Affirm Holdings, Inc. ($AFRM)

A consumer-led recession or the consumer is about to take a major downtick, whether it's oil prices or some other things that we don't know about yet. Let's go through that? How do you manage that downside scenario -- what do you see first, which says, hey, we need a meeting on this. And then how does that progress through the system?

Yes. The analogy that we've used historically is that we always have our hands on the steering wheel. So even in benign times, which I would classify today, frankly, is a pretty benign consumer and credit environment. even in benign times the entire executive team is looking at the weekly credit report that the risk team is pulling together. As I mentioned, first payment at both the 4-day mark and the 30-day mark. Those are really, really important and primary internal KPI that we manage the business to. And that, as I mentioned, that really is that sort of check on for the most recent cohorts when they get to that first repayment event, did we get the underwriting right for where the consumer is today in terms of being stressed or nonstressed -- and so as much as the macro is, of course, important to our business, headlines about the cost of oil or rising unemployment rates. If it's not showing up in the delinquency rates internally, we're probably not actioning change proactively, to be honest. We can do some things on the margin due Burton durations a bit more or potentially increase the level of down payments that we're asking a consumer. Both of those take risk out of the system. And we can do that in a really lightweight or small wave if we want to. But honestly, like in terms of actioning a company-wide change in our underwriting posture, we would really have to see it show up in the first payment delinquency data. I've been putting aside something like maybe COVID where it just was sort of this Black One event honestly, the company actions very quickly there. But if it's sort of a typical distress is starting to creep into the consumers' financial lives, then it will show up in our data because, again, I think we originate enough loans that it's going to show up there, and we'll start to work on getting the models to sort of fit where the consumer is. And it may mean that we take some risk out of the system. We can do that in several different ways. We talk a lot about loosening or tightening so much more nuanced than that. I think the IR team does a bit of a disservice to the incredible risk team that we have internally. But yes, we can raise the minimum credit score for given merchant, that's typically the bar that the transaction has to clear to be approved. We can be incredibly fine grained. Some of our programs run it is the score that you need to clear and we can go to 92.7% or 92.75%. So we can be incredibly fine grained there if we need to take risk out of the system. I already mentioned increasing the down payments for sort of the marginal borrower if historically, if the consumer has some skin in the game with the transaction via down payment, we tend to see better credit outcomes there, better repayments -- we can shorten the length of the loans, that takes out of the system for us to -- we can increase the APRs if we wanted to or needed to. There's just -- there are several things that we can do to take risk out of the system. And the good news is we can action them all really quickly. I think we've got a really well-defined playbook for how to roll this out across the merchant base.

And the change is going to be gradual, right? It's not like Wednesday afternoon, everything is fine, and then Thursday morning, like everything exposed, right? This is going to happen really gradually. So where would you see -- if you were to -- this is the other thing we should talk about and everyone says, "Well, firm really hasn't seen a down cycle yet.

Yes. You could say, well, late '22, early '23, was kind of sort of it was short-lived. But when would you -- how long would it take for you to start saying, okay, our rates are starting to look like they're forming a trend I would think on I would say typically, if the new cohort of first payment delinquencies comes in and there's an elevation in terms of delinquencies versus expectation. That's all it takes to prompt an investigation. And when we do the investigation internally, we want to make sure that we really understand what's driving the deviation, right? Sometimes there can be a messaging bug or sometimes something could have changed with the merchants configuration. And so we're typically looking to understand is the stress that we're seeing in the data? Is it limited to 1 merchant? Is it geographic? Or is it broad-based? And so the answer to that question will sort of inform how we triage the remedy. And again, we can be really quick in terms of raising the underwriting approval thresholds and about other sort of levers we have to take risk out of the system.

And so we were -- in the company that I ran, that's how we ran risk as well, right? Like there was 1 merchant where a few locations where driving our loss rates through the roof, so we just turned off those locations. So everything was fine. -- right.

Okay. So you look at all that data, and this happens gradually over a few weeks/months, depending on what you're seeing in the data. Yes. But again, it's like really, that first payment delinquency tells us a lot -- that's the -- is the underwriting right. Like is the model set up properly to predictably drive credit outcomes? And if we see drift, we take that really seriously.

