Join Motley Fool contributor Brian Stoffel as he breaks down the business model, Riskified's (NYSE:RSKD) proprietary AI system, and its risks.

To catch full episodes of all The Motley Fool's free podcasts, check out our podcast center. To get started investing, check out our quick-start guide to investing in stocks. A full transcript follows the video.

This video was recorded on Oct. 6, 2021.

Emily Flippen: Welcome to Industry Focus. Today is Wednesday, October 6th, and I'm your host, Emily Flippen. Today, I am joined by the Fool's Brian Stoffel to talk about Riskified. It's a newly public business that's looking to solve online payment fraud. Brian, thanks so much for joining. I think this is your first time on Industry Focus ever, despite the fact that you are a prominent figure here at the Fool, so I'm really excited.

Brian Stoffel: Well, thank you, Emily. It's exciting to be on it, I enjoy the show I listen to it all the time.

Emily Flippen: Well, today's show is especially fun, selfishly because you mentioned that you want to talk about this business in today's episode. I have to say, I mentioned this is an online payments fraud business. It may not sound like the type of business that I would inherently be skeptical about. But maybe 10 minutes into my research for the show, I was foaming at the mouth. I was ready to go as like this is another Lemonade. But as I got deeper into this business, it grew on me more and more and more. As I sit here today, I actually think I'm really excited about this company. I am excited that you brought it to me and that we have a chance to talk about it.

Brian Stoffel: It's an interesting company, thank you. I will also say that your initial reaction, if you look at the stock what it's done recently, was probably justified. It's down quite a bit from its all-time highs. But we see this all the time with super small recent IPO companies. I don't think it's anything investors should get worried about. But I do think it's worth digging into because this is very unique. There's a lot to wrap your mind around.

Emily Flippen: Well, with that, tell us a little bit about how this business got started. I love a good origin story.

Brian Stoffel: Sure. The company's Co-Founder and CEO, Eido Gal, he worked for a company that got acquired by PayPal. While he was at PayPal, he worked in their fraud division. While he was there, he realized what a big deal chargebacks were. Now, a chargeback is like where you send the product to someone and they say they didn't get it and who knows, maybe got lost in the mail, nobody knows. But the bottom line is, is that you lose the product and you lose your money. It's a bad deal. But if you're the consumer than most of us are who pays for things regularly, you might ask yourself, how often does that happen? It can't be that big of a deal. I know that there's some people that steal those Amazon packages that are on your front doorstep, but come on. My question to you and I'm putting you on the spot here, Emily, the economist tried to estimate how much is lost in sales, not only from chargebacks, but also from vendors saying, "No, I'm not going to sell you this product because based on what we know, we think it's going to lead to a chargeback." Percentage-wise, do you have any idea?

Emily Flippen: Percentage-wise? I would think it have to be less than 10 percent. But I'd say that ignorantly thinking, based off your question, it has to be bigger, but initially thinking, yeah, less than 10 percent of sales must be rejected, right?

Brian Stoffel: You're close. It's between 10 and 15 percent, which is a sizable chunk if you're saying, "I'm denying these," and they couldn't get approved. He worked at PayPal. He saw this was a problem. He got together with his co-founder, who is still on board, and they started building basically an AI machine that could figure out based on a ton of variables that I'm not even going to pretend to understand, whether or not someone who comes to a merchant is going to lead to a chargeback, or they're not going to lead to a chargeback. I can get into the results more if you want, but that's the basic idea. This is especially important, I will also say for international vendors because I'm in Wisconsin. If I run a local business, I might have a pretty good idea of who my customers are. But if I'm running my business in Wisconsin and I'm selling to the other side of the world, I have no idea what to expect. This is especially important to those types of e-commerce vendors.

Emily Flippen: I definitely wanted you to get into it because when I start to read into this, we're getting into this black box that exists anytime a business starts talking about AI. I always roll my eyes and this is where my skepticism came in. I felt the same way about Lemonade, and admittedly haven't looked at Lemonade in a while. But when I did my initial research, I remember feeling like they're using a lot of buzz words, they're saying, "We have this proprietary system. We can tell who is going to be a good person to bring into our network versus somebody who is maybe going to commit fraud or be riskier person so we can make a better decision about what their premium should be." I heard Riskified saying a similar thing throughout their perspectives, essentially saying, "Oh, we have a good idea about when a customer comes to a company, if they're going to result in a chargeback or not. It's all based on our special technology that just gets better with time." Anytime someone says we have something special, its proprietary, but it's in this black box, my red flag start getting thrown a little bit.

