -- or mainly know you for the vending machines, whatever, I would love to hear sort of your take on the mission, your consumer value proposition, how you view the opportunity going forward and sort of what you needed to build a company like Carvana that fundamentally, in our view, changed the used car buying experience. So massive growth and how have you done that.

So let me start with this. I think like the -- one of the ways that we try to think about like any business, I guess, the like mental model we bring into all this is that constantly underneath the surface, customer preferences are changing just as a result of new opportunities that customers have in every industry, a new experience that they have that generate new desires in them. And then I think also underlying technology is changing. And I think that when you've got a given customer problem, in this case, selling cars, I think the best answer is a function of basically like what's the best intersection of technology and customer preferences available at the time. And I think it's super interesting when you look at automotive retail, that basically for like a 75-year period from the '30s up until roughly when we launched, there were like all kinds of superficial changes to the way that automotive retail works, but there weren't super deep changes. And so I think that's like interesting. You guys ask yourself why because if you start from this place of kind of believing that the market is efficient and that businesses automatically get like generated and migrating the direction with customer preferences, you would kind of expect there to be a lot of movement. But I think for all kinds of interesting historical kind of accidents, the business was kind of defined by massive fragmentation and many, many small players that didn't necessarily have a lot of first-party capabilities. They were using a lot of third-party systems. And so I think that, that made it so it was very difficult for the market to evolve. And then I also think that because of the underlying complexity of buying a car, which I do think is like a much more complex transaction than most transactions that consumers partake in and certainly most transactions online, which I think has all kinds of interesting implications, but because of that complexity, I think that like the divide from the old way of doing things to a new way of doing things, whether better or not, was just like this very long path to try to like navigate across and get to the other side. And so our view is that, that's kind of -- those fundamentals led to like the stagnation in automotive retail. But then that stagnation came paired with like when you have many, many sellers selling largely undifferentiated products with largely the exact same cost structure, like if they want to make money and to be able to transact in a way that's economic, they're going to need to find a way to monetize the transaction as fully as possible. And basically, the entirety of the competition in the market moves in this direction of like how do I monetize each customer better than my neighbor because that's the way that I survived. And I think that, that lack of variation, paired with that need to economically survive kind of in the context of lack of underlying business model variation, generated like the backroom and putting people in a box and selling them thousand dollars of products they don't understand because that required that you monetize the transaction more fully. And so our view was not to sell cars online. Our view was what are preferences, what are customer needs, what are technologies, and now how do we, in a way that is actually fundamentally different, kind of create a new business model that serves those needs and creates real variation and experience and then, hopefully, real variation in expenses as well, so now we don't have the need to move to the box, right? We don't have to have this back room where a customer walks out with a stomachache not understanding what's going on. And we thought that if we could do that, we could build a lot of demand. And I think a byproduct of all the things that are necessary to build that is that it is very hard. It requires a complicated transaction platform that is more complicated than most e-commerce solutions. It requires an underlying supply chain that is first party that now becomes the new best answer under the context of there being a site unseen transaction platform that kind of can work at all. And so we had to build all these different things. But when you build all those things, it ends up being a constraint to the speed at which you can move because you grow at the slower of demand growth and your ability to satisfy with your supply chain, but it also necessarily becomes a moat because all those hard things that you did, someone else has to do as well. And I think that the combination of kind of the customer experience we've been able to build with that model and kind of the speed that we've been able to build it and the difficulty of others replicating it is what has led to our kind of continued growth over this period from 2013 to today that we hope to continue going forward.

Got it. That's lessons learned. These are super helpful as we think about building a growing business. Maybe taking a step back and thinking about the state of the online autos market, which I'm curious, we're a year into the pandemic, right? The vaccine is here, so there's light at the end of the tunnel, which is thankful, right? We're all thankful for that. But where do you think car buying online is today -- online car buying is today? Cantor and others think about and talked about a 1% penetration. Others talked about 10%. Carvana's market share, I think you just mentioned 0.7% to under 1%. Just thoughts on where do you think car buying is today online and I guess, most importantly, where it's going. E-commerce, in general, is about 20%, 25-plus percent penetrated depending on how you manage that. So do you think online car volume could be higher than that 25-plus percent? Do you think it's lower? Like how do we think -- how do you think about the market where it is today and where it can go? Could it reach e-commerce levels?