Yes. Is it a foregone conclusion that in a macro slowdown you guys automatically have lower growth? Is it definitely a foregone conclusion? Or could you say, "Hey, if the traditional lenders pull back more people are going to want to go to buy now pay later, we could actually see steady growth in a downturn even though we're scaling back on our risk.

I think it's definitely possible. I mean, I do think maybe firm and by opalater broadly are taking continuing to take meaningful share within U.S. consumer wallet. So it's not out of the question that we could continue to grow through a downturn. But again, I think -- the most important thing is that we're driving predictable credit outcomes. That's important to a firm, but it's also really important to our funding partners. And frankly, we're so aligned with the consumer. We think it's good for the consumer, too. We don't want to put loans in the hands of consumers that can't afford them and won't be able to pay them back. We don't profit from that. It's not good for the consumer. It's not good for their financial lives. It erodes trust, right, with -- between the consumer and the firm. So really, again, we want to get it right. I think we have every incentive to get the underwriting right. Let's go through the -- let's continue on this track of the down cycle whenever it will happen.

You guys perform -- well, I'm assuming margin stays relatively flat, right? You manage to the margin -- maybe growth fluctuates. Maybe it's down, maybe it's steady, maybe it's up a little bit depending on how the market is changing.

But on the other side of this, if we kind of look forward, people are going to say, you guys manage through the cycle pretty well because of the -- you are underwriting a specific transaction at a specific point in time your algorithms are terrific on the risk side. And so the margins stayed flat. You didn't have any major losses or provisions or anything like that, like the traditional creditors will have -- and yes, maybe growth fluctuates. It's down a little bit, it's up a little bit ever, but you guys sell through, and now we're back to business. I think that's right. Every recession is different, of course. So it's hard to be precise in this hypothetical. But I think what's important to remember is that -- we set up our financing programs. And the other important thing is like every merchant is different. Every merchant has a different cutoff because there's just different product mix, there's different economics in each of our merchant relationships. But -- in theory, we're setting every merchant relationship and every financing program with a merchant such that the last loan that we approve is breakeven or better. And so if you do introduce stress into the system, what are we doing? We're raising the threshold for approval and we're taking sort of marginally profitable or probably breakeven loans out of the system. And so it should not be a significant drag from a profitability perspective, we are giving up growth, of course. That's the KPI that we're going to see potentially soften. But again, we think that's important for the brand promise that we have to consumers. We think it's really important to the funding ecosystem that we've built as well that like if there is stress, we're going to course correct and we're going to manage through whatever credit environment we're in.

Okay. All right. I'm done with the downside talk about happy stuff. Did I kill that enough -- did I beat that 1 good enough. Let's talk about underwriting. I was fascinated that you guys aren't really leveraging cash flow underwriting at all up till now, which is -- for those of you who aren't familiar with cash flow underwriting, it's getting 90-day transaction data from a third-party data aggregator where you can see inflows and outflows in the underlying account. The fact that you guys were able to do what you do without seeing that data is fascinating to me. In my business, we wanted that data and we couldn't get it because we didn't qualify in the transaction type. But talk to me about cash flow underwriting and how much better you think this could make your system? Could you actually see -- could it -- are people underestimating the impact it could have to volume growth just because you're going to have much better data in addition to the algorithms that are already top of class.

