Upstart Holdings, Inc. (UPST) Earnings Call Transcript & Summary

So happy to host Dave, and I love seeing Sanjay in the back. So it's a lovely day here. Dave, thanks so much for participating. We love to have Upstart here and hear more about the story. So thank you for coming.

Yes. Great to be here. Thank you.

So let's just kind of kick it off, right? The most common question that we get from investors is just the understanding of the differentiation in the lending model, right? And I'm sure you get this all the time. I'd love to just hear it from your mouth, like how do you describe the differentiation that you guys offer in terms of credit scoring? And what are you guys building there that's creating a barrier around the business?

Yes. So differentiation in pricing risk and credit is really the centerpiece of our business. So it's what we focus on more than anything else. And it comes down to applying vast amounts of compute with very sophisticated AI and machine learning algorithms and enormous amounts of data to make a better, smarter credit decision. And depending on your background, it may not be shocking that, it's feasible to build something better than a FICO score or a bunch of sort of human-created rules, which is what dominates the credit industry today. But maybe, I think if you're a technologist at all, you can appreciate that enormous amounts of data and computation can probably do something much better, more accurately predict risk. And that's the heart of what we do. And it comes down to -- I would say some people either can't conceive that you could make a smarter credit decision. It almost feels impossible, which I think if you come from the credit or banking background, you have skepticism built in there. Or maybe you alternatively think, well, everybody must be doing this. So are you really building anything different? And the truth is neither of those are true. It is neither impossible nor easy. And we've been, for 8-plus years now, building increasingly sophisticated models that predict the cash flows of a loan in a way and also sort of in very related, but different, also automate and eliminate all the friction of actually issuing a loan to a consumer in a way that makes the economics pretty dramatically better. And it's -- so maybe another way to just say it is that any technology business opportunity often is focused on markets where there's enormous inefficiency. And that's sort of a Jeff Bezos like thing. There's enormous inefficiency in consumer lending because models are very crufty and basic. It's not very hard, honestly, to build a better model. And there's a lot of reasons why others aren't or haven't and maybe that's a -- go there if you want to. But that's the heart of it is with better technology, a lot more data and the types of teams that can build sophisticated machine learning-based models, which aren't -- they don't exist in most of the lending industry. That's what it amounts to. And I think the results -- the financial results of the company that we've been able to deliver, demonstrate that super high growth and profitability aren't really by accident. They're happening because we're really part of a secular change in how credit origination is done.

Lot -- actually there's ton to unpack there. Let's go first in terms of just 7 banks you guys talked about last quarter are now no longer using the FICO score, right? And so there is some sort of adoption that you're seeing within the financial system of moving towards a more advanced model, right? Like why is that a big deal though? So help us understand, what's the second thought that's behind that as we see that number start to tick up over time. What does that tell you?

Yes. So we have 48-plus banks and credit unions in our platform. And the way we approach them, it's important to say is on their terms, like there's enormous amount of knobs and dials in limits they can implement within our system to manage their own risk. And a lot of it based on their own history and what they know about, what they're comfortable about. So for example, A lot of banks might just say, look, we just don't lend to anybody below 680 FICO. Well, if you want, you can enforce that in the system now. You can think of our system as having a first pass, which is the bank's rules, and a second pass which is our risk systems that are quantifying the risk of an application. And we do that because it's really important that we beat banks where they are. I mean they're very conservative. They're highly regulated. They have all sorts of reasons to not want to just jump into the future full bore. But having said that, once they're on the platform for a while, they start to see how it performs. They start to see actually that minimum credit score they have. And there's a bunch of other rules, that happens to be kind of the headline one, if you will, because it's a well-known thing, is unnecessary. It's not actually predicting risk. It's unnecessary, and it's actually excluding a bunch of people that they could successfully lend to in a responsible manner. So that's -- the heart of the truth is credit scores are not meaningless or worthless, but they're extremely limited. And as banks get more comfortable with us, they tend to move in a direction away from having these hard rules because they develop more trust in the system to make -- do the right things for them.