Sure. So I mean, I obviously am incredibly biased. So let me caveat my answer with that. But I mean, I think it's -- I think it's potentially very exciting. So why would customers choose to transact online as opposed to in-store? I think that when you're thinking about our products versus other products, in general, customers are looking for a great selection. They're looking for great price and a great experience, like that's true across products. And then there's this question of like, okay, all those underlying customer preferences that you're trying to build a solution for, like how much better can you do in automotive versus brick-and-mortar versus other industries that have moved online? And I think that debatably, you can do like very, very well relative to even other verticals. And kind of the simplest way to explain why I think you can do really, really well relative to other verticals is that like consumers weren't actively complaining about the way they bought books, right? It wasn't a point of anxiety and frustration for consumers. The product lends itself to going online and you've got all kinds of convenience out of it, but it wasn't like a negative experience before. I think automotive retail, for like all the reasons discussed a moment ago, I think it's starting from a place where there's like more anxiety wrapped around it. And I think that building a real kind of national supply chain that kind of connects inventory together and gives customers greater selection with faster turn times so price can be low, like I think there's some deep fundamentals that are powerful in automotive retail. So I don't want to quantify it, but I think that like there's potentially a lot. And what I would say, too, is I've got like this simple mental model that I use. It's kind of 2 part to think about how big an industry online can get relative to the sum of the industry and then how big the biggest player can be kind of in any given part of the industry. And I think that like, in general, the amount of volume that in theory should move, in this case, online, but move in any direction to a new business model, is a function of like how much more like economies of scale and positive feedback does the new model have than the old model? Because if there's more economies of scale and positive feedback, all that does, that generates increasing customer benefits relative to the old way of doing things. And so that should cause more and more transactions ultimately to move online. And so I think that like this is a business where there have never been economies of scale. There's been, in fact, I think, diseconomies of scale, which is part of what's led to all the fragmentation to automotive retail. And I think all of a sudden when you move online, you have real economies of scale in selection, size, in delivery time, in brand, in all kinds of things, and the ability to buy cars from customers because you can now ship them all over the place and you have a retail type where you can sell everything so you can get kind of the maximum value. So I think there's real economies of scale that are very interesting, and I think that, that bodes well for the business model of selling cars online. And then I think as it relates to who should be the biggest player and how much of the biggest player win by, I think that a reasonable mental model for that is how complicated is the sum of the business to build. And I think the harder and more complicated and more investment it takes to build the business model, I think in general, you tend to see bigger, more outsized market shares from the biggest players. And I think that -- I'm sure that I underestimate the complexity of many other businesses that I'm not as close to, but I think at a high level, stepping all the way back, I think that like automotive retail demands its own supply chain. Like that's hard, right? You have to have your own logistics because you can't ship using third-party shippers because there aren't third-party shippers that ship cars in a way that is customer-facing. The transaction itself is way more complicated than merchandising and taking a credit card payment because you've got financing and a trade-in and you've got registration and regulatory compliance and all kinds of different things. And so I just think the business is more complicated than many other e-commerce kind of verticals. And I think that, that creates market share potential for the biggest player.

Super insightful and helpful. I mean, I think this is -- clearly, we're in the early days of where this could go. And let's get into the business a little bit more because I think everything we just talked about stems from what Carvana is doing, I think, with investments here going forward. One of the things that sort of stood out to us after earnings was a comment that demand outstrips supply. Yet awareness is still -- I think it was unaided, still 10% or so. And so I wanted to dig in a little bit more, maybe start with the IRC and logistics perspective, and then we'll work backwards to the consumer side. So one of the questions we just get often is understanding the impact COVID had on supply because I think there is always a confusion. Is it because there's not enough used cars out there? Or is it something else? And so can you talk to us a little bit more about what limited production in 4Q, really the back half of 2020, relative to demand? Was it acquiring vehicles? Or was it a reconditioning process where everything need to be retooled to be 6 feet apart? Or if you can provide insight on that, that would be helpful.

