Driving Innovation: A Fireside Chat with Ernie Garcia, CEO of Carvana & Executive of the Year

Thank you.

Let me start with what the vending machine maybe is because it's very unclear why that makes business sense at least superficially. But I think one of the places where... let me start with this: the entire car market is people switching cars to each other. That's like all it really is. All the cars exist in the world. People are driving them. They got to switch it to someone else. So I think when you're trying to build something different, you got to look at like what is the experience that exists between people that are switching those cars and what are the expenses that exist there. And in the traditional way of doing it, the biggest expenses are basically paying commissions for people that are delivering the sales experience and managing financing and everything else. And then the second biggest is either real estate or marketing. And in general, I think marketing is a very important function in any business. But the best dollars you can spend in marketing are generally improving the customer experience. If you can spend your dollars improving the customer experience, it makes your customers more likely to come back. It makes them more likely to talk about the experience. Whereas, if you just spend your money getting their attention, you get customers to come, but there's no feedback inside of the system. And so, it's kind of like a linear relationship instead of like an exponential payoff. And so I think when we were thinking about how we wanted to build the business, we wanted to create a really fun experience for getting the car. Getting a car is usually something you're going to tell your friends about. You're gonna tell your family about it. It's like exciting. And then buying a car, I think, is sometimes like a stressful experience where you're afraid you're going to get it wrong. And so we wanted to build fun into it. You know, we started delivering cars to customer stores. That's how we started. We realized it would be less expensive if we did like a hub and spoke logistics network instead of a direct, you know, from like a large center of cars to customer stores. We wanted to stop in a city center somewhere. So, we bought a little spot in Atlanta, which is our first market. And we spent like 20,000 bucks knocking out a wall and putting in a glass garage door. And we were trying to figure out like what was that experience going to be named? And the person who runs marketing for us said, 'We'll call it a vending machine.' And I was like, 'One, that doesn't make sense.' And two, it's like a massive exaggeration of what it is, but okay, let's try. And so we called it the vending machine and we had never gotten any coverage like as a... especially back when we started 2012, 2013, there weren't a lot of startup companies in Phoenix. And I don't mean to come at like the media landscape, but I think a lot of times people don't want to write positively about you unless there's some other positive data point that exists before that moment. So we couldn't get any press to save our lives. When you're a startup company, you need press because it's like the only free marketing you can get and you don't have any money. But by just calling it the vending machine, we were on the homepage of Yahoo for like three straight days. And we went from, you know, like when I looked at Google Analytics, see how many people on our website, it averaged like 120, 200. And all of a sudden it was like 55, 60. And I remember my wife who's sitting there. I remember, you know, sitting there watching whatever show we were watching at the time, probably Game of Thrones or something. And I had set up my phone on Google Analytics so I could see it and pretend like I was watching the show, but really I was just watching the number tick from the 50 to 60 because I was like so pumped that that many people had chosen to come to our website. And so I was blown away by the power of just the language of vending machine. And so we started to research online like how can we build a real vending machine? And we found a cool vendor in Germany that had a chance to build something that we thought would be like a real car vending machine and ended up trying to figure out that trip. It basically, I would say like we were 1% away from not taking the trip at all because we had other things that were coming up. We were trying to raise money. But found like a little sliver of a window where we could do like an overnight flight, basically be on the ground for 12 hours and come back. And so we went out there and did it. We like, you know, figured out the whole deal with the manufacturer and like structured it and then we built our first vending machine and that was in Nashville. And customers absolutely loved it and it led to a ton more free press and coverage and customers also would pull out their phones, record it and share social media and say it was awesome. And so I think that's how it came to pass. And it was like one of these awesome discoveries that I would say was not hyper-rational in its origin. I think it was like so often in business we think that like everything has to fit in a spreadsheet, but people like awesome. People like cool. Like that's a real thing. And so I think when we built something that we at least thought was cool and awesome, people responded. And so I think that was like a huge innovation that's really cool. And then what was the rest of the question? What's next? Well, we're a public company, so we get to play with Bill Koi and not answer those questions and just say tons of cool stuff is coming. But, tons of cool stuff is coming.

