All right, kicking it off today, so I'm Will Nance. I cover payments and fintech here at Goldman. Excited to have Eido Gal from Riskified here. First time joining the conference, so really excited to have you guys. Eido has co-founded the business less than a decade ago and has overseen the company's growth and partnerships with top e-commerce merchants in the industry. So Eido, thanks for joining us.
Great to be here, Will.
You know, I believe it's the first time having you out here. I wanted to maybe level set a bit. You know, the company has been public for about three years now. Where is Riskified in its journey as a public company, and what is your vision for where it can go over time?
So look, I think hopefully if you talk to our customers or kind of investors, do references with our customers, they would say, "Hey, they're... We've tested them, we tested them against our internal teams, against other solutions, and we think they're the most accurate at identifying e-commerce fraud." I think that's the situation today. I would say that over the next eighteen months, I hope and believe we'll be able to expand into solving more significant use cases, or just more use cases, I should say, around policy and dispute management, and be a more integral part of our merchants workflow, workflow.
And probably slightly longer horizon, let's call it thirty-six months. I think we would be doing more and more things like the enhanced auth bank authorization that we talked about the previous earnings, where we're leveraging some of the unique data that we have and leveraging that within the payments ecosystem to increase auth rates or performance for our enterprise e-commerce merchants. So I think that's kind of the trajectory product side that we're headed in.
Got it. That makes sense. So maybe could we unpack a little bit just how you think about the growth algorithm for the company? You have this kind of land and expand sales motion with customers. You've also added new products and verticals, new geographies. So how do you kind of think about what the growth should be for this business longer term?
I'm gonna move a bit like this.
Oh, yeah, that's great.
You're right. So I think, look, oh, by the way, related to the IPO, we incrementally started investing more in the go-to-market-
Right
to go over kinda the global opportunity. And I think we're seeing, you know, thinking about the growth algorithm, obviously, there's adding new merchants to the core Chargeback Guarantee platform, and I think that's been going really well for us. The contribution from that segment this year is gonna be higher than prior years. It's gonna be higher than the previous year, so that's going well, and a lot of that is the expansion, some of the other things we talked about. Another part of the growth algorithm is around the newer products that we discussed, right? The Policy Protect, the Dispute Resolve, and I would say that's growing exponentially. That's great, the three X growth year over year.
Mm-hmm.
Unfortunately, that's still a small base, right? So about 0.5% contribution to the gross margin is what we've shared. I do feel great about kind of the pipeline, the traction. If we can continue to see some of this, you know, percentage growth, it should be meaningful, relatively quickly. The net dollar retention, the other piece of the growth algorithm, which historically, even pre-COVID, has been the 115%-120% range, has been lower over the past two years, around the 105% range. This year, we think it might dip below, slightly below 100% as well. And a lot of that, or predominantly, that's because of the macro pressure that our merchants are seeing in a lot of the discretionary categories, whether it's travel, luxury, the home category.
That's been a headwind recently, and I think retention metrics have been kinda somewhat consistent in the 98% range over the past few years. This year, they might dip slightly below that to the 97% range. I think that's how we kind of imagine the growth algorithm.
Got it. That makes sense. You know, as you, as you've kind of pointed out, the macro environment has been really choppy. The consumer's been dealing with high inflation for a while. There's been a lot of sort of reallocated spend under the surface of what looks like kind of more stable spending patterns. So, you know, what have you seen in the business, and is there anything to call out more recently?
I think we've seen. Let me highlight a few verticals, and I think it's consistent with what these verticals have been saying publicly, and it's kind of well understood, but maybe how that impacts us is what I should explain a bit more, right? So you have the home goods category, relate to furnishings, people are remodeling, you know, high interest rates, less people are moving, less houses, houses being bought. So that's been trending downwards. Fashion, especially kinda the high-end, the aspirational shopper, but also the highest end luxury has, especially on the e-com side, has been trending downwards relative to kind of expectations at the beginning of the year.
And I would say that travel, while it's still strong, we're not seeing the type of explosiveness that we were seeing in previous ca-
Right
... in previous quarters. Where there has been an increase is, you know, kind of low-end shopping, whether that could be, you know, fashion, otherwise electronics, some verticals around, you know, kind of groceries, the food deliveries, those have been seeing a more marked increase.
Got it. That makes sense. And then, I guess, a different way of asking the macro question, I think specific to your business, I mean, oftentimes we ask about trends in consumer spending, but I guess there's also the question of trends in fraud, and I think most payments companies, over the long term, are focused on reducing that. What has been the kind of broad trend over the last couple of years in just the prevalence or the number of attack vectors in the fraud space?