How do you show them that though, right? I mean if I think about the opportunity there is you're basically widening the funnel, right, as you move off of the FICO score. So is that just a sales process? Or is that -- like how do you educate them on the opportunity?

Well, it's not very hard. They can log into their what we call performance console. They can see every loan that they have issued, they can see the performance of those loans. They can see what the loss rates for different pools were predicted to be and what they actually are. And they can go, wow, I can just look at this and see that credit score was actually not very predictive here. These sort of risk tiers that we define for them are actually much more useful and predictive. So they finally go like, why are we doing this? And so it's very simple. I mean, ultimately, it is SaaS software, if you will. They log in, they run their program. They get to see the results of it, and they get to make decisions about do we want more volume or less? Do we want to take more risk or less. So we really -- again, our goal is to meet banks where they are today, not to like make them -- make this crazy leap into the future. We're going to take them one step at a time, and I think that's served us well.

That makes sense. Switching a little bit. So you mentioned horizon and default rates last quarter. It's more of the normalizing the credit environment overall. I think we all get that. But help us understand why this was expected. And what changes does that mean for you guys, right? Is that just kind of you guys being offering higher rates kind of a period of time ago? Or what exactly does it kind of look like? How does play with the system?

Ye. I mean it's, first of all, important to realize lending. I mean, it's not surprising to anyone who's in the industry, lending is a highly cyclical industry. It's very tied to the state of the economy, and things change quickly when the economy changes. So anyone who's been in banking or lending forever, of course, knows like you're trying to watch the tea leaves or read things or see and react and do things responsibly. And ultimately, our system, we believe, is much more adept at reacting quickly doing the right things through a changing scenario. To directly address your question, when COVID hit in March of 2020, almost everybody -- almost every bank in the country probably just put the brakes on hard. It was such a -- unemployment went from 4% to about 14%, literally in a handful of weeks. Just kind of -- I mean that's never been seen before and nobody really knew what we're in for. So everybody on our platform, and I think everywhere, really hit pods. And very rapidly, we started to see -- actually, our loans were doing fine, and the banks turned things back on quickly, but they didn't necessarily know that. They didn't have any -- enough history with us to know that. But what also happened and not just to us, but across the industry, is default rates were just ridiculously low. And this, of course, was because the government was sort of air dropping checks on everybody. The stimulus out there and also the fact that consumers weren't spending. So they're -- sort of debt carried was dropping. It really, in effect, maybe shocking to everybody, it was actually a net positive for the consumer in terms of their financial situation. And that amounted to loss rates that were probably -- our sort of true north is what are the loss rates relative to what we predicted them to be at the time the loan was originated. As long as that's good and always getting closer and closer, that's what we want. And what we were seeing is loss rates, 40% to 50% below what we estimated them to be at the time of origination. Now left to itself, our models would quickly just go, oh, great, the world has changed. It's time to lower rates because people don't default anymore. But -- and this is where it is important, an AI-based system doesn't know what it doesn't know and the humans need to intervene. And in this case, we introduced, 2 years ago now almost, the notion of stimulus and how that is buffering credit performance. Now we also knew -- so anyway, we communicated to our bank partners and credit unions and loan investors like you're getting an enormous surplus right now. Default rates are half of what they should be. And you're getting enormous amount of sort of excess profits, if you will. But guess what, it's not going to last forever. We fully expect when COVID is sort of coming to an end and the stimulus dries up, we will see default rates return to, without knowing much more, 2019 levels, right? And that's essentially what we've seen. It happened starting, I'd guess, maybe in Q4 and has continued. And now I'd say we're probably pretty close to where we were pre-COVID. So yes, the default rates have increased a lot. It is as we expected. We are now in a position where the world is normalized. Now the world in 2022, of course, has all these other variables all of a sudden happening that are a little unclear. Fed raising rates, the prospect of recession, geopolitical issues causing other sorts of disruption. So the bottom line is like there's no such thing as, oh, everything is just normal again. We're just dealing with the next set of disruptions. But I think what we're doing over time is proving to our bank partners that this is the system you need to deal with a noisy, crazy background economy because there's no reason to believe it's ever going to just be steady state again. We may or may not hit a recession next year, some debate about that. So that's the heart of it is that lending is a cyclical industry. We aren't going to change that, but we can certainly help lenders navigate that more adeptly.

