That's good. You think we're at a stage where we don't need to give the overview of the company anymore?
I think so. I think so. Fantastic. All right, well, welcome back, everybody. I am very pleased to welcome Eido Gal, CEO of Riskified, here today with us on stage.
Thank you.
Thanks for being here. It's funny, we were just talking about whether this is necessary or not. I don't think it is, but just for the sake of level setting everybody in the room, why don't you give us a very brief overview of, you know, Riskified's model and how the model has kind of evolved over the last stretch?
It's already on the page. We've captured, you know-
It's on the page, yeah.
Riskified helps e-commerce companies manage their risk and understand who their good customers and bad customers are, and give them a different experience based on that. We think we're kind of the largest in the world doing it, and work with over 50 publicly traded companies, including names like Wayfair, Prada, Booking.com, so really across industries and geographies. We do something that's inherent to every e-commerce transaction. We look at that transaction and decide if it's fraudulent or not.
Fantastic. Artificial intelligence has been in the headlines a lot lately. You know, you guys have been in the business of kind of training models for, for years. You know, how, how is your—what is your approach to AI? How, how do you think about AI in the context of your business?
Right. So really, when we started, you know, 10+ years ago, we said, "Hey, we think we can leverage AI to build a smarter fraud solution." And I would say that... And I'm sorry, I might get into a slightly longer answer on this one.
No, that's fine.
Um-
Please.
One, one of the things that we did early on is we understood that one of the most important things for supervised machine learning, and what we do as a form of supervised machine learning, is having the most accurate and unique and proprietary dataset. So for the first initial years, we had large teams that were just looking at transactions and tagging them and saying,
"This is a good transaction, this is a bad transaction." And later on, the machine learning trains on that dataset, and we think that we have the most unique and proprietary dataset to train machine learning models to identify fraud, and that's very unique to us. The second thing is we do a lot around feature engineering, all right?
So something that we shared on the earnings call yesterday is that we've had a lot of success in the ticketing and live event space, okay? So when you do feature engineering and machine learning, it means that, you know, you take an analytical insight, and our insight was that fraud behaves differently between NBA and finals tickets and Broadway show tickets, right?
One is very high risk, one is not. So you would say, "How do I get the machine to understand that this is an NBA ticket and this is kind of a Broadway show ticket?" So it turns out you need to build, like, semantic text analysis to extract the name, and then you need to categorize the name into something, and then you need to assign a risk score to the name based on a histogram.
So we do a lot of feature engineering to help the machine understand, right? So that's another component of our machine learning stack. And the third one, because we really operate on enterprise e-commerce companies, we tend to not have, like, a general machine learning model that just works on hundreds of thousands of merchants, right?
We have an engine that really optimizes features and data per merchant, okay? And that's a way to drive really optimized performance, right? To get that last extra 100-200 basis points of approval rates, to reduce chargeback rates by that incremental 10%-20%, you need to do that. And we've really built an engine that does it at scale. So it used to be that you need to have a handful of data scientists optimizing model, optimizing performance, optimizing features.
So we've just built an engine that does it automatically, and now we're able to have customized models with customized features operating at scale for enterprise e-commerce clients. I told you it was a slightly long one.
No, that's quite interesting. Has the drivers of chargebacks and/or fraud or anything else that you're sort of attempting to solve for, does this evolve constantly, so you have to sort of change your, you know, what your models are sort of solving for, you know, perpetually, or is it somewhat more static than that?
Always.
Always.
Fraud is very dynamic, it's very creative, and it's not just within credit card fraud that it's changing. You find an attack vector, you block it, they try something else, right? But they've also expanded from just using stolen financial information into abusing other areas of the merchant's website, right? So we see a lot around policy abuse.
I don't need to steal your credit card. I'm just gonna call the merchant and say, "I never received my package," and nine times out of ten, if I'm a first-time customer, I'll receive a new package or a refund. That's really, really easy, okay? So there's a lot of abuse and fraud going on in those areas.
A lot of times I would receive my package, the goods, and then I would file a claim, a chargeback, saying, "I, you know, I never received the package or good anyway, or give me a refund." So there's a lot of abuse in fraudulent chargeback claims. So we have our product around, you know, kind of dispute management. So we see creativity and changes within credit card fraud, but we also see the aperture expanding into different forms of abuse.