Yes. Look, I mean there's a whole team at a firm focused on developing and then testing and ultimately graduating the next underwriting model and things like cash flow underwriting are just 1 ingredient to go into sort of the incremental gains that we see every time a model rolls out. I think we're on -- we call it POS, point-of-sale model. I think we're on 13 now, it's around for about 15 years. So it's a huge effort. It takes about a year to develop these models, sufficiently test these models and prove to ourselves that the models are going to do better in terms of conversion rate for the merchant credit outcomes for us, approval rates? Like there's a whole bunch of criteria that are often diametrically opposed here to make sure that the model is truly better for everybody involved. So it's a really high bar again, cash flow underwriting is part of that. We ultimately -- we want to find a way to get to a yes with the consumer if we can, right? And so down payments are a part of that. internally, we call them step-ups, asking for more information, that's where cash flow underwriting would come in. We really want to make sure that we have the fullest possible picture of the consumers' financial health before we get to that ultimate decision. So yes, I mean, I think it's really -- the early results are promising, and it's a really interesting way to sort of step up and try to get to that marginal approval where we can. In your mind, what's the optimal percentage of transactions that would go through cash flow underwriting I think the other thing to keep in mind is just most of our consumer transactions today are what we call repeat borrowers, right? Meaning it's not their first a firm loan. So we tend to over a consumer's life cycle, we tend to rely more and more on the consumers' repayment history with us, like they're a firm file start I think where it helps us is just in some of the growth areas around acquiring new users at a new merchant maybe. So can be really compelling there because the early days of a merchant program are really critical to get on that steep trajectory of growth. Yes. I know that firsthand. So cash flow won't necessarily be applied to everybody, but in certain use cases for certain parts to be really compelling. Yes, totally. -- love that. Let's talk about credit. We've done a lot of work post the Post the controversy not contrary I forgot the name Ridge. One of your credit people en -- so thank you. That was a middle age moment. Rob, thanks for bailing me out. So when Stone reach came out, we really delved into it and your creditors love you. They love the paper and you take the same risk that they're taking. And I think 1 of the things that stood out to me early on was you guys said like, look, we will forego growth to make sure that our creditors are happy because that's our lifeline for the next for the -- as far as the eye can see in terms of growth. So we don't really see a problem there of it. The ABS deal you just did was really great. the agreements you have with all your people in the credit facility are terrific. It's kind of like that concern has kind of come and gone -- do you still get a lot on that? Or is it I think we still get questions there, especially for investors that are maybe new to the story. I just wanted to understand all the mechanics and sort of the broader funding ecosystem. But yes, if you ask me for 1 KPI on the the health of our funding ecosystem, I would say the terms that we get in the most recent or the next ABS deal is probably the best leading indicator of just the health of that environment in that market. And right now, I mean, honestly, we're seeing spreads at all-time lows on a like-for-like basis across recent deals. And so it feels like the market's really healthy today. And they like quality. And if things go south, they're going to flock more to quality. And if you guys are viewed as that quality paper, then I think that's okay. I think that's a good outlook for you.

Let's talk about guidance a little bit, switching gears from the investment forum, you built a great reputation for kind of beating and raising guiding conservatively make it look easier than it is. I know you guys -- it ain't as easy as you guys make it look. But if the 25% medium-term GMV growth target proves to be conservative over time. At least -- you forgot 2 words there.

Yes. I have greater than finest Yes, it's important. Sorry, -- that's a lot my glasses to get that one. So it's the at least 25% growth people are going to love that correct. By the way, I'm just thinking about all the investors on the call here, they're going to love that correction. Where did the upside come from? Got you. I mean if you look at sort of the drivers of our growth today, today, right, both point of sale and the direct-to-consumer businesses are both growing far in excess of at least 10% we led out for both of those programs. So again, if we outperform that, it's because we're sort of maintaining what we're doing today, right? And then we've got sweeteners on top that we called out with international becoming a bigger part of the story over the next several years. And then we've also got I think, really good irons in the fire around both the firm edge and AGTC 2. So look, I think we've got 2 programs and projects that could be meaningful in the medium term that we're not ascribing any sort of growth to in that at least 25% target in agent and a firm Edge. So I think those could be contributors to outperformance. And again, just the base business today both sides, POS and DTC are both growing significantly more than 10% today. So I don't think we're we're not asked -- we're not like envisioning a world that doesn't already exist. I think it's just about continuing to execute and continuing to sort of grow with the primary drivers that have fueled us to date. It's worth reiterating that the genic and Edge aren't really in the numbers yet -- they're not in your -- that's all. And I think that's appropriate, right? Hopefully, you've seen from us over the years that we tend to be pretty measured about signing up for big numbers for new programs. We have a lot of confidence that we've got the right playbooks and tools to make these programs large where we can, but we do take a pretty balanced view and a pretty conservative view on new things generally. So it Doesn't mean we're not excited about it, but just in terms of how we guide I think that's the right way to do it.

No sense of setting up for failure and false expectations, right?