So I think in a business the ability for AI to make an impact is constrained by the scale of your business because you need to be generating enough data to be useful. It's constrained by the amount of data about your product that exists in the world already. It's constrained by how vertically integrated your business is. I think like there's always like an interesting question about vertical integration versus, you know, horizontal businesses and like I think the argument for vertical integration is complete control, the argument for horizontal businesses is that you can focus and focus is a tremendously important thing. I think as AI becomes more important, I think vertical integration becomes a better answer because one of the frictions of horizontal businesses like layers of, in our case, think about like a finance company and an automotive retailer like they have to coordinate to give you a single experience. If their data is not structured really well and able to be passed into the same system, it's unlikely the AI can even provide you with, you know, great answers about like, you know, what car could I have? Like if you want to ask the question, what's a good car for me? You know, I'm graduating college and I've got a dog and I want a payment that's less than $500 a month and I need it by Wednesday. Like, to be able to answer that question, you have to have like a logistics system that is well structured and integrated into the agent. You need to have your finance processes and policies that are integrated in that agent. You need to have knowledge of your inventory. You need to have integrated with your search API. So, there's a lot of stuff you have to do. As you do that though, you can give these incredibly high quality answers instantaneously because the agent can do all those calculations very, very quickly. So I think it's really cool because it not only reduces kind of cost in the system and changes I think like the entire landscape of like the structure of businesses that are most likely to be successful. It also gives customers an answer that would have taken a very well-informed person 20 minutes and it can do it in one second. And so I think it's amazing. Yeah, basically it's amazing and I think if you want to go play on our website, our agent is called Sebastian and you can start to ask these questions and depending on what like test branch you're in, you'll see different levels of sophistication in the responses, but generally it'll be pretty sophisticated. And then you'll see some cool stuff if you go to like the search page. Most of the cars will have three dots underneath them on the search page and you kind of swipe them and that's a different form of AI that's basically taking our fixed images of all these cars, is converting into a 3D model and it allows us to do animations that are kind of like beautiful to make you like really understand what that car is like. There's other features you click into a car and you see the BDP. We've got something we now call WIS which is basically all LLM informed. So it's taking all the data about that car, the types of people that buy it and it's creating little two sentence words about like why you're going to love this car. So I think there's more and more stuff that's just showing up all the time and it's actually really fun. And then the other thing that's happening that's wild is the speed of development is also accelerating and that's happening a little bit slower but it's definitely happening. You can now, you know, I think like there are companies out there, there are certain companies that will say like we're doing, you know, 95% of our lines of code are generated by AI. That's I think you have to be pretty generous in the way that you calculate that to get to those sorts of numbers. But I do think it's very real that it can be 30%, 40%. And it's very true that it was 0% like 18 months ago. So I don't quite know where that goes, but it's definitely accelerating the pace of innovation.

You know what? I think that's the ultimate goal. I mean like, people always ask about the size. I mean, I can't imagine that any of you guys care that much about used cars, but I mean, I'm just gonna come at you with it. But like, people always ask about the size of the market, and so it's a 40 million unit per year market. It's a really, really big market. It's the used car market in the US is probably the single biggest single product retail market in the world. And the way it works is basically there's 270 million cars on the road and people swap them out every six or seven years. And then that math just reduces to there's 40 million transactions per year. But like if you think about your own experience, you want a car more than once every six years, right? Like you would rather have a car every year. But the reason you don't is basically because there's economic frictions. It's expensive to take a car from one person who has kind of made peace with the dent and the rip in the seat and, you know, the smell and get it ready for the next person who's like this is my new car. None of those things are okay. So the more that car is turned over, all that work has to be done and that creates an expense. There's expense to pay the people that provide the retail experience for you. So that's an expense. There's oftentimes a cash friction because you have a loan that probably has negative equity. If you want the new car, you got to pay off that loan. You got to get your next car. And you can't necessarily finance it all. So it's not only an economic friction, it's also a cash friction. I think as you make the experience simpler and cheaper, you enable people to do the thing they want to do, which is you buy a car every five years. And if people bought a car every five years there'd be 55 million transactions per year and the market would just instantaneously grow by 40%. Which is cool. So I think part of what we hope to do is build an experience that's simple enough to not only, you know, take market share but to grow the entire market because it is a market that is just recycling product. It's not a market that's fundamentally constrained by the product being manufactured.