Brian Stoffel: No, I get it. It's them saying, "Trust us, don't worry about it, just trust us." I had the same reaction. It's funny you say that, Emily, because I had the same reaction when I first looked at Upstart in January and I said, "Come on, how can what you have to be that special when there are these banks that have all these people." Obviously, I've been fairly wrong about that. Upstart has done very well. I think their revenue in the second quarter was up 1000 percent. Actually when it comes to AI, the way that we view it, it is very difficult. But Tom Gardner made the point that Google, their main moat is basically AI, it's their algorithms. So it can work, the problem is it's not as obvious to us the same way that Walmart has purchasing power. Anyone who spends a little bit of time can figure that out. That's a moat. Riskified has AI. Really? Show me. You're showing it to me. I don't understand it because I'm not a data scientist. But let's get into the results because I think that's where we can start to see evidence of a moat start to prove itself out. I'm going to use one of their biggest customers as an example, Wayfair, where a lot of people get their furniture and home furnishings. I get a lot of the stuff from their. Wayfair uses Riskified. If Wayfair uses Riskified, it helps them in two ways. I think it's important to understand both. One, it cuts down on the expenses that Riskified was using to prevent fraud. It cuts down on their operating expenses and then it increases their revenue. Because Riskified is approving a certain number of transactions every month that normally would not have been approved so you're cutting costs and you're gaining revenue. Riskified went back and looked at their 10 largest customers and they do have concentration risks that are worth mentioning. But they looked at their 10 largest customers. What they found was, was that expenses related to fraud at those companies were cut 39 percent. That's a pretty good deal. Revenue increased eight percent on average. It's not only cutting costs, but it's increasing revenue. Then here's the kicker, Emily. If Riskified approves the transaction that later leads to a chargeback, Riskified will cover it. Even if there wasn't a 39 percent savings and an eight percent revenue growth, even if it was just one percent, one percent, it's still a no brainer because it costs almost nothing to use. Those 39 percent savings are net of the fees that are paid to Riskified. It's almost silly not to use it.

Emily Flippen: I will say it sounds compelling. But when you start telling me that they are offering what is essentially a guarantee to the purchaser, that this transaction will not result in a chargeback and if it does result in a chargeback will cover it. That to me sounds like an insurance company. I have a hard time understanding how this isn't considered an insurance company.

Brian Stoffel: That's actually a really good point. I kick myself for not having looked at that before because in a way, it is. The funny part is, it's the hook or it's the customer acquisition aspect like they use that chargeback guarantee is almost like, "Hey, you got to use this. You've got no excuse for not using us." Then it goes from being AI just an AI product almost like a SaaS product, although it's not because they get paid a portion of their gross merchandise volume, but it almost goes from that to being an insurance company. But if we look at those numbers too, there's a couple of things to point out. One, Riskified doesn't pay cash in most cases for the chargeback. What do they do? They offer credits for that retailer in the future. If they were guaranteed to say pay risk by the $1,000 and they had a $500 chargeback, they'll still get that $1,000 of Riskified services, but they only pay 500. Riskified is not paying that out of their coffers. They're losing revenue, but they're not paying it out of their coffers. That's one. But two, it's really important to look at the trends with the chargebacks. In 2019, they used about 42 percent of their billings on the chargebacks. That's a lot. But part of that's because they're only collecting about, by my calculations, about a quarter literally one quarter, 25 cents on every $100 spent across all their merchants. That's not much. Using 42 percent of your billings on that makes sense. But that fell from 42 percent to 37 percent in 2020. They didn't reveal what that figure was in the most recent quarter. But here's what they did say. That their cost of revenue is almost entirely their chargebacks. While revenue grew 44 percent, their cost of revenue grew just 27 percent. From that, to me, we can assume that chargeback percentage is continuing to fall. To me, that is the clear sign we have available to us of saying this does look like a moat. The AI is improving, they're approving more transactions and they're spending less on chargebacks.

Emily Flippen: I was excited. When I started to see these numbers come out, I was like, oh, I'm going to take Riskified and I'm going to run it through the metrics we use to evaluate any other insurance company and show you just how bad this business is. But when you look at that percentage of billings that's lost chargebacks, the closest metric that you can compare to in the insurance industry is something called the loss ratio. Investors who are familiar with the insurance industry know that decent loss ratios are usually in the area of 40-60 percent. That 37 percent I was somewhat disappointed, but I guess now excited to say that's lower than what you would normally expect from a decent insurance business. To the extent that you want to evaluate Riskified as an insurer, they're actually performing really well. In fact, well, they obviously don't breakout metrics like the combined ratio, which is a key performance indicator for insurance companies. If we take out their stock-based compensation and seeing that the majority of these operating costs are billings, their combined ratio is probably less than 100 percent. Meaning they're not actually losing any money on the customers that they're bringing in. Again, this is unlike when I looked at Lemonade, both of those metrics were horribly negative or greater than 100 percent in the case of combined ratio. But either way, when you look at the financial performance of this business, even taking that added level of skepticism to say, I'm going to treat this like an insurance company pretend like the software aspect isn't really there, it's still pretty impressive. I have to say, my job is a little on the floor right now.