Talk about your kind of land and expand strategy. You know, how does the customer kind of life cycle unfold? And then I wanna get in a little bit into some of the new products that you've launched. But maybe first, with that first question.
Well, because we work with, again, large, global, multi-billion dollar GMV merchants, they have a large existing stack. Before Riskified, it's several teams, it's a lot of different tools. It's hard for us to come and say, "Yeah, rip and replace everything with Riskified day one." So they tend to prefer to start on a segment on, you know, whether it's a geography, one of their product lines, and then they test the solution.
They say, "Wow, this works really great." We've already built a great relationship with the client, and then we upsell them more volume over time, right? So let's say it's a $10 billion GMV merchant, they start at $1 billion, and over the next several quarters, they would provide us with more and more.... The cross-sell is when we start with, let's say, the Chargeback Guarantee product, and you say, "But you know what?
We're only saving you half the cost structure that you have. You have a lot of loss in, you know, kind of policy and abuse and refund claims," and that's when we cross-sell some of the Policy and Dispute Resolve products.
Mm-hmm. And on the growth algorithm, more generally, so you have a sort of, you know, land and expand strategy. How much of the revenue growth comes from, you know, existing versus new customers? And maybe parse apart the growth algorithm a little bit, too. You know, you've got other drivers in there, you know, new products, new geographies, perhaps, you know, whatever it is.
So, for 2023, I think we shared NDR, right, has been around 105%. And historically, during COVID, pre-COVID, that was 115, 120. So what we've seen coming out of kind of the COVID is we've had various macro weaknesses.
Mm.
It could start at, like, the home and home fitness category, right? Like the Wayfair and Peloton that everyone was renovating and buying, and then it takes time to digest that growth. Now there's some more macro when, you know, kind of certain pockets of luxury and high-end fashion, maybe certain pockets in travel are kind of going down a bit.
So we're definitely seeing a different macro environment than we were seeing before. So most of the growth has been coming just from the new clients, from some of those cross-sells, more so than historically. But obviously, we're hopeful that as we do see some more recovery, and we anticipate that recovery in kinda the quarters ahead, that would, you know, inflict positively on us.
And let me drill down on that last thought a little bit. So the idea is that the commerce kind of backdrop is still unsettled, sort of post-pandemic, and there's some kind of normalization that kind of could be occurring. Maybe talk about that a little bit more. I-
I think that's exactly right.
Right.
When you look at, you know, the growth that some of these categories have had during COVID, then now it's, you know, taking time for them to normalize, so they've stopped dropping, a lot of them, but they still haven't reverted to growth.
Even if the growth is not gonna be at the same level that it was pre-pandemic, we would still anticipate growth, right? Some of the categories, because of the current, just general macro situation, you know, aspirational shoppers within the luxury category, people buying $700 handbags, they've kind of dropped off, and we see that on, you know, the earnings of some of the companies that we work with. $10,000 handbags are still going great, fine, at the same rate, if not better, but that aspirational shopper has been impacted.
So we think that once that comparable period lapses, and we revert to a normalized growth environment within that, because there's nothing structurally wrong with, like, either home furnishings or luxury, that would reflect positively on us as well.
Walk us through your new products. You know, they're not so new necessarily, but walk us through the new products, you know, the ancillary products.
So-
You touched on some of them, but, yeah.
No, yeah, I'll go into more detail. We talked about the chargeback problem.
Yeah.
The chargeback product is when someone files a claim that says, "I did not perform this transaction, someone stole my credit card." The consumer gets refunded, the merchant is liable. Right, that's where we protect the merchants with our Chargeback Guarantee product. Now, we'll screen the transaction, and they won't accept any transaction stemming from a stolen credit card.
Our Policy product, okay, tells them that we'll look at, you know, anytime someone initiates a refund request and says, "You know what? I never received the item," or, "I received the wrong item," should you honor that request? Is that a legitimate request from a legitimate customer, or using the same technology that we've developed in machine learning models and features, is that actually a bad actor that's just trying to, you know, kind of steal that money from you?
