Business Breakdowns
· Fraser
[00:00:00] S1: At the highest level, we think of Affirm as the American Express, or the payment network, built for the internet and e-commerce. And so if we walk through a transaction, I'll help give a sense of why that is the case. To use your example, a consumer's on a merchant's website — Amazon, Target, whoever it may be — and they want to buy something typically reasonably material. The average purchase price for an Affirm transaction is about $250. So let's say it's a couch. Let's assume this person is not already an Affirm customer, but when they get to the checkout, they see the offers for Affirm financing. So there might be a 0% interest loan over three months, a 20% interest loan over six months, and a 25% interest loan over 12 months. Now, at this point, these are just theoretical offers, because Affirm hasn't actually pulled your credit history yet. And so the consumer decides that these loan terms sound interesting, and so they click on the Affirm button at checkout to set up an account. They provide a few details like their mobile number, name, date of birth, and the last four digits of their Social Security number. Affirm then matches not only the data from the transaction — so what is the merchant, what is the basket size, what are the SKU details — with the data from the credit pull to finalize the terms of those offers. Assuming Affirm is still happy to underwrite the transaction and the consumer is still happy with the finalized terms, the consumer selects the offer that they're happiest with and they check out. The consumer then pays that plan in line with the terms agreed at the point of sale. And that's a very key distinction versus a credit card. The interest rate is fixed, the repayment schedule is fixed, and the balance is paid to zero according to the rundown of the plan. In terms of the actual flow of funds on the merchant and the Affirm side, let's say the purchase price was the $250. Affirm pays the merchant the $250 less whatever fee they agree Affirm receives for the transaction. So let's say it's $5, and that's called the merchant discount rate, or the MDR. That MDR varies widely depending on the merchant and the type of loan, but we can get into the specifics of that later. In this instance, let's say Affirm keeps the $5 and sends the $245 to the merchant.
· Fraser
Affirm then also makes money on the interest payments paid by the consumer when it is an interest-bearing loan. So the merchant's happy — they've potentially secured a transaction from a customer that they otherwise wouldn't have converted. The customer is happy because they were able to purchase this couch on attractive terms. And so collectively, the mechanism makes a lot of sense for all the parties involved. Now, on the funding side, about half the loans Affirm originates get sold to third parties and moved off balance sheet, and about half remain consolidated on balance sheet but still utilizing third-party facilities. So the things we would highlight in that transaction — the customer acquisition is extremely efficient. You can imagine the more merchants that have Affirm checkout, the more customers see those offers, the more customers Affirm acquires. There's very real network effects, very much like an American Express. And the second thing we would highlight, and we're going to probably talk a lot about this, is just how different that credit underwriting process is to a credit card and how different that customer experience is to a credit card as they're paying back the loan. Half of US credit cardholders today have a revolving balance. The average credit card balance in the US is $7,000. There are mechanisms with credit cards — whether it be variable interest rates, compounding balances, late fees — that mean people can end up in really unfortunate scenarios where they're constantly trying to pay down debt over many, many years. That process does not happen with Affirm. Every loan is underwritten on a transaction-by-transaction basis, and they're designed to be paid down to zero. So these are much better for consumers, and the data shows they end up with much better underwriting outcomes.
· S0
[00:00:00] S0: I've actually heard Max say that the credit card has been the best form factor for payment, and he couldn't really imagine that being disrupted, but obviously he's working on doing that. And his background is in cryptography and anti-fraud at PayPal, so he spent a lot of time thinking about quantitative ways to underwrite. Maybe walk us through some of the ways that Affirm is approaching underwriting that traditional credit cards haven't been able to build up.
· Fraser
[00:00:00] S1: The clearest and biggest differentiator is that they're underwriting transaction by transaction. If you imagine a credit card, you're approved for a general-purpose line of credit — maybe it's a few thousand dollars — and you're told, "Good luck, don't spend it all at once," and you're left to figure it out on your own. Affirm is doing every single transaction uniquely, whether that is interest rates are changing, the economy is changing, the type of purchase you're making. They're always feeding that through their models to adjust who they're lending to, how much they're lending to, and what rates they're lending at. And so that's the core structural difference. When you then go a layer deeper, the business has talked a lot about the use of AI and transformer-based underwriting models, particularly at their recent Analyst Day. And so they're basically seeing, every time they release a new underwriting model — which is on a, call it, six-, seven-month cadence — they're getting materially better underwriting outcomes. And this is as they load more and more data into those models to train over time. And so what you're seeing is the core model of underwriting on a transaction basis gives you a step-function better outcome. And then also Affirm's sophistication in the team and the models they're building gives them another step-function change in underwriting outcomes. For context, when we compare credit or underwriting metrics versus businesses like Synchrony, Affirm gets better outcomes, and that's reasonably material when you think Synchrony would talk about having 75% of its customer base being prime, whereas Affirm would talk about, call it, 50% of their customer base being prime. So despite the differences in the customer base, the Affirm process ends up in better underwriting outcomes.
· S0
[00:00:00] S0: It's an interesting segue, because I was going to ask about what the expansion opportunities would be from better underwriting. One of them is you can service more potential customers, whether that's prime or non-prime. So that's one. Are there other things, whether from a demographic perspective or otherwise, that being much better at underwriting opens up — different pools of merchants or consumers that maybe other peers or other card payments wouldn't want to touch?
· Fraser
[00:00:00] S1: It impacts the whole network. Merchants are happy because they're getting more sales — there's more consumers that browse the website that can actually transact. Consumers are happy because where maybe a credit card wasn't approving them for a limit, maybe a $100 loan for a very short repayment schedule can be approved by someone like Affirm. So they're also getting access to credit. And typically what we see when, again, Affirm is talking about their business is that frontier of: we can approve more of these people across more, let's call it, credit score bands and end up still with very, very strong repayment rates and very low delinquency levels, because they're optimizing the dollar amounts and the terms for each individual. I do want to highlight there is a big misconception that all of BNPL and all of these financial technology businesses that have credit aspects to them are generally lending to non-prime customer bases. Like the credit card industry and other traditional lending industries, it's the full spectrum. So yes, you get some players that are predominantly lending to subprime, but then someone like Affirm, as I mentioned — their average customer is actually generally prime, particularly when you look at the dollars they're lending rather than customer count. It's also just not young people. It actually skews perhaps older than you would think — ages 30 plus is the age bracket. And so the product isn't subprime lending to people that otherwise couldn't have access to credit. It's a much broader pool of the population. And I would re-highlight and double down on the statistic that half of US credit cardholders are sitting on revolving balances. And so there's this huge portion of the population that haven't necessarily been served by a great product best suited to them historically, and Affirm is targeting that part of the population. They're not targeting the American Express Platinum cardholder that wants points and airport lounges.