Brian Stoffel: Yeah, and here's the thing. I think it can make it below 30 percent because they've been doing this for longer than anybody else. That head start we learned from Upstart, it does matter. It makes a difference that they've got this backlog of data that they're feeding to their AI. I believe it could fall below 30 percent. But the other thing to keep in mind, because it's going to be a little bit bumpy, is that they've said as they enter new verticals, which basically just means as they enter new industries, their chargeback ratios once they get started are usually much higher. That makes perfect sense because the AI is just starting to learn what leads to a chargeback and what doesn't. But they fall over time and there's smaller percentage of overall revenue. Don't be surprised if there is some lumpiness, but I would not be surprised if five years from now, that ratio was below 30 percent. Then Emily, here's the other thing that is just if you are a software as a service provider, super exciting. I know I sound like a bear or a bull, a huge bull, stocks down big.

Emily Flippen: I will get to their risks.

Brian Stoffel: Yeah. But their dollar-based net retention was solid when they came public a couple of months ago. I think it was in the range of about 120, maybe 117. But travel and ticketing are two of their largest verticals. Now, travel and ticketing and because they are paid as a percentage of all sales for their merchants were not existed because you couldn't go to concerts and people weren't traveling on airplanes. If you didn't include travel and ticketing, their dollar-based net retention was 158 percent when they came public over the trailing year. We almost never see numbers like that. Like that's better than CrowdStrike. It's because it's a usage-based model. It's based on the amount of gross merchandise volumes that its merchants sell.

Emily Flippen: Let's get into that because there was an option when Riskified thought about setting up its business, they could go to subscription route where you just pay a flat monthly fee to have access to their services. Or the usage-based model routes. There's so much back and forth across different industries that which is better and I think Riskified is proving out that in their case, at least that percentage of gross merchandise value is great because they succeed when the merchants succeeds. They make money by charging that percent of GM-V. But the fee does depend on the customer base. It's not a matter of we take an X percent fee up every purchase, it's perceived how risky that company is. If you're talking about a company that mainly does sales into markets that their system, their AI system determines have the highest amounts of fraud then they is going to be a higher percentage fee. They actually use metrics from property and casualty insurers to actually determine that discount rate that they're going to be using to figure out what premium they need to be paid. One thing I will mention is that they actually only get paid for purchases that are approved by their system. They do have an incentive to approve as may charged as possible, but at the same time because they reimbursed for any fraud that happens. But the business was covered against, they don't want to approve anything that results in a chargeback. It actually works for both Riskified and the merchant.

Brian Stoffel: Yeah. It's a beautiful job of aligning incentives. On that point, Emily, I think what's important to say is that a company like Wayfair, we don't know what the agreement with Wayfair is. But Wayfair might only kick a certain percentage of their potential purchases to them. It doesn't mean that Riskified is reviewing 100 percent, they might, but they have agreements to specify which ones Riskified will review and which ones they won't.

Emily Flippen: One of the things that makes it really appealing, but I imagine the salespeople at Riskified must love this aspect of the business. But as an investor does make me a little bit worried and it's something that I put down as a risk, is a fact they actually guarantee a certain amount of approvals. It's great because it's salespeople can go to the customers and said, "Hey, if you're only approving say 92 percent of purchases that go through your platform, we will guarantee that we approve at least 95." That's great because from the business angle, again, it goes back to that average revenue. Their revenue is going to increase and if it doesn't, then they get that chargeback guarantee. But it does mean that the pressure is on. They're essentially looking at the current systems that the customer has, whatever they have for fraud and they're saying, we can do better than that. In fact, we know we can do better than that so we're going to bet money on it. If they can't do better than what they expect than that 92 percent or whatever percent it was, then they end up losing money. I don't see that happening right now, but it's I guess a risk to keep in mind.

Brian Stoffel: The thing to remember with that is that it is a risk to keep in mind at the same time Riskified will be hurt and its shareholders hurt, and its co-founders are the biggest shareholders, individual shareholders. We will all be hurt if they say we will approve more than they should be saying. What we are placing faith in the management team to do is to not go into some company that has a risky business that it's selling to and saying, we're going to prove 100 percent because if they do that, that's really bad news. We're trusting that when they reup these contracts every year, they are using the data that they have to make a smart amount of percentages that they say that they will approve.