If you all look, there was a fairly large New York Magazine article this past week about a very interesting case of policy abuse that was a $multi-million-dollar abuse, and it was interesting. It was very kind of... It was a 22-year-old Miami college student that hacked into the databases of UPS.
So that's the type of stuff that we're up against, okay? And it's very, very pronounced. So some of the merchants that we work with right now, we can pretty easily show them that 10% of their refund requests are fraudulent, and they don't have to honor that.
So when you think about, you know, from a P&L perspective, you know, 20% of orders could end up being return requests or refund requests, and having 10% of that be fraudulent, and that's, you know, money that they were paying out of pocket. Then we tell them, "You don't have to. That's not gonna turn... You know, you're not gonna insult a good customer. Okay? That's not gonna turn out as a chargeback.
That's actually a fraudulent actor. Let me show you how." So that's been great ROI, and while it's still early days for the product, we've been seeing very encouraging traction and proof points there. So definitely feel like there's a lot of opportunity. Dispute Resolve is when there's this entire manual, laborious process. "I got a chargeback. How do I file that chargeback with the bank?
It's, "I need to build a compelling evidence file. I need to parse out a lot of data." We automate that, and we generate the file in a very smart way using AI, and we, you know, kind of reduce operational overhead, automate the process, reduce cost, and increase win rates. So that's been very well received as well.
Without giving away any company secrets on the Policy Protect side, what types of... This is just a minor curiosity, what kinds of indicators, what types of variables are signaling to you that that transaction was, in fact, a fraudulent, you know, false return?
It's usually identifying, you know, there's really a blind spot. This is a customer I don't identify. It's a first-time customer, or for the merchant, it just looks like that. And for us, some of the underlying technology, the clustering, the linking, the network effect, the ability to identify proxies and network devices, helps us understand it's actually not the first-time customer.
It's actually, we've seen him either in other companies doing something similar, or he's trying to hide his traces using these various mechanisms, and we're able to uncover that. So we know, you know, once you're trying to hide your identity so much, it becomes pretty easy to block that.
Mm.
That's one of the kind of the more common elements that we see. On the other hand, sometimes we see that it's actually a nefarious actor that's taken over the account of a great actor.
Mm.
Right? 'Cause fraudsters understand that, you know, if they're able to hack or log in to an existing customer's account, sometimes they actually have a stored credit card, and that's really helpful. But sometimes, you know, the merchant wants to give these good customers such a great experience that they don't have a lot of controls. They enable them to do a lot of different things. So there, it's also we need to make sure that they're not taking over the good accounts.
What about the kind of P&L impact of the new products? You know, how should we think about the impact today versus what you expect as we go forward in terms of the ramp?
Yeah, and they're still early stage. I think what we shared on the Q4 earnings call is that they grew 3x year-over-year, the contribution from newer products. They're still lower, you know, low single digit impact on the overall revenue. I think this Q1, we shared that about 0.5% of the margin improvement was due to the newer products.
When you think about our core product, the Chargeback Guarantee, it has inherent structurally a lower margin than the traditional SaaS product because we pay back for chargebacks and mistakes that we make. The new products don't have that. They have more traditional, you know, kind of SaaS, like 80%, 90% gross margin. So we're seeing an uplift in that area as well.
Talk a little bit about your, your sales and distribution strategy, and I wanna get to how does, how does kind of selling in the new product sort of fit into the existing process. But how do you, you know, how do you organize your, your sales and distribution?
Traditionally, we've been very much an enterprise, kind of direct sales force. You know, the way that merchants interact with us, they like to pilot, they like to provide us with a CSV. Based on that CSV, we provide them, like, a menu of options. Here's the fee and approval rate that we can provide you across these different geographies and industries, and here's how we recommend starting.
So there's some customization and handholding there that fits the enterprise motion. We leverage partners to get kind of more referrals and introductions. You know, we work with the Deloitte Consulting Group so that as they work with the CFOs of some of these companies, they can say, "Hey, you know, you should think about Riskified. They can help transform this part of your business." So that's been great from a lead gen perspective.
In the future, we would hope and plan to get more mid-market and SMB distributions through some of the platforms, whether it's kind of Shopify or other aggregators, but that's not something we're currently very involved with. And the new products have just helped us have more continuous selling, and solve additional problems. Let me give you two concrete examples.
