The Exchanges

Q loan tape and vintage history do they have to build to drive their costs from 12% and one percent warrants down to, you know, eight percent or bringing a bank on top of the credit fund or whatever to drive the blended down. What did you back then have to grow your loan tape to to get significant savings below that 10% cost of capital on your first hundred million?

A You know, uh, that's a great question. Um, I think there's a couple things. The weighted average life of an asset, even though you write it for 24, 36 months, you know, people end up paying it off in 18 months, or they pay it off in 16 months. So, you know, you, you know, within about a year, a year and a half, you have a pretty good idea of how the, how the curves are going to work, uh, because most of your losses are really front-ended. In the first six months, you'll see about 60% of your losses on a vintage analysis curve. So it's relatively easy. Once you get past six months, They can kind of predict the rest of your curve. And, and so a year into it, you've got a couple of turns of products. Um, but you know, it really wasn't until we got over, call it a hundred million dollars of, of, of transactions until they said, look, you got enough scale, enough predictability, and a couple of turns of the product that we feel comfortable in giving you better, you know, better pricing and better advanced rates.

AI assessment note: “wasn't until we got over, call it a hundred million dollars of, of, of transactions”

Q Well, that's what I'm asking. So that two to that two 20, like that capital, you started your balance sheet on day one. Was that like a prom note on the operating company or was it actual equity that your friends and family, they put money in for equity in the business?

A A little combination of both. Some of, some of just pure equity. Some of it was, uh, you know, more mezzanine type structure, right? So a little combination of both, but it allowed us to use that money to leverage, to be able to get us, uh, you know, access to capital, uh, be able to, to, to test our models and, and, uh, you know, make sure that our models were as predictive as we were hoping they were going to be. And they, they, you know, they became that, uh, over a short period of time and therefore then the cost of capital, uh, Continues to go down as your, you know, as your performance, the models are better. It's magic that way, right?

AI assessment note: “A little combination of both. Some of, some of just pure equity.”

Q What number grew by that amount? The capital deployed or the revenue?

A The, the, the, uh, the funding levels. The funding levels. And we'll talk about revenue here in just a second, but the funding levels themselves. So we grew, we went from roughly, um, about, you know, we, we went to, uh, to 2.1 billion And 21, uh, which is a, you know, a huge, uh, you know, a huge growth, but we slowed down a little bit during COVID in 2020. Uh, but then we had, you know, just significance would close a little under a billion, then went to 2.1 billion. So we had a significant growth kind of year over year, you know, this year we're already starting out about 800 and about almost nine hundred million dollars.

AI assessment note: “The funding levels. And we'll talk about revenue here in just a second”

Q So that was your sort of thesis. And then it scaled from there, I guess, take me, take me back to one of those early deals. So I'm a consumer. I have a great, you said a credit score above what?

A So typically in the early days, I credit score would be around six, 26, 25, maybe in that area. Okay. And they were somebody who's Well, it was either somebody that was lightly, you know, had light credit footprint, right? They were just getting established, and nobody could really kind of put all the other kind of API and data information together to be able to tell their story. So there's a lot of other things that you would look at outside just credit. You might look at phone bills. You might look at rent history. You might look at some other things to tell the story of their willingness to pay, right, or their ability to pay. And so those are the things we, you know, we really focused on. We focused on, I mean, one of the problems you always have is fraud, and we focused a lot on KYC, know your customer. Uh, and we were able to get a very predictive outcome on those particular scores. And so, you know, this was typically somebody that was either on the way back up, you know, I had gone through a dip or just had a very light credit footprint.

AI assessment note: “typically in the early days, I credit score would be around six, 26, 25”

Q warehouse facility. The problem is you end up with unused fees if you can't attract customers and deploy it quickly. It sounds like you just told me in 2018, you had four hundred million in capacity, but you did three hundred sixty million. That's very good optimization in terms of actually utilizing what you, what you took down on the facility size. How did you plan that so, so well?

A Well, you know, um, the, the nice thing about growth is, is, um, you know, our models have really led, our AI models from a marketing perspective have really led who we go after and how big that TAM is, or that total addressable market. So we knew basically based on our efforts, what we needed to do to grow the next ten million, next twenty million dollars a month, right? And so what you're really trying now to do is line up capital, right? The capital that you need. To, you know, to, to backstop that advance rate and grow and continue to grow the business. And so, you know, that was all kind of coming together at that particular point. Uh, and really limiting factor was capital. It's just how much capital you have on the books. Uh, that was really more of your limiting factor, uh, more so than the capacity of the lines.

AI assessment note: “our AI models from a marketing perspective have really led who we go after”

Q And, and what you know better than anybody, what are these kinds of companies getting valued at today? Is it a multiple, like, let's look at 2021. Is it a multiple on your 330 or do you get a multiple, you know, lower multiple on loans done? 2.1 billion.

A You know, it's a challenge. This is the challenge for, for, uh, for valuation, right? The challenge valuation is a very fast growing company earnings lag. So, so typically, these are discounted cash flow models and really looking more at a multiple of revenue. Typically, a lot of times it's either forward or, or, you know, post, but a lot of times they're, they're looking forward now because on a really fast growing revenue company, they'd probably look more like in our case, the six hundred million dollars than they would look at the three 30. Um, you know what I'm saying? So, but that is a valuation challenge because a fast growing company will continue to have massive scale and efficiencies of scale and growth and revenue that won't show up in earnings in the first year. So you really kind of need to do it more as a discounted cash flow. We see primarily over a five-year period or even a ten-year period, uh, to pick up the value of the, of what you're creating.

AI assessment note: “they'd probably look more like in our case, the six hundred million dollars”

Q most, most businesses like this have trouble scaling for two reasons. They have to fight yield compression, right? Because more money will plow into the market, right? If it's a known, known asset class, right? The second is your Google ad expense goes up, right? You have to have CAC arbitrage somehow. So how have you fought both of these, you know, headwinds on both sides keep scaling so fast?

A Well, look, I think the question has always been in the fintech space. Uh, is it scalable? Is it predictive? Right. And is it sustainable? Can you grow it? Can you continue to grow it at the, at the appropriate rates? And, and I think we've answered those questions. Uh, you know, I think, you know, first of all, uh, the biggest challenge you have is you have to be able to optimize your cost to acquire a customer. And that really comes through your ability to renew a customer. I mean, we have an Net promoter score. We work really hard on our customer, uh, to make sure that our customer is, is having a great experience because they have a great experience. They come back. They see you as kind of that trusted advisor. So I mean, right now, uh, about 25 to 30% of our base runs in renewals, just renewals alone. These are customers just coming back.

AI assessment note: “optimize your cost to acquire a customer. And that really comes through your ability to renew”

Q And today, even back then in your pro formas, when you're, you know, building in sort of a charge off or a losses or bad debt expense, is this two percent, three percent, four percent? What do you build in as a buffer?

A Well, you know, really what you're doing is the AI models have done an amazing job of predicting risk. Uh, and so, you know, what, so the way that the AI models work today is it predicts risk, puts you in a category, uh, and then tells me, okay, basically here's what the risks are going to be. But then pricing is the next kind of, uh, uh, uh, optimization tool that we use. And we have about five different buckets of credit grades or risk, right? And we, but we now are up to 400 different pricing points with inside of those grids. So we are getting really, really good at giving you the right product at the right time at the right place with the right terms and conditions that you can understand how affordable it is for you to, you know, finish a project or, or to, to, you know, to resolve some consolidation of bills or whatever it is that you need to do.

AI assessment note: “really what you're doing is the AI models have done an amazing job of predicting”