So one of the things you said there that kind of pricked my ears up was just the fact that your system recognizes stuff faster, right? So do you feel like we're now -- is it stabilizing? Or kind of what's the second derivative, right? Or what...

Yes, that's a good question. We do see it stable now. We saw essentially default rates starting again, I don't exactly know the month last year, but it's second half of last year, maybe towards the fourth quarter. And I think we now see it pretty stable and pretty similar to 2019. So we kind of view it as, at least in terms of that sort of notion of stimulus going away, COVID going away is sort of mission accomplished, if you will, because we think it's now behind us. We're also in a seasonal -- usually, the first quarter is a trough of demand. It's usually our seasonally slowest quarter. So there's some of that always in there. But no, we feel good. We're in a good position. Our system did all the right things. And now we're back to at least with respect to COVID and the stimulus and the effect of that. I feel like we're largely beyond it. There's probably some parts of it that's still remaining.

Okay. Transitioning again, auto, right? So you guys set a $1.5 billion auto target. Look, that's an impressive number. But it feels like it's very dependent on auto refi, right? And the bigger opportunity is originations. So just talk to me around the progress that you're making on origination. How do we think about that portion of the auto market? And what are you guys doing to drive that?

Yes. The terms we would use is auto refi and auto purchase. But in any case, yes, we've been at refi -- auto refinance for a longer time. So maybe not surprising to us that it's the first one to sort of lift off, if you will. And it has. I mean we -- last year, since we've been public, we were saying, look, auto is our next thing. We're focused on it. We have a lot of confidence that the market opportunity is large and we can enter it. And then we came into this year and we're able to say, hey, it's happening. We've achieved liftoff and put a number out there as you suggested. And auto, what we call auto retail or auto purchase from a lending perspective is something that we decided to get into a while ago. A little over a year ago, we acquired a company called Prodigy to provide the software that puts us in the dealership. And I think it's never say never of what could be larger or what have you. I think we view the current auto refi product is much more similar to what we've already done. So we know the drill towards improving the models and scaling up, improving the conversion rates and all that. And that's midstream right now. I think it is fair to say we view the purchase opportunity as much larger because we've always had what we call proprietary product, which means a smarter credit decision built in, more automation built in. But with the auto purchase product, we have proprietary distribution as well because we own the point of -- so it's a little bit like as everyone always likes to note, in our personal loan business, we work a lot with LendingTree and Credit Karma. And in those cases, they're the first thing the consumer sees. While we're in the car dealership, we are LendingTree and Credit Karma and Upstart all rolled in one. And so it's an enormous opportunity. That's -- I think we feel confident that, that's going to -- that has the potential to be our largest business by far in a few years. Just because it's moving very quickly, but we're still putting a lot of -- there is integration to do. Dealers have some kind of archaic management systems. They used to run the dealership. You need to work with all those. So there's some sort of technical hurdles to get through. But none of those are, in our view, remotely risky. There's just a bunch of process. But we are very, very optimistic on that product. But it is fair to say, when we put that number out there, we feel very comfortable that the refi product has achieved liftoff. And hopefully soon, later this year or at some point, we'll say the same about the auto purchase product.

For purchase, just help us think through, right? So there are existing number of lenders that are already out there that are offering a lending product to their consumer as well as to the dealer because they're part of the equation, right? That a huge portion of revenue for dealers. So just help us -- like how do you guys win that competitive set of -- Upstart becomes that purchased loan?