Cool. So first of all, thank you for the opportunity to clarify that because I do think our language choices of saying kind of like supply side constraints oftentimes lead people to believe that we mean we can't get access to cars, and that's not what it is. It's basically having the entire operational chain, from buying a car to getting a car to a customer, be -- like have the capacity to match demand. And so I think that, that is where our constraints lies in that kind of operational chain in sum total. So when we think about going through COVID, what I would say, like the simplest way to think about our primary constraint that we're facing today, which is the reconditioning center, which is where you take a car that you've already bought and then you put around $1,000 of parts and labor into it, to fix whatever is broken in the car, to bang out dents, to replace the windshield, to fix rips in the seats, to kind of go through and clean the vehicle up and photograph and prepare it for sale. All of those different things that you do, that's where our constraints are. And so those facilities, like the Modal facility, is a 4-line facility that runs 2 shifts. So there's kind of like 8 shift lines of people that you're hiring and ramping up. When you go through COVID, what happens is you just start to see more call outs. like generally -- actually, to set the stage a little bit, people are more than 6 feet apart, and these facilities are oftentimes largely open air. So they lend themselves pretty well to a COVID world. It's not like a place where you have to make massive changes to your underlying processes. So that's not the problem. What happens is people start to kind of call out because they get COVID or they're afraid they have COVID. And then we have to contact trace and send other people home and people have to wait for a negative test. And when you've got kind of an assembly line structure, it's very efficient when it's functioning properly. And it's built with a certain amount of robustness. So if a certain number of people call out, it still functions pretty much the same. But if you have enough people kind of checking out of that supply chain, that becomes problematic. And it can actually be more problematic, like if you lose 25% people, you might lose 75% of production in that chain, right? It's not necessarily just kind of linear. So that can be difficult. And so I think given the conversations that we've had after the Q2 -- I guess, on the Q2, Q3 and Q4 call, it can give us impression that we've been massively constrained that entire time, and we haven't been able to work out of it. And I think that is largely true from a distance, but I think the kind of -- if you zoom in a little bit, what's happened is the first wave of COVID really hurt us. And then that kind of wave started to subside, and we started building inventory again. And then we got hit with the second wave of COVID. And then that kind of really hurt us. And so inventory shrunk up because demand outpaced supply. And then kind of we went down the backside of that, and we started building inventory again. In the shareholder letter, we provide that data of total available units. Immediately available units is the terminology we use that started to grow kind of at the end of the third quarter and early fourth quarter. And then we got hit by the third wave of COVID. And we started to shrink inventory again. And then we provided this statistic on the call the other day that from December to the time of the call, we've basically grown production levels by about 40%. That's largely because now we're kind of coming down the curve of the third wave of COVID. And so you're getting a lot more efficiency out of those existing inspection centers. And then we're simultaneously ramping up the 4 inspection centers that we opened in 2020. And so that's led to this kind of unsustainably fast increase in production recently. So my hope is that as long as we don't face a bunch more waves of COVID, hopefully, we will recover relatively quickly and start to get production levels back up, which we view as very exciting because we do think there's a lot of demand there that we're not currently accessing.

So if we could drill down that 40% increase on production. You mentioned new IRCs coming on, the third wave of COVID, and we'll get to -- maybe after this, we'll go to questions from the audience because it's growing, which is great. Is this -- is the ramp here -- talk about the -- I guess what I'm trying to get at is, have you retooled the processes now that when you're entering, you do the temperature check and such that you're fighting off the COVID, if there's another wave, you sort of know how to handle it? So what I'm trying to get at is, like what changed the processes to get that 40%? Because that seems like it should continue.

So I do think we've gotten a lot -- I mean a lot, lot better at managing through COVID. That said, the vast, vast, vast majority of transmission that we saw inside of Carvana did not occur on the grounds of Carvana. Like it's people living their lives outside of work that get COVID and then come in is where virtually all of it came from. So I think in that way, it's not really -- like what we have to build our process around is being robust to unexpected call outs, right? That's what we have to try to build our process around. And there's just so many ways to do that, but there's -- there's not a lot of super efficient ways to do that. So I think that we've gotten a ton better at it. We've got all of our processes for contact tracing and for taking temperatures at the locations and doing all those things. But I think like as long as these vaccinations start to work and we don't see really large waves of COVID, I think we should ideally be in way better space going forward. But you never know exactly what the future holds.

Sure. Maybe another question on IRCs and talking to the comment earlier, whether it's online autos gets to e-commerce penetration or doesn't, or however many -- however it goes and the complexity of it. What do you think is the gating factor of that from a Carvana perspective? Is that IRC build-out? Or is that the network that we're building in terms of distribution? I ask because, obviously, we know we'll see, what, 10 new IRCs between today -- this year and next year. That's 1.25 million capacity. The goal here for Analyst Day, not [ legal ], but [ eagle ], was 2 million at some point, right, and going forward. So I guess my point is, as you get to that e-commerce penetration levels, is IR -- are IRCs sort of like the -- I don't use -- want to use the word pinch point because it's not a pinch point. But as you think about the buildout there, IRC is sort of like the gating factor to share gains?