Let me start with this. The first company that we tried to start was an idea that I know works because two other companies have done it somewhat successfully since. And basically the same team or at least a large part of the same team that has built Carvana was part of that team and we blew up in about a year. And I think that the reason I say like that's what happened and the idea worked is because I think people think that like what matters is the idea and I just don't think the idea is what matters very much. I think what matters is being ambitious or dumb. And I think it's a combination of both enough to try something hard and then I think just solving all the problems that come up along the way. And so I think that to me after we failed in the first business, you know, one of the lessons was like, okay, we didn't know anything about this business when we started, we didn't have like any experience around it. And so it took us weeks to figure out things that other people knew in minutes because they were around the industry. And so, you know, your speed of being able to make decisions and make impact is a function of how quickly you can solve problems. And I think that was something that slowed us down, made us worse in that business. I think there were a lot of other things, too. But I think that was one of the things. So, we wanted to go to something we knew. I had been, you know, around automotive my whole life. And so I felt like I understood it. But then when I would ask people like, why is buying a car kind of a bummer, right? Like why do people not like it? Basically got answers that were just like very unsatisfying, incomplete, and reduced to kind of like, well, that's the way it works. That's the way it's always been. And I think that that answer will always sit between you and anything new. People have a default assumption I think a lot of the times that like the things that exist in the world around us were very intelligently composed and they probably were but like the world changes, people's preferences change, technology changes and the people that built all the things that were here to begin with or when at least from your perspective, they make mistakes all the time like they do things highly imperfectly. So, I think you have to just be willing to like pick a direction that you feel strongly about and then just go like march and then you get punched in the face a thousand times. You got to keep getting up and you got to solve the problems one at a time. But I don't think that there is such a thing as a or there's very rare cases in the world where your idea can be so unique and beautiful that it's like easy and fun the whole time and it's like, you know, private planes and champagne. Like that's I think usually it's like a brutal grind where people call you dumb most of the time. And so I think you have to be willing to tackle that. And you also have to know that I'm not sure you want to know to be totally honest. But the truth is however long you think it's going to take it'll take five times to 10 times as long. Knowing that may not be good. So like I'm sorry if I burden you that knowledge because I think if you know it, you're less likely to try. But I think it's the truth.

I think it'd be a version of what I just said. I just think everything worth doing is very hard and everything that you do, you're going to need other people that care that come along with you for the journey. And the only way to get people that care, that are resilient enough to fight like all the battles you're gonna have to fight, is I think to try to build real relationships with them. And I think that's hard, but I think you do that by saying things that are true and not convenient. And so I think, yeah, it's like I don't mean to try to make it heroic, but I do think that trying to do something different is a battle and you need a lot of people to come along with you if you're going to do it.

So the most important advice of, you know, and I hate to call it that, suggest that I think it's right which I think it's right but it's not obviously right that I would give to all of you guys is one, don't be afraid of risk. I think like maturity and I put it in quotes because I think it's very unhelpful to progress is like the worst thing there is. Maturity means, you know, you compromise. You avoid unnecessary risks. You do things the way that all the smart people that came before you tell you to do it. I think that like you're young people and you should go try to either like if you're really risk tolerant and I think you should sit with that because most people aren't, go start something and go try to do it yourself. If you're reasonably risk tolerant and even if you're not that risk tolerant I would say go work at a company that is either small or flat. And what I mean by flat is just like there's not many layers of bureaucracy because I think you want to be close to real things getting done. And I think that if you're in a company that fails, it's the best thing that will ever happen to you. Like I think that the first sort of way people think about it is like I should go work at some big company where I've got a stable job, the best thing that will ever happen to you at your age is that you go work for a company that fails and you get to be close to it and see how it failed and why it failed. Like there's nothing better there and you'll go to your next job interview and if they're smart, they're going to love that you experienced that. So to me, like something that I always really struggled with when we were first starting, we'd be interviewing people and a lot of times the people that are most capable have a lot of options and they're sitting there telling you like, you know, once you like break in and you have a real conversation, they're like, 'Well, but isn't this risky?' And like it's hard because when someone is asking you a question where your motivations are very clear and they know what your incentives are, it's hard for them to believe, you know, the things that you're saying that are convenient for you to say. So I would try to like break through, but that would always be my message. Risk is like the greatest thing that could happen to you right now like I think Carvana worked in many ways because we failed in the first things we tried to do and so I would just say, you know, take risk, you know, be in companies that are flat, be close to the real work. The mature thing to do is to build processes and try to build like a big group underneath you with a bunch of direct reports that have direct reports that have direct reports. All that means is you don't see anything real like real life happens where the customer meets your product and like data matters. But it doesn't matter as much as just seeing what actually happens every single day. And so I think you want to be close to real stuff. And so what I try to do inside Carvana, and this is partially selfish, but it's also partially because I think it works. I like to pick at any point in time like three or four projects that I'm in the middle of. And ideally I'm sitting next to the engineer that does the actual work or I'm sitting next to the designer that does the actual work as opposed to like having many of those projects, you know, sort of like filter up and then you discuss them for two seconds at a high level and you don't really influence or change anything. So I try to do that and then we try to have our kind of management team that's about eight people that also all have like three projects they're deeply involved in. And that can sound from a distance like micromanaging, but in my opinion, it's really like mentoring. Like I think real mentoring isn't about sitting in a room and telling people the rules. It's about being there solving a problem and having a person watch the problem be solved and figure out what did work and what didn't work and what is your approach and what would I change and what is my style. That to me is like real development. So anyway, I think be close to the work, take risk.