So think of it as right now the software, what we call Upstart auto retail is going into dealerships at scale. It was in about 100 dealerships a year ago when we purchased the company. It's at maybe 4x, 5x that number today. So this software is what the dealership is using every day, day in and day out to sell cars. And it's part of that to show financing options to the consumer. Now until today, that -- those financing options had nothing to do with us. They are the ones that come through the legacy dealer systems. They come from people like GM Finance. They come from Ally, they come from Westlake. So everything from subprime to sort of the OEM manufacturers offers running through our software. And then suddenly now in a very small number of dealerships as we're just testing this, you're seeing offers from our bank partners in the same place and competing. And we are increasingly confident because we're now starting to see this that our offers are going to compete really well across a lot of the spectrum. And so suddenly, we're in a position where we don't have to -- we're not going to win -- our banks are not going to win everything, but we can also go out to anyone, any lender who's already working with dealerships, and that could be any of the ones I mentioned and say, hey, there's no reason you can't work through our system as well. And in fact, as usual, we'll have the same value proposition, you can improve more people at lower loss rates, like that's a pretty good value proposition. So our view generally is there's no enemies for us in here other than when our software is out there, our lenders can work through the traditional system. Their loans will be available through our system or they can work through our lending platform, and we think we'll have a better proposition to them. And that's the heart of it is there's maybe 16,000 or so, maybe 20,000 dealerships that are franchise dealers and large independents. And there's kind of a long tail. But I think when once you're in sort of maybe 15,000, 20,000 dealerships, that's full market penetration. And we're well on our way. I mean, I feel very confident we're going to continue to grow that presence really quickly. And at the same time, in a sort of parallel of it, push our lending products in there. And that's going to be an enormous win for the consumers as well as for the dealerships and as well as for our lending partners. So there's kind of like all these 3 parties that, in our view, all of which can win here, and that's pretty cool.

Yes. All right, contribution margin. Right. So we just talked about auto, that's going to be a headwind as it ramps, understood 2022. But you guys also have multiple new products coming online. I want to talk about that later. Help us understand contribution margin kind of over the next 3 years, right? How do we think about you guys just managing contribution margin, if you will, or as an output, just given the product ramp and everything coming online now over the next 3 years?

Yes. So I would say, first of all, I mean, we have the luxury of having an extremely profitable first business. And in some sense, I might say it's more profitable than we would have attempted in the sense that at our stage, of course, we're mostly focused on growth. I think it's great that we are a unique fintech company that is actually triple-digit growth and as well as generating a lot of cash as well as GAAP profitable. So there's just not a lot of flaws to find there. And we don't want to change that. But having said that, we are at an early stage. These are very large opportunities, small business loans, eventually mortgages. And again, we're just at the very first stages of auto. So we're not necessarily -- we are certainly not trying to optimize for contribution margin, but we love being a cash-generating business that controls our destiny, particularly at a time like this, the world is a little crazy. It's awesome to be so profitable. So I would just generally say, look, we go towards sectors that have a lot of inefficiency where we can have high contribution margin. It certainly takes us a while in each segment to get there. I mean there's an exploratory phase and all that. But I think the good news is that you would expect our core business to dominate the economics until auto gets to a certain scale and then the contribution margins will come into place there. And we'll, by then, be into another couple of product areas, but they'll be small at that time, so that they won't terribly distort the contribution margin. So I guess, in other words, as long as we continue to grow and add new things to this very large business, the contribution margin will go up, it will go down. But a new business like us jumping into new business isn't going to crash the whole company's contribution margin because, frankly, it's -- we're always starting from a very small position.

Okay. Okay. All right. Kind of a big picture question here, right? So you talked about being a tech company last call, right? And so I guess this got me thinking and one of the things about tech companies is that some of them now have very large brands, right? And so how do you think about Upstart in terms of building a consumer brand that starts to know Upstart as a product, as a consumer brand versus kind of being the tech partner that you've kind of talked about?