So I think that, that's certainly the gating factor today for us. I think the gating factors are the sum of the chain, which is complex. Like to run this business properly, there are a lot of things you have to physically do. And physically doing things are hard, and scaling them comes with risk that you have to be careful about, right, that you don't -- like a major risk is you try to scale so quickly that you mess up the customer experience and then you kind of mess up your brand. So you got to be really cognizant of that because customer experience is like the original and single most important goal in the whole thing. So I think making sure that you scale the chain in a way that continues to deliver great customer experience is important. And then I think something else that's really important is like it's expensive to invest in all of the ways that you need to, to grow quickly. And in ways that are obvious, things like building inspection centers, but also ways that are not obvious. You always have low tenured people. If you're doubling every year, your average tenure is very small, right? So you have very low tenured people, people less efficient, and you have a huge percentage of your more tenured people that are doing interviewing and training and like all those sorts of things. You're probably investing a lot of your efforts in things that are supporting growth instead of things that are supporting efficiency. So that's expensive. But I think like the way that you pay for that is in gains in GPU, right? Like I think one of the reasons why GPU has been such a big focus for us from the beginning is if you can kind of grow your GPU by getting really efficient at monetizing everything, now your incremental transactions, even if it's expensive during the period of transition from scale A to scale B, you can afford it because you've got these big GPUs flowing through that can help you fund it. And so to me, I just think it's like -- I think it's building that chain. It's like building this system that works and then running as fast as you can without falling on your face the whole time and making sure customer experiences are great and that economics work out and that you're well capitalized on all these different things. I think it's like the sum of many problems. Like I don't think that there's 1 or 2 things you point to you say, that's the thing. Like that's the thing that's hard and that's the thing that creates the moat. I think it's 1,000 little things, all added up, that are hard in combination and create the moat in combination.

Got it. Okay. So let's go to customer questions or investor questions here in a second, and we'll get back to the other questions that I've got here. Let's see. And by the way, those listening, if you have more, feel free to put them in. We're getting a good amount in here already. But one question is talking about capacity. It goes, is it fair to use the capacity metrics that you released with 4Q earnings as a framework on what retail units could be in '21 and '22 if the demand environment continues to remain strong? Is that a way to think about it?

I do think it's good. Now let's like develop the framework around it. So first, let's start with what that question ended with, which is you grow with a [ slower ] of demand and supply. So let's like assume that you have sufficient demand. We certainly have a lot of demand today. And then you kind of move to, okay, now how quickly can you grow the supply chain? Now like the first thing you need as it relates to reconditioned vehicles is the physical facility capacity. The next thing that you need is if we go back to that modal inspection center that has these 8 shift lines, you have to hire all those lines, train all those lines. That takes a certain amount of time, and we historically at least have gone through and done 1 line at a time. So we're doing those like in series instead of in parallel. So you can kind of look at our historical -- like generally speaking, going back to your earlier comment, Ron, that like our inventory today is not materially different than our inventory from 4 years ago, which is materially different as long as you kind of zoom back enough from our inventory when we first started. Like we, generally speaking, have kind of sold all the cars we've produced is like 1 way to characterize our entire life. So when that's true, you can kind of like just look at sales and you can figure out like, okay, what is -- what is our annualized production capacity is effectively very similar to whatever our sales was in any given quarter times 4. And you can kind of compare that to the facility capacity we've had, and then you can look at how long is it taking Carvana in the past to get from a certain level of facility capacity to that many actual sales. And that gives you a sense of what our lag times in the past have been. And those lag times are kind of made up of the time to hire and train all those lines. And then it's also made up of -- by like the degree to which we were choosing to sprint into that capacity. And so I think like that's like the way that we unlock that capacity is like you got to build the facility first and then you got to hire and train, and that doesn't happen overnight. It does take some time.

Okay. Another question comes in here. How vertically integrated are your logistics operations? And how differentiated is this from peers pursuing online car retailing models?

I think it's pretty differentiated today. And it's -- I mean, it's vertically integrated to the point of like we have now one of the largest fleets of trucks that moves cars in the country. A lot of the bigger fleets that move cars in the country are basically moving them for the OEMs, from production facilities to dealerships. Our fleet size now rivals many of the largest there and is growing much more quickly than any that at least I'm aware of. And then that kind of is supported by an entire like first-party logistics, like software system that is figuring out which cars go on which trucks, at what time, what is the current capacity of trucks, what's the right way to connect a car location to a customer location to get that done in the quickest and most efficient way. And then that includes like the last-mile logistics as well, which are the single-car haulers with uniformed advocates that deliver the car to the customer. That's a, in my opinion, a big, big undertaking. And it's done in a way that is complex compared to anything that's existed before us. Like I don't think you could just buy, for example, one of these big shipping companies that ships cars from OEMs to dealerships and have like the same kind of solution because they're solving a way simpler problem. It's like pull up to the factory, load 10 cars on the back of this thing, drive it to a dealership. It's not, okay, like this car is going through these 4 legs. This car is going through these 7 legs. This car is only going through 1 leg. How do you load these cars on the truck? And where do you do that? And how do you set these up? And when do you hand off keys and keep the trailer fully loaded? And when do you kind of take the cars off and reload them onto different trailers to reoptimize? There's like a lot of problems you have to solve there. It didn't look like any logistics optimization problem that has been solved like many, many times by traditional shippers. But I just don't think that, that same degree of optimization has been applied to automotive shipping, right? Like people do all kind of smart things about like which boxes do you put in which location of a UPS truck. But that's like a problem that's been -- being optimized for 20 years. And I think that we are kind of the first ones that are optimizing this problem with respect to vehicles on like vehicle haulers.