In Q1, this was the first time that we had standalone sales for policy and dispute, right? So we were talking to a merchant, one of them said: Look, I don't feel like I have a chargeback problem. Here's a list of things that I'm trying to solve for this year. Policy is number one, right? Like, fraud, risk and fraud is number whatever, eight, nine, ten.
We were able to sell our platform or integrate our platform while only selling policy to this merchant in Q1. You know, we anticipate and believe we'll be able to cross-sell the chargeback or additional products now that we have an integration and a relationship with that client, and something similar happened with dispute. It's definitely helping, you know, kind of facilitate more conversations.
In that context, help us think of the sort of addressable market for your business. I mean, there is obviously e-commerce, but I have to imagine that there's particular verticals or particular types of merchants who might have more challenging, you know, chargeback-related issues than others. Yeah.
I mean, look, e-commerce is massive. Let's say $6 trillion, but you're right, we shouldn't think of all of that $6 trillion. Maybe we can reduce the SMBs, you know, that we're not a significant portion right now, but you're still left with 70% of that market.
You know, if you want to be conservative, you can remove certain parts of what we call China and India, because they have their own kind of different payment networks and systems there. Maybe some portion of Amazon, because we're not gonna take over that entire line of business, but do work with anyone a tier below them, even, you know, kind of 50-100 billion GMV merchants.
So you're still left with $multi-trillion-dollar opportunity, and within that, you know, whether it's travel, live events, groceries, food deliveries, electronics, fashion, apparel, payments, remittance, I think we hand a hand in all of those different categories.
Mm.
Some categories, like live events, we have, you know, we have a much bigger footprint and kind of own that category. Other categories like food delivery services, you know, we—it's a growing business for us, but we don't own it yet.
Mm.
But to your point, it's not just the risky trend, the risky stuff, right? Because that's why I mentioned, like, food delivery. People used to say, "Hey, is that even, you know, risky, dangerous?" Look, it's not as dangerous as global fashion and digital gift cards, but it's still pretty dangerous, okay? There's still a lot of abuse and fraud happening.
Think about how many times and how easy it is to claim on that app. "I never seen my order, give me a refund," and how instantaneous that is. Just think about how open that is to abuse. Think about the fact that they sell digital gift cards that are basically pre-loaded with cash that you can resell pretty easily on the marketplace.
Think about the fact that some of these delivery services now, they would go and buy groceries and alcohol. They would go to Best Buy on your behalf and purchase stuff. There's a lot of risk, even if it's not like the riskiest stuff. They also need to have good protections.
And you touched on travel, and I know you guys have been a beneficiary of the sort of post-COVID, post-pandemic, you know, travel recovery. You know, talk about that vertical in particular, you know, that's something that's been an interesting kind of growth driver. How has it evolved for you guys, and what are your expectations with travel?
Yeah, I mean, it was a scary rollercoaster, right? Like, a third of the business was travel, and then it disappeared during COVID, and then it rebounded significantly and helped offset some of the softness in the other categories. We definitely still are big believers in that industry, and that it'll continue to grow. We're looking forward to kind of a strong Q3, which is kind of consistent with every year, the travel, kind of summer, summer travel season.
I don't think that it's growing as much, and we are seeing, I would say, an uneven distribution between merchants. So we have some merchants that are experiencing some softness in the travel space, particularly in EMEA, but I don't think it's something very structural or industry-wise. I think it's just a bit more merchant-specific right now, is what we're seeing.
I see. And then what is the geographic overlay to your business? You know, how much of your business takes you to other geographies? What's the sort of non-US opportunity and strategy?
Well, the US has always been our, our kind of biggest, our biggest geography, but that's been decreasing in size and importance over the past few quarters and years. I think it went down, 65-ish%-- 60% even. And we've been seeing a lot of growth, especially in, in APAC and Latin America. Those have been the kind of fastest-growing regions for us.
Specifically APAC, you know, with 30%-40% growth over a number of quarters. Our other Americas actually grew a bit slower this quarter, but that was more an offset because of some luxury fashion in Canada that was growing slower. Latin America continues to grow very quickly for us in similar ranges, and we're seeing a lot of opportunity there as well.