Emily Flippen: Another thing to keep in mind in terms of how that revenue is recognized because it gets confusing while they do have a few other product lines, virtually all of their revenue right now comes from this fraud management business. The revenue recognition is weird there because while they're getting a percentage at this GMV, they're also trying to go to that insurance process estimating how much of this is going to have to be paid out in the future. Again, not in cash, but actually just decreases from future billings, that credit that is being handed to their customers. Portion hits historically been around, 50-60 percent of their revenue is recognized in the month that it occurs so when that GMV flows with the merchants, they recognize that revenue that month, but the rest, 40 or 50 percent is recognized over a period of six months as consideration for that guarantee. There is a little bit of lag time you can say in their revenue, assuming that their models hold up in that chargebacks don't end up larger than what management expects. Their backlog is a pretty interesting predictor of what future revenue is going to be.

Brian Stoffel: Yeah. If you keep that in mind and you track that over time, you get a really good idea for what's coming down the pipeline in the quarters to come as well.

Emily Flippen: Well, I feel like we've been really, I guess bullish here. I justify this is a really interesting business. I haven't seen anything like it. I like the fact that I can't consciously think of any single company that is competing with the Riskified, at least solely on their core business right now. But there are certainly risks here. Brian, I think you're a shareholder yourself. What risk standout to you?

Brian Stoffel: Sure. There's a whole bunch risk. The number one risk that every investor should and I'm going back to the basics here is that I don't understand the business. Here I am on the Motley Fool Industry Focus podcast, I might not understand the business, but look, everybody needs to have the humility of being able to say that that might be the case. But beyond that, I would say that one, this is pointed out by someone at Twitter, if hackers somehow figure out a way through or way to exploit the AI and make a whole bunch of purchases that end up being fraudulent and Riskified has to pay for that or give huge considerations for future revenue, that could be like a black swan that could really sideline the company. I don't know the intricacies. Management's probably the only one who does have how they're making sure to avoid that but it's something that's at least worth saying. There's always the potential for people bringing it in-house because once they see how well this works, they might want to do that. Then there is competition actually, Emily. If you go, for instance on Shopify's third-party marketplace where you can buy things. There's a lot of people that offer the same thing. Here's the difference and I'm not saying it makes it good or bad is most of those have subscription-based products. Riskified used to list their tool on Shopify's marketplace. They don't anymore because they decided to go this usage-based model and only focus on merchants that have over $75 million annually in spending. It was a business model decision that they made but it doesn't mean that there aren't others out there who are trying to do the same thing.

Emily Flippen: That's interesting, I think for smaller businesses too, a lot of payment providers offer some level of fraud prevention. A lot of companies, admittedly, probably companies that are less than 75 million in sales of smaller businesses but what they'll choose to do is just either self-insure or trust that their payment providers is going to flag any issues related to fraudulent sales. But what stands out to me I think is the landscape is certainly interesting. I worry a bit about just how many opportunities there are there. I really do think that they are doing well with their current customers. You mentioned, Wayfair, but other customers include Macy's, Wish, Foot Locker, Prada, Skullcandy. All on that sweet spot in terms of not too large to bring it in-house, but not small enough that they're not going to be going based on the usage-based model. I worry a little bit that when customers get too big, they'll just choose to self-insuring. I think Amazon is a good example of that. I don't know how much losses Amazon just retains as a result of porch pirates or fraudulent purchases but it's enough that Amazon probably isn't going to be a Riskified customer, they's probably just choose to retain that risks themselves. If they are able to pull on a big customer like that though, my mind would be blown.

Brian Stoffel: That's a really good point, Emily, because they do need to pull in big customers. Eventually, at some point in time, they're going to reach a point where their GMV growth is pretty much going to match whatever the economy growth is because if they're not adding new customers then that's where it's going to be. I still think that there's probably more of an opportunity out there for them. Importantly, as those chargeback loss ratios go down, even if they keep the same amount of GMV, their profit will go up because they're saving less for those charge-backs. But it's still worth considering they do need to land medium to big sized customers continually moving forward for it to fulfill its potential.