Got it. Got it. Interesting. Let's switch focus a little bit more and talk about sort of the demand side of the platform and the equation, so consumers to a certain extent. I want to start a market rollout. And I think we -- I think the comment was 80% of the population by the end of the year was a goal. And we're always getting the question, when you look at the map, Northwest is pretty barren. Does that 80% include the Northwest? Or maybe I don't -- that's not the question. The question really is, talk to us how do you choose to expand markets in any region specifically as you -- on your way to getting to like that 95% population coverage longer term? So any percent of like shorter term and then how do you choose.

Yes. So I think the goal is to get to 95% plus, which is largely like you can think of as kind of like coverage of contiguous U.S. is kind of like the simplest way to think about that. And then I think like the next question is, how much do you expand in any given year? And from like a long-term perspective and from a brand building perspective, we want to expand as much as we possibly can. From the perspective of facing and dealing with the constraints that like we're facing right now, you debatably shouldn't expand kind of at all. But then you have to kind of like think about it. Like the way that you try to thread that needle is you think about the way that demand kind of ramps up in markets, and then you kind of say like, okay, maybe opening markets right now is putting undue incremental demand on the system, which isn't perfect, but it's like low enough levels over the next 12 to 18 months because markets take a while to ramp up, to where it's worth it to make sure that we're still marching out in that period that's 12, 18 months from now, where we think that we're going to be okay from a supply side perspective. So I think that's kind of the next question, which is like how much market do you want to open up? Because I do think we're largely in a place now where if we wanted to just like open it all up, we probably could. It's not operationally difficult enough at this point where we couldn't open up all of it very quickly. So the second question is just amount that we want to unlock. And then the third question is where do you unlock the amount that you're going to unlock. And in general, we're going to pick the easiest like amount that we want to unlock that we can. And so going up to like the Pacific Northwest, like that's kind of the last remaining big population cluster that we haven't yet addressed, but it's also pretty far removed from the nearest population cluster. And so it's kind of always been lower priority as a result of that. And then we haven't given specific expectations of what we're going to open. But obviously, we're marching to the place where you're starting to get close to opening everything, so it's not far away.

Got it. That's helpful. And to that end, how do you build the awareness? So now that you were talking about 95%, not that far away to getting nationwide, call it, which you're sort of already there. But you mentioned 10% on any brand awareness on the call. We mentioned that earlier today as well. Just marketing is more national than regional. Can you talk to just how do you build awareness overall? And is it fair to think that Atlanta's 2 and -- 2 1/3, call it, share is potential for national -- potential for markets as we get to, call it, national awareness over time? Like is there a way to think about Atlanta to total? Yes. I'll stop that in my opinion.