What does it mean for you guys to enter a new market? I mean, is that just basically you get a file of transactions from that market, a merchant in that market, and you have a model that you've trained to basically be effective, or is there any deeper, you know, work that needs to be done to, you know, quote, unquote, "enter" a new country?
Yeah. We always, you know, take the strategic view. If we're entering Brazil, we're entering Japan, are there localized data sources? Is there anything we need to support local languages that could impact our features? You know, obviously, there's an entire go-to-market strategy that goes into it. Luckily enough, we're not entering blind, because we work with so many large global retailers, we already have reviewed hundreds of thousands of transactions from those geographies on behalf of those merchants. Right?
So we have these global brands that are selling into Japan, selling into Brazil, but we already have a good idea of what's going on there, and we use that to gain the first merchants. We show them, "Look, it's not that we don't understand this geography or, you know, these transactions. We've already done this before.
But obviously, once we have, you know, kind of a sufficient footprint or a number of merchants and transactions, that's where we can get into more customized modeling and performance. And there, once you have both a customized modeling, which means you have superior performance, and you have, you know, kind of a good number of proof points, merchants to help you sell, that's where you really see the acceleration. And we feel good about what we've been seeing in those regions over the past-
But there's no necessarily, you know, in-country build required per se. It's not like when you think about a typical-
In-country adaptation.
In-country adaptation.
Yeah.
Okay, fair enough. And this is also maybe something you touched on a little while ago. What about the opportunity to move kind of down market, you know, to small and mid-sized merchants? What will that—what do you think is required in order to sort of... You know, you mentioned on the distribution side, maybe you'd need some force multiplier kind of-
Yeah
... distribution relationships. What else?
Look, whenever we talk about it internally, the first thing we say is, let's remember that this is, you know, kind of in the range of 20%-25% of the market, right? It's not, that's significant. We need to go after that. But from the remaining 75%, we still have so much opportunity to go after, just on the enterprise side, that we are laser-focused on that, right?
Like, we're very proud of the $120 billion in GMV that we did, you know, kind of last year. That's still nothing compared to those trillions, and we think there's a lot of geographical opportunity ahead of us. We think even in the core geographies like the U.S., we still have a lot of growth opportunities. I think the product platform is significant. So those are the core focuses.
As we layer into the SMB, number one is how do we get distribution, right? We're not going to do enterprise or inside sales for SMBs. And how do we adapt the product, which is really, you know, kind of the Rolls-Royce, the Ferrari, whatever type you want in it, and, and simplify it in a way that an SMB can consume it, okay? And also minimizes the support requirements are on, on our end, right?
Because we don't want to be the one teaching everyone the entire world. Right now, we're working with professional teams that understand how to leverage that. So I think those are the product things that we need to solve and the distribution things that we need to solve in order to get into that SMB space.
Purely hypothetically, and I hear you loud and clear, that this is not a necessity.
Mm-hmm.
at this point, because there's a lot of, you know, addressable markets still on the enterprise side. Would it be more to sort of a, like, a platform play, where you would create something that's a little more easy for a smaller merchant to kind of plug into? Is that, is that something more standardized, I guess? Maybe that's a better way to put it than-
The ideal way for it to be is for it to be integrated into part of a stack that provides a similar service for a wide range of SMBs, right? Whether it's a payment stack-
Mm-hmm.
whether it's an e-commerce hosting and performance stack, right? But part of that stack to be powered by Riskified.
Mm-hmm. Talk about the competitive environment. How do you guys—who or what are you competing with? It's gotta be a pretty broad spectrum inside.
It's a broad spectrum, and it can get, like, noisy between what's real competition and not. So I'll try to simplify it. You have legacy solutions. Some of the names are Accertify, Cybersource. They're dedicated solutions to fraud for predominantly enterprise companies. You know, they're losing market share, they're not gaining a lot.
They've stopped developing, you know, kind of a decade ago. But those are kind of the legacy and entrenched incumbents, and they still have most of the market. Okay, you have the payment gateways or the payment facilitators, the PayFacs, and everyone has something tangential to fraud, anyone from a Worldpay to, you know, kind of an Adyen and Stripe. They're not the most sophisticated of solutions, even though these are great companies. We don't see them competing with the customers that we work with.