Okay. So it's -- everything you're seeing today, plus a couple of other things that could be more meaningful et cetera. Okay, cool, and you did say at least -- at least 25 -- that could be the title of about right here -- maybe we'll do that. Okay. Let's switch to R LTC guidance. I thought you'd go to like 3.5% to 4% and kind of -- but 3.75%, that's a tight rate for RLTC. You and I have had this conversation offline. I think RLTC was viewed as kind of like a all-in measure about consumer health, but there are so many factors in there like gain on sale and all -- there's so many factors in our LTC. It's not really a proxy -- a pure proxy for consumer health. So the 3.75% to 4% like is there a message that you're trying to send folks with that narrow range? I think that maybe the broadest message would just be that we feel like we have good visibility into the business. And I think that's rooted in the fact that we've been really active on the ABS side from a funding perspective, same on the forward flow side as well. And so we have a pretty good sense for where the funding is going to come from. Of course, we're going to grow and we're going to need more funding along the way. But just the sort of layers that we've been able to put into the funding base, we've done 3-year ABS deals that are at fixed cost of borrowing. So it does insulate us against a movement upwards in an upwards rate environment. Similarly, with the trajectory that we're on with a firm card and also just where we are with both Shopify and Amazon, our 2 largest merchant programs, we just feel like we've got really good line of sight into how those programs should perform over the next several years, certainly from a profitability perspective, and that gives us confidence that some of the error bars, the AirBar being a point wide. It just -- that felt like more room than we needed for these next several years of operating.

Okay. So it's a tight range, which I think is good because it almost like removed that from the equation kind of thing, right?

So even though people probably still have a stroke if they see a 3 handle, but somewhere around 4%, I think, is great. I remember the comment we made at a group launch. I said, Rob, if you do 390 if you do 41, you like walk on water and you looked at me like see -- and again, it's just -- again, like coming in, I mean, we're at the tail end of our budgeting cycle for for next year and just as we think about all the levers as we think about all the building blocks for growth and profitability in a given year. Again, we just -- it just feels like the error bars we have on those are just a lot smaller than maybe they were 5 years ago when we established the 3% to 4% range. And I think it's good, it's good for the narrative too because it pushes everything to GMV growth and margins. And that's really what you -- at this stage of your development.

That's really what the story is about. How fast are you growing and are you increasing your profitability? And it doesn't need to be more complicated than that. Okay. Let's switch gears again. We got a little under 10 minutes left, as CFO, how do you think about capital allocation? -- we need to over the next couple of years. You see the focus, obviously, very heavily on organic growth as you should. GMV is accelerating year-over-year ex Walmart. How do you think about M&A? How do you think about buybacks? How do you think about dividends Talk about how you rank all the.

Yes. I mean I think just maybe pointing to some of the things that we've done historically. We've been pretty active around buying back the convertible issuance that we did in 2021. that first tertile bond that we did is going to mature in Q4 of 2026. So we've been chipping away at that, especially there was a period in time where those bonds are trading at a pretty meaningful discount and we felt like that was kind of a no-brainer move in terms of capital allocation with sort of buying those back, buying back that future liability at a pretty meaningful discount. And so I think that served us well. And so when there's been opportunities to allocate capital to something that we think has a really, really high return. We've done it. I think we're still very early, though, in generating cash. I'm really proud of the cash that we generated in the last year. But it's still early days there. And we spent a lot of time today just right here talking about recession planning and recession scenarios, and we do stress testing ourselves within the capital team and the treasury team to make sure that if that rainy day comes that we feel good about the balance sheet that we've built and our ability to fund the business in any environment and through the cycle. So it's cash is an important part of how we think about stress testing and scenario planning for a downside scenario. So it's it's a high bar for us to sort of distribute cash externally. That's true with M&A. It's true with buybacks. It's true with dividends. I would say -- we are -- we do have a team that fields both inbound calls on the M&A side as a potential acquirer. And we also are out trying to meet people in our industry and in adjacent industries, it wouldn't surprise me if in the next 5 years, we did something on the M&A front, but there's a really high bar for those opportunities given our own internal development shops. And it's going to have to be the right opportunity and it's really hard to predict the timing of that. It's going to be idiosyncratic at bike. So look, I think over time, buybacks are probably potentially a way that we could return capital to shareholders, but I think it's just -- it's early enough for us today that we haven't made a commitment there. I mean personally, I'd rather see M&A because it will drive the growth story. I think buybacks and dividends, I think dividends at your stage or not really appropriate. Buybacks, you could argue when the stock is down, but the stock is so volatile. It's like it was 83%, the 43 and how it's back up. It's like you're not a hedge fund. But -- let's talk about M&A, bring us into the Exec Committee meeting when M&A is being discussed. And is there -- what would make sense for you guys to do? Would it be, hey, this company has a product that we think would go really well and a bunch of users and active members that we think we -- is it that? Is it something tangential? Is it -- take us like into that conversation about what you just generically. Sure. I think it's more likely to be something tangential. And I think tangential for us can either mean tangential in terms of a product category or tangential in terms of geographic market. Like if you look at arguably the best M&A deal that we've done to date, it was probably the acquisition of PayBright in Canada, where we had an opportunity to acquire and merge with frankly, the market leader for Binola later in Canada. And it was a business that looked a lot like a firm, they thought really deeply around how they treated the consumer, they were winning an incredible roster of merchants. They had just won Apple's first-party hardware business in Canada, right? So I think it's something like that, where we look at the business and we feel good about how they've treated the consumer, we're not looking for revenue models that are overly dependent on fees, right? That's a huge part of our promise to the consumer. So yes, I think it's a really high bar, but I think ultimately, the corporate development team, they should be expanding our product development efforts, right? They should be able to sort of get to the 2 or 3 things that are maybe below the line for internal development, but still long-term valuable to the overall strategy for the business.