So I think -- so first of all, let me start with this, like we did find a stat that, nationally, we believe that we now have on the order of 10% unaided awareness. Roughly speaking, I think that we provided a stat last year that in Atlanta, we had on the order of 10% unaided awareness. So we haven't provided an updated stat for Atlanta, but like basically, that at least gives you a sense of where Atlanta was a year ago and then how the rest of the country is generally catching up. Now I think that, that's actually a very useful metric for us to follow. But I think of it much more directionally than like quantitatively. I think when you see this 10% stat, there's like this thing in your brain that wants to be like, okay, so then if they get to 100%, they'd be 10x bigger. And I can just like do that math. But I think that like that's like a massive oversimplification of what reality is because I think like when you think about what a customer needs to transact, it's like what are the things -- what are the hoops that they have to jump through to decide to transact with you? The first thing is they have to like be aware that you exist at all. And so let's call that, for this purpose, unaided awareness, but you could argue that when someone's actually in the purchase funnel, they start to like gain awareness that is above what is traditionally called unaided because they start to be aided as they're searching. So maybe that number is like a little bit higher, right? I think it's somewhere between unaided and aided would probably be like my starting point. But then they need to understand what you do. So like even when we survey people that are aware of Carvana, even unaided aware, if you ask them like, does Carvana offer return policy? You see all kinds of answers across the board. How big is Carvana's inventory? Answers across the board. How do you get a car from Carvana? Answers across the board. So you still have this like understanding hurdle you have to cross. People have to understand what you do to be open-mind transact with you. And then last one is they have to trust that you're real, right? They have to trust that the 7-day return policy is real. Like knowing it's a 7-day return policy, like that still doesn't necessarily give you comfort unless you really believe in like the institutional credibility of the company. And so I think you have to march through all of that. And to me, like more is back, right? Like pick your measure. I don't care if it's aided, unaided, understanding, whatever, just more is better. And the more that people know about you and think about you and the more top of mind you are and the more trusted and understood you are, the more you're going to see transactions go up. And so in that game of like more is better, and we are using this one particular measure of unaided awareness, I think that it's probably very early in the game of building our brand, Carvana. And so I think there's a ton of upside there. And so then when we think about that, then you -- you face all kinds of questions about, okay, how do you build your brand and how do you do that efficiently? And I think that, by far and away, the most efficient thing you can do is just deliver great customer experiences over and over again and have people talk about it. I think the next most efficient thing that you can do is invest your dollars into things that kind of make those experiences noteworthy. These customers want to talk about them. And then from there, I think you face like the more traditional marketing question of like, okay, where do you put your advertising dollars? Do you put them in direct kind of like converting channels, things that are very deep funnel? Or do you put them in very high funnel things that are generating that awareness and understanding at a higher level? And so in moments like this, where we're constrained pretty heavily on the supply chain side of the business, we're going to elect to put -- we're still going to invest our dollars, but we're just going to be investing mostly in the brand side of the equation, not in the direct conversion side of the equation. And so that's generally how we're running that today.

Got it. And someone is asking this in the queue, but in a certain way, are demographic characteristics of these new buyers any different than when Carvana first launched from historical customers? Or...

Surprisingly, so here's what I would say is, I think what you hope is -- let me start with this. Like when you're building a business, I think part of what you're trying to do is you're trying to build a business that appeals to some set of underlying customer preferences. And then I think that like the more like esoteric those preferences are, the more you're going after a niche. And the preferences that we are trying to serve are like very, very down the middle, not esoteric at all. It's better selection, better price, better experience, right? That's...

Maybe -- can I maybe put a twist on that or -- have we crossed the chasm, so to speak? So it might be the same type of customer, but now people are more willing to buy online because it's more normal.

Yes. So I don't know that I know the answer to that, to be honest. But -- well, let me finish the previous thought just to make sure I make it to...

Yes, yes, yes.

No, no. You're good. So -- but I think we're trying to build a business that supports kind of like very down the middle customer preferences. So I don't think we're shooting for like a subset, and we've never seen a subset of customers. We've always seen a distribution that looks in age and demographics and credit in virtually everything, a lot like the average buyer. And then because it was so much like the average buyer before, what I think you're hoping if you build a business that is "crossing the chasm or getting more mainstream," I almost think that you're hoping that your distribution of customer attributes, and I need to mention that you're measuring, is approaching the distribution of the population, right? Like that's what you're...

100%. Yes.

But I think we started so close to that distribution that it's hard to measure migrations. But then I think you asked the right question, which is like, okay, but maybe psychologically, are you crossing the chasm and moving in these different directions? And I don't think I know the answer to that. I think that a good way to measure it is with unaided awareness, and then a good way to measure it is just demand on the site. And I do think demand on the site today is higher than it's ever been by a decent amount. So I think we're happy about that, but then I don't know how to measure like this, like moving from early adopters to kind of early mainstream, whatever the different groups of customers are in that mental model of crossing the chasm.

Got it. That's helpful. Let's maybe -- so much to talk about. We have maybe 10 minutes left. So let's see if we can get through...

You want me to give shorter answers. I really would. I don't know how. I'll try, I'll try.

No, this is wonderful. I'm just saying there's so much to talk about. So let's talk about sourcing. So I understand the sourcing. We talked about IRCs. Done. Let's -- I think that's understood. I want to understand just retail sourcing from consumers, right? I think 90% of retail units sold at IPO was in auctions back in the day, and now 65% of retail units sold were from consumers. So I wanted to ask about, sort of piggybacking on the awareness comment, how are you building awareness that Carvana is buying your car? And then how do you make sure that you're not buying a jalopy, so to speak, right? And this gets into a bigger question of like how do we think about wholesale units overall and then into CarvanaACCESS. So I'll be specific on just like building awareness for car buying because there's a lot to that.