Yes. We're an interesting business because we're kind of this B2B technology provider to the banks and credit unions. But at the same time, we are a consumer brand. And we like that mix. I mean, it makes us hard to comp against anybody because we don't quite look like anybody. But at the same time, it's really the source of our advantage is having both sides of this. And I would say Upstart as a consumer brand. I view it -- I like to think of it as an Internet brand that will sort of mimic some of the most successful Internet brands. By that, I mean, you kind of know what -- I mean I was for 8 years at Google before founding Upstart. So I'm obviously pretty intimately aware of Google's history and its brand, you trust it for certain things. Now whether you do today as much as 10 years ago and I left, I'll leave that to the audience to decide. But there's a reason people start at Google when they're looking for something. It's because they trust if it's not in Google, it probably doesn't exist. And then Google will probably get you to the right place. Similarly, if you look at Amazon, I mean, I know I personally will almost always just buy something on Amazon, if it can be there even -- I don't need to know it's like exactly the lowest price available, but I trust Amazon to do all things right and they generally do. So I like how those bands resonate.

So what do you -- what is that for the consumer though? Is that the best credit outcome?

Yes. I mean, it's trust that if you come here, you're going to see the best rate available to you and you're going to have the very best process, which is, of course, no process. Nobody wants to go through hurdles to access credit. So as we expand the categories we're in, we want us to be the place where you can think, when I need to access credit, this is my first stop because I know this is the best process, the best rate, the best everything. And we just have to make that a little bit more true every day. That's an Internet brand. It's distinctly different from someone who's a bank and becoming a digital bank because I think digital banks will always have a very specific demographic they target. You sort of have to, I think, particularly the way the market is evolving, a bank is hard to be everything to everybody. But an Internet brand actually can, right? You actually can say, look, this is the best process. We're going to connect you to the right product because we are really -- we're not a principal or an agent. Maybe there's one way to say it. And I think that's actually really important because the upside and the potential to build something very large is there when you actually can reach out and serve virtually everybody.

We're under 5 minutes. Happy to open it up if anyone has any questions.

Can you speak a little bit to kind of what you're replacing within the bank organization as far as a lead gen, I'm talking about how your bank partners can manage or how you sign a big partner to grow that component of the network?

Yes. So well, essentially there's probably a couple of scenarios, but we could be bringing a product class that the bank is not in today because in personal loans, for example, most banks don't have meaningful personal loan business because historically, it just wasn't economic to do it. It was a marginal business, not very profitable, sort of a favor to good customers or something like that. But suddenly, it's enormously popular with consumers now. So in that case, we're kind of introducing a completely new product to a bank that's probably not -- had it in a meaningful way. And they can certainly rethink how they allocate their balance sheet and what they can offer to consumers, et cetera. So it's that case -- in another case, may be more common in the auto side is suddenly, we just have a better product for them. I'd say auto refinance is historically what I would call a $0 billion market. It's a good notion, but it's been so difficult for the consumer to get through a process. You got to go get DMV, you need to get a notarized signature, et cetera, that most people just don't do it. But we're eliminating all of that. And suddenly being able to -- a bank to offer to its own customers refinance of their auto loans, which is just a great way to increase the sort of, what they would call, share of wallet is right there. So in some sense, we're a fully digital product. Yes, we can connect to an in-branch experience, but we predominantly are a digital product. And that is something they mostly aren't -- they haven't made the leap to today in a meaningful way. So it all tends to be quite new.

What do you see is the biggest governors to your growth over the course of the next years?

Yes, that's a good question. I will just say this, first and foremost, it is execution in the sense that I don't have any question the market opportunity is there. I don't have any like inherent fear that some competitors are going to run in front of us and swallow up the opportunity remotely. But executing in this market is sort of a multidisciplinary thing. There's a lot of things you have to get right. And we could screw things up. We could -- this could happen slower than that. Technology infrastructure, can it scale? The way we manage our teams, can we make decisions fast enough? And can we deal with whatever the economy throws at us. So mostly it's execution, and I spend a lot of my time and energy, like thinking about how to upgrade our quality of execution because we're going -- we've already, I guess, gone from a one product company to a multiple product company. And we're going through different economic transitions and all that. And it's a very regulated industry. So dealing with regulators and lawmakers and consumer advocates and all that part. But that's the heart of it. It's execution, and I feel like we're in a sweet spot to do this right.

Okay. So we're at time. But Dave, thanks so much for joining. Appreciate it.

Yes. Thank you