They don't have meaningful market share in our space, and they have more limited data and integrations than us. They have a more limited platform. Some of the problems that we solve around policy and kind of dispute management, they're not even on their radar. So, you know, we think we don't view them as main competitors today. And on the newer generation side, we have a company—we have companies called Sift, Signifyd, and Forter.
And I think the easiest way to just kinda do a matrix there is, you know, that Forter and Sift are more tools used by internal teams. They don't provide that Chargeback Guarantee very much, and it's, if you're a team that wants to keep managing this process internally and build that, then you might use some of those tools.
The Signifyd is more similar to us in the fact that it's a chargeback guarantee. It's a solution that replaces that and then has a thesis that believes that merchants should not be managing that part of the business. Us, with a better network effect and scale, we can do it better. They tend to focus a bit more on kinda SMB and mid-market, have less geographical and vertical exposure than us, and overall, kind of a smaller private company right now. So I think that's the main competitive set we're up against.
Sounds like more in-house solutions, at the end of the day than-
At the end of the day, the biggest thing we need to convince someone is that you don't need to manage this internally.
Yeah.
Okay, you should trust us to do it better. You, you'll receive a better ROI against approval rates and chargeback rates. Once we get that point across, then things are easier.
Mm-hmm. Changing channels entirely. You, you guys have pulled forward your Adjusted EBITDA profitability with two profitable quarters now. Talk about that path and the sort of variables and levers that we, that we think about, that you think about in terms of driving profitability from here.
Yeah. I think in the past two years, we've been, together with kind of some of the macro shifts and the overall market dynamics, on a pretty significant journey to kind of quicken the path to profitability. And the Q1 performance, that's 1,200 basis points of EBITDA improvement year-over-year. So we've been able to reduce OpEx, improve margins, obviously grow revenues at the same time.
So we're really happy with that trajectory. Obviously feel that 2024 would be, you know, significantly profitable year relative to the negative $8 million that we had in last year. So really excited about that journey and the performance so far, being able to increase the guide there.
The organization now, in terms of staffing and talent and everything, is where you need it to be to execute?
Yeah. About two years ago, we increased expenses pretty significantly just before the IPO, and we said, "Look, we think that there's this platform opportunity ahead of us. We wanna build out this enterprise, go-to-market motion, and, you know, kind of some of the newer geographies." In enterprise, you need to have some boots on the ground, you need to have some support.
And we said, "Hey, you know, we feel we've done that." We have the team in place to go after that platform opportunity on the product side. We have the team in place to go after that global go-to-market opportunity. So we feel confident in our ability to maintain OpEx where it is, definitely in the quarters ahead as we continue to scale revenues and flow that through to the bottom line.
We've shared kind of an updated midterm guide on the preview on the Q4 earnings call of kinda 15%-20% adjusted EBITDA targets for 2026. That was just two months ago, so no changes there. Obviously, still kinda poorly have that in place.
Capital allocation. You guys have been buying back some shares. I think you have another allocation. Talk about your philosophy there and maybe weave in, you know, M&A. Is that something that you contemplate could accelerate your strategy?
Yeah. So we've been very active in the buyback, and I think this quarter or in the previous quarter, we bought back approximately 4% of the shares outstanding, significant amount the quarter below. We were hoping to start the buyback even a bit earlier, but just because of regulations in Israel, we had to obtain some approvals. Look, the way we think about capital allocation is, you know, we're either gonna have the money sit in the bank and gain interest, right?
We can either buy back some stock or we can use it. We have more than we need for operational reasons, or we can, you know, do some M&A that would accelerate the growth or the long-term opportunity of the company. Whatever we've been looking at over the past 18 months has either been just not interesting enough from a product perspective.
The two opportunities that were interesting from a product platform perspective were not priced attractively enough. It didn't make sense. You know, we're trading at, you know, what we believe to be depressed levels, but still, 2.5-3 times gross profit. They were expectations of, you know, 20 times gross profit, so we're not gonna do that deal.
And at this point, you know, it definitely continues to feel that, you know, given the long-term opportunity, as one of the largest shareholders, if not the largest, I would definitely like to increase my holdings by buying back more stock on the open market.
Fantastic. I think we're out of time. What a great conversation. Thank you so much.
All right. Thank you.
Nice to have you. Thank you.