Yes. So I do think that, first and foremost, you just want to build the brand Carvana because then you have an asset to build on top of. And then I think the next thing that you want to do is you want to make sure that you add a value prop to that brand, which is we will buy your car as well. And so we have spent -- over the last couple of years, we've given the stat at different times that on the order of like 20% of our advertising, on the order of 1/5 was maybe the language we used, was focused on buying cars from customers. So we have tried to get that message out there. And so I think that's how we're trying to just kind of like add that value prop to the brand itself. And I think interestingly, a lot of customers start from a place of not thinking they can sell you a car and not thinking they can trade in a car because you're an online platform. And so I think that's been like an education that was more important than we appreciated early on.

That's helpful. Got it. And from a condition perspective? And what it means to like wholesale units, maybe that's the way to ask it.

Sure. So let me start with this. I think it's incredibly important when you're buying something as complex as a car with as much potential value variability as there is in a car to make sure you don't get negatively selected and get beat up because then you can just -- you can lose really badly in that kind of a setup. But what I would want to start from is, I think a lot of times when people are evaluating a new business like ours where you're buying a car sight unseen, they can start from this place of implicitly assuming that the old way of valuing cars was a lot more effective than maybe we think it was. And so then all of a sudden, they start pivoting off of, okay, well, here's where you're worse here, here's where your worse here. And so like how are you going to come up -- how are you going to handle all those places where you're worse? When we first kicked this thing off, like one of the big questions that we always got was, you can't take trade-ins from customers. They're going to lie you. They're going tell you all kinds of things about their car. You're going to get crushed. You can lose several thousands dollars per car. This isn't going to work. And that was like over and over said to us. So right out of the blocks, like when we were still part of DriveTime, we did this incredibly simple test where we basically said, okay, let's like -- this is a thing that we can test and we can figure out if this is true or not. Like everyone is telling us this is true, but it's not hard to just prove one way or the other. So let's build like a very simple model with the data we've got that tries to predict what we think each car will transact for that's going through auction. Then let's take a bunch of the buyers that we have, the physical buyers that value cars all day long and let's have them go value a bunch of cars. And then let's tell them the rules of this game are you value the car, you write down your number, you step away, you don't bid in the market. And then our algorithm is also going to value all those cars and let's just see who's closer to right on average. And in our very first version of the algorithm, our average standard error was like $1,100, which means if we thought the car was worth $20,000, like 1 standard deviation in either direction would be like the car actually transacted for, call it, $19,000 or $21,000, right, which is a big miss. If you're -- if like 1 standard deviation is $1,000, it was $1,100. That's tough. And like you obviously are going to have somewhere 2 to 3 standard deviations off, and you're going to also -- you're going to win more where you're off by more. Okay. So that doesn't look good. But then we went and we said, okay, now what's the data for the physical in-person buyers? And their average error was $1,000. So like they were also missing by $1,000. And so what it just comes down to is like when you look at anyone who's buying cars in any process at all and then selling them in a new market, the variation in price from kind of when you bought it to when you sold it is already super broad. And so you're trying to get to like a [ mean ] that's in decent shape. And if you can collapse that distribution as tight as possible to your advantage, like that's a big deal. But we started from this place where the distribution was already so wide. And you know that's true because wholesalers are a thing that exist in the market. There's an entire market of people that go to auction, buy cars in person, go sell them at another auction a couple of days later and make big money on it just by running the car over and over again and basically exploiting the variation in those prices across many different runs. And so I would start from this place of I don't think people are that good at valuing cars in person either. And so like you're starting in a simpler game than you thought you were starting in. And now you have the advantage of an algorithm with tons of data and no bias, and that's helpful.

And that's only grown and gotten better and better as time goes on. That's super helpful. Let's see. I'd be remiss tons of questions here. We're not going to get to all of them. But from a GPU perspective, I'm getting a lot of questions on GPU, and we haven't gone there yet. And so we're coming up on time in 6 minutes. The questions, in general, are asking about GPU growth to a certain extent and on retail and wholesale. And specifically on retail, what can -- where are you seeing parts to drive that retail GPU up higher going forward?