Emily Flippen: The last risk that I'll throw out there and maybe an overblowing the risk of this. It just stood out to me when I read it through so I might as well mention is that at some point, maybe this is legally considered an insurance business. I'm not a lawyer, so I'm not positive what that definition is. You're probably right that it has something to do with the lack of consideration in cash and comparison to just future credits that differentiates this business from an insurer. But insurance companies have to go through a lot of hoops and a lot of regulations regarding their business practices, especially around things like liquidity and metrics. If at some point, whether it be in the United States or elsewhere, while the majority of the revenue does come from the US they are, I believe incorporated in Israel. Depending on which jurisdictions that are operating in, possibly someone could interpret this business as an insurance company, which then may change the regulatory framework in which it works.

Brian Stoffel: It might. I don't know the laws are in Israel. I do know that there is some consideration for changing laws in Europe that are coming up as well that management has talked about on the conference calls but I would argue, and maybe this is a silly argument for me to make. I always use Southwest Airlines but I don't know about you, Emily. Do you know why I use Southwest Airlines other than I love it but do you know the other reason.

Emily Flippen: Why?

Brian Stoffel: Because I can change my reservation to any time, which if you think about it, isn't that the same thing? Southwest compensates me if I have to change. Its like trip insurance. If I changed my mind, I'm not losing anything, I don't get the cash back, but I have to use it with Southwest within a year. What Riskified is doing is not all that different than what Southwest is offering. Southwest is saying you can change your mind, Riskified is saying you won't get a chargeback that if you do, here's what we're going to do for you. I could see an argument being made that it's not quite an insurance company.

Emily Flippen: I like that argument I hadn't thought about it that way. Before we wrap up here, I know that you have a framework for which you analyze companies and you do it a lot. If people aren't familiar with your Twitter account, with your YouTube account, I believe as well you do stuff with Brian Feroldi. It's a wonderful resource, so I encourage everybody to listen to it. But you also have something that you called the Anti-Fragile Score. What can you tell us about that and how Riskified stacks out?

Brian Stoffel: Sure. The Anti- Fragile Score is based on the writings of Nassim Nicholas Taleb who was a former trader and now just does whatever he wants. He's a very interesting person. We have an interview between him and I in our Replay Hub so go there and type his name in if you want to see it. But he said the world breaks down into three things. Things that are fragile like a glass, like a fall in the ground. Things that are robust like a piece of iron it's not going to change when it's exposed to stress and then things that are anti-fragile. Things that will get stronger when exposed to chaos and stress up to a point. Our muscles would be a great example. Our muscles get stronger actually because they tear. As long as we give them enough time to rest and recover, it will grow back stronger. What I wanted to do was find companies that do the same thing. Obviously, every company will break if it's put under extreme stress, but I want to find ones that will get stronger under normal amounts of stress. I look at things like their mission, their moat, the optionality that the company has, their cash balance because that plays a huge role in the anti-fragility If you've got lots of cash and free cash flow during an economic crisis, you can get way stronger because of it, because your competitors don't have that. Then I look at things like skin in the game, like are the founders involved? How much do they own? Do employees like working there? Those are the things. Then there is one other risks which is important to point out because Riskified has it which is concentration risk. When you are counting on a few customers for a lot of business, there can be one decision made in one board room on one day that overnight changes your business. When Riskified did this, I look for companies that score between a seven and a 12 on my framework. You can visit the YouTube channel to see this. It's Brian Feroldi YT because I do all this with Brian Feroldi, YT is for YouTube. But it scored an eight and a half, so it's in that investable range, 12 and above is the best but this company, there's too many unknowns to give it that. The big points off came from that concentration risk.

Emily Flippen: I love that framework and I haven't had the chance to go back and listen to the analysis that you've done over Riskified on your YouTube channel with Brian Feroldi but investors who are looking for more information should definitely go, check that out to follow Brian on Twitter. Listen to us here on Motley Fool Live via Motley Fool subscriber. But Brian, in the meantime, thank you so much for joining. This is really enlightening, we'll have to do it again.

Brian Stoffel: Thanks for having me, Emily. I really enjoyed it.

Emily Flippen: Investors and listeners, that does it for this episode of Industry Focus. If you have any questions or just want to reach out to say, hey, don't be afraid to shoot us an email at industryfocus@fool.com, or tweet us at @MFIndustryFocus. As always, people on the program may own companies discussed on the show and the Motley Fool may have formal recommendations for or against any stocks mentioned, so don't buy or sell anything based solely on what you hear. Thanks to Tim Sparks for his work behind the screen today. For Brian Stoffel, I'm Emily Flippen. Thanks for listening and Fool on!

This article represents the opinion of the writer, who may disagree with the “official” recommendation position of a Motley Fool premium advisory service. We’re motley! Questioning an investing thesis -- even one of our own -- helps us all think critically about investing and make decisions that help us become smarter, happier, and richer.