Okay. That's great color. Let's focus the last couple of minutes we have on AI. You guys touched on the shareholder letter, how it's driving productivity. Remind us how it's doing that from the cost side, I guess, which is kind of table stakes at this point, but how is -- how you see it kind of driving growth as well?

Yes. I mean we're still at a point in our development where we just -- the only shortage is capacity to get these ideas built, right, where there's no shortage of ideas. And so we've really pushed the product in engineering teams to utilize AI more in their development cycles. And we actually did an AI tooling week in late January, where we've sort of shut down development at the company for a week to let some of the early sort of leaders in terms of utilizing AI to firm work with the rest of the engineering team to sort of show and share wins. And it's actually been incredible. The uptake of AI usage. We shared in the letter, a table that showed the percentage of our pull request, so sort of the software features that are being shipped and integrated into the code base. I think we're now up to like more than 60% of a week pull requests coming from AI-aided development. So the team has really embraced these tools and it's a meaningful step function change in terms of the throughput of development that we've seen. And then if you double click and go a little deeper for me as a CFO, we're also seeing really nice efficiency like our cost per pole request is actually going down, even though we have brought on some new vendors and there's some new costs in terms of tokens, the rate at which we're using these tokens to ship software means that we're still more efficient in terms of the cost to develop a feature or a piece of software. So I think that's really important. I think that all said, we've done a pretty good job of getting our arms around the spend early, and I think we have the right tracking in place. And so there is, of course, still work to be done around optimizing. I think 1 of the things we're starting to do a bit more around is just making sure that the model or the token type that we're using is rightsized for the job, right? We don't want we don't need the cutting-edge cloud token to sort of check the weather, right? Still like making sure that we're we're using the right model for the right job. That will be another layer of optimization that we'll do. But yes, it's been really awesome to see the acceleration in development internally.

Are you monitoring the cost of tokens -- of course, I mean the answer is yes. But like is it a big piece of the growing piece of the P&L, you're like -- that's a big -- that's getting to be a much bigger number, and that's good because we're offsetting that cost with not hiring as much in certain areas and things like that. So the efficiency of a token versus a human is obvious right?

Yes. And again, I think we've got a really robust road map internally. And so we are continuing to add to the team. We're doing it, I think, in a pretty measured way in terms of head count growth. But right now, we're adding the token costs, and we're growing the size of the team as well, and that's working for us. I think we're really happy with the throughput we're seeing -- but yes, we need to build financial plans that are aware of all these things. I think we've done a good job of that and all of the token costs are factored into our near-term guide and our medium-term guide as well? We could end it on this one. I think at the last month we held, it was right around the time, 1 of your kind of sort of not competitors, major head count reduction. And you said, "Hey, if you look at your gross profit per employee, it would be what they would be post RIF. So you're run really efficiently. It's obviously a really incredible management team.

So do you see AI more as a continuous growth driver as opposed to a cost reduction? Or is it a combination of both? We do. I think right now, it's very much more the former -- like I said we are still continuing to add human beings to help develop more software, and they're using AI tools to get there faster and more efficiently, but we haven't done sort of AI-driven layoffs.

Okay. All right.

I have a bunch more questions, but we're out of time. All right. Thank.

Thanks, Rob. I appreciate it. Thanks, everybody.