I think the biggest potential driver of retail GPU up over time is continued progress in monetizing cars that we buy from customers. So I think the right way to think about the retail contribution of cars bought from customers is take roughly our average wholesale GPU per wholesale unit, which was about $350 in Q4, and for the full year, I think, was something like $600, though I'm not sure that's right. So please like check that, but was something like that. And that was probably too elevated because Q3 was abnormally high. But if you look at that over multiple years, it's been kind of like moving up consistently. So you want to take the percentage of cars you bought from customers that were sold to customers. So in this case, in Q4, 65% times that number. And that's kind of like the contribution that you're getting versus if you bought all your cars at auction. And so right now, we're doing so well on the percentage of cars that we're selling to customers that were bought from customers that I think there's potentially some upside there. There's also potentially some downside. We're already 13 points above the long-term model that we set for ourselves 2 years ago. But we've got a lot of room on kind of approaching our long-term goal of making $1,000 per wholesale unit. And so I think that, that's probably the biggest fundamental contributor. And then from there versus right now, I think there's also -- there are some costs in COGS that are flowing through because of how quickly we're trying to ramp up our inspection centers. So I think that, that's like a transitory benefit, but kind of versus today, that's a tailwind. And then there's just a question of where do you set price in the long term, which, for now, we're assuming is going to be in the same place.

Okay. And really quickly, how excited are the significance of the first securitization SEC-backed to prime?

So I mean, I think that -- I think, in general, we're really proud of our finance platform. And I think we've got a really, really high-quality finance platform through every lens that matters, starting with the customer experience, the simplicity of the experience they go through, and then ending with our ability to monetize it with structuring and credit scoring and credit pricing kind of sitting in between. And so I think that moving to that kind of SEC-registered deal is another material step in monetization.

Yes. Okay. On the earnings call, Ernie, I think you mentioned around -- we've been seeing a lot of third-party units on your site for some time. On the earnings call, I think we -- you got a question and you sort of said, well, we might be entering the next phase. I don't want to comment on that too much. But any insights on like how you view third-party units or third party or becoming more of a marketplace might be a way to ask that question?

Yes. So I'm going to apologize in advance for like talking circles around this issue. But I would say that you can kind of think about that as, one, everything that we can do, every product, every feature, every adjustment to process, every like adjacent capability we add to Carvana, we always view through this lens of like, what does it do to the customer experience, first? What does it do to unit economic, second? What does it do to scalability, third? And you can think of some of these things that we'll test around the edges as kind of being ways that we would limit, we would shorten our value chain and, therefore, drive more scalability, but also, therefore, give up some unit economics and, therefore, take some customer experience risk. And I think that in any of these tests that we run, we'll try to just get a sense of what's possible there. And we'll design our test to try to isolate those 3 things and understand what moves. And then based on the answers to those questions, we'll kind of decide where we go next. And then we're purposefully just trying to stay away from talking too much about any specific test just because there really are -- there are 60, 75 different teams inside of Carvana working on all kinds of different things all the time. And several of those things, I think, are the sorts of things that could get a lot of intrigue and attention just because there are analogies of other companies that have done similar things that are interesting enough to where they're like big ideas. But many big ideas that we work on won't work, right? And hopefully, some do. And I think that the opportunity we have in front of us with just our core business is so big that we want investors to know that we're going to do our best to think of everything that's interesting around the edges that we could and should be thinking about. Then we're going to try to prioritize those and test them to understand them as best we possibly can. But we also just have this huge, huge opportunity we need to stay focused on, and we're very confident about our ability to go attack that and execute against that opportunity and do really, really well. And so we want to try to keep focus on the thing that we really understand deeply where it's just an execution question as opposed to the things where there's execution questions and product questions.

And this is the last question, sort of a derivation off of that, because I love the comment on the framework, customer experience, unit economics, scalability, and the difficulty in growing a lot of these businesses that you -- or building a lot of the business that you're doing. That's raising awareness, that's reconditioning cars, that's the last-mile delivery network, as you said, that hasn't necessarily scaled, buying direct from consumer and auction market with access. Can you talk about -- like we're in the building phase of this. Is there a point where you might see excess capacity in any of these phases such that, and I'll say it outright, like they could be their own businesses outright?

So I don't think that should be the expectation in the near term, especially given the opportunity that sits in front of us by connecting that value chain into a single thing. And then the value that I think is generated by the kind of compound differentiation that occurs with more steps in the value chain. So I think the longer a value chain is, the more differentiated it can potentially be. And so we have a ton of opportunity in our longest value chain. And I think the value chains that we would think about the most from there would likely be pretty long versions of our value chain because that's where you have the most moat and the most differentiation.

Got it. This is great. Well, Ernie, I think we're out of time. To all those that asked questions -- all those have asked questions, I apologize we didn't get to all of them. But Ernie, this has been wonderful. Thank you very much, and looking forward to the next time. So...

Cool. Well, thank you very much. Appreciate it as well.

Congrats on everything.

Thanks, everyone, for listening. Sorry, my answers were too long.

Absolutely. No, they were wonderful.

No. They're too long according to your face. Everyone saw it. Thanks. See you later.

All right. See you, Ernie. Goodbye.