Look, I think fraud is adaptive, and it shifts, and it never really goes away. And, and for us, that's great. Because it's this type of game of cat and mouse, it's not something that's ever, you know, gonna quote, unquote, "solvable." And there are always new vectors. Now, with the rise of, like, ChatGPT, there's a marked increase in chatbot fraud, where, you know, again, the fraudsters are leveraging some of these text capabilities and, you know, forcing or kind of conning agents into giving them credentials to log into accounts, right? So that's a new attack, and every few weeks and months, we see these new patterns of attacks. For us, because of our kind of focus, dedication, network effect, and engineering capabilities, it's easier to detect and to stop that.
To merchants that continue to try and manage this complexity, and internally, it continues to be an increasing challenge. So we think that, you know, the increased sophistication of fraud and fraud-as-a-service are overall, you know, positive for us long term.
Yeah, that makes sense, so I wanted to talk through distribution. The e-commerce marketplace has been become increasingly concentrated over time into larger platforms. How has that impacted your go-to market? And then how do you think about partnerships more broadly in the e-com ecosystem?
I think we've done a good job of really focusing on enterprise e-commerce companies, even the marketplaces there, and, you know, kind of enjoying their growth as we scale with them. One of the food delivery companies we work with, you know, they're expanding into a lot of categories, offering this as a white label service for some other merchants who don't want to build this capability in-house. So in that sense, I think we're well-positioned to grow with these larger platforms. Specifically around the SMB market, which is probably kinda 20% of, As we define SMB, I'm sure everyone defines it slightly differently.
Sure.
The e-commerce marketplace, I think that's where we wanna have much more of a partner-first approach versus kind of a direct sales motion, and that's probably something that, you know, I would say in the next eighteen months or so.
Yeah, makes sense. So maybe you can talk through just pipelines and customer acquisition. What have been the recent trends there? How have pipelines trended? And, you know, when we think about sort of the catalyst for a merchant to look for an outsourced fraud solution, what are kind of the top two or three in, you know, swing factors there?
So look, let me start from the end, right? I think we kinda shared, hey, this year it's gonna be the best performing, you know, contribution-wise of new logos. So pipeline has been healthy to support that and feel good about that. And the reason it's been, I think, so healthy is the combination of more global reach, right? We talked about the newer regions. Maybe we'll mention Japan, Brazil, APAC, LATAM, that have been ramping nicely, and we've also been able to land some of the product platform as new sales from day one, which has also been helpful in kinda contributing to that growth there.
Yeah, and then when you think about... Once you have a customer, that land and expand motion, what does that conversation with the customer look like to get a greater share of their volumes?
I think it's about building a trusted relationship and being able to, you know, to showcase, "Hey, you know, we promised you a 97% approval rate and a twenty basis point fee, and look, you're actually receiving a one percent higher approval rate than we promised you, and fraud and chargebacks are under control. You know what? We noticed that you have these other segments, and your cost profile is twenty-five bits there. You know, we think we can manage it for twenty bits and incrementally increase your sales as well," right? So once you've proven performance and show that, it becomes an easier process to gain more and more volume share.
Yeah. That makes sense. And then pipelines this year, you mentioned, this being one of the biggest years in terms of contribution of new logos. Just how do you see... You know, what is sort of the visibility you have into kind of pipelines and deployments, as we start thinking into next year?
I think we always have this rolling cadence where, you know, and I'm sure this is similar to almost all companies, especially enterprise e-commerce, you know, for the next one to two quarters, you already have kind of the people integrating-
Mm-hmm.
- Definitely for the next quarter.
Yeah.
Probably mid to large-sized companies can always happen in a two-quarter cadence.
Yeah.
Three to four quarters out is probably the pipeline that we're building right now. So I think that's the overall trajectory.
Got it. Makes sense. You've had some kind of one-off fraud events over time, and I'm sure there are also a bunch of events that we never hear about that you catch quickly and nip in the bud. What are the conversations like with your customers in real time as those events happen?
That's true. By the way, one of the fun things about running Riskified is, you know, on top of all the regular public company metrics, we also have fraud events and CTB and everything else to manage internally, so it keeps us on our toes. Look, I think that the. What we try to do is to provide the merchant a sense of control. Okay, the merchant, from our perspective, needs to feel that they're managing the business and are making these decisions.
You don't want them to imagine that there's this black box that is saying that, "Hey, something is happening, and now approval rates are going down." So it's a lot around communication, showcasing it through the tools and visualizations, working with the merchant, their leadership and security teams, to understand the fraud of what is happening, mapping out the options in order to stop this fraud event, and kinda collaborating on a solution, right? Where there's broad alignment around the room, what needs to be done right now in order to stop this.
Yeah. I guess, how do you balance in an event like that kind of the need to act quickly versus the need to also communicate, stay in front of the customer and kinda make sure that they're involved in the decision-making?
They're aligned. If there, again, if there's a massive fraud event, it's in everyone's best interest to make sure that the fraud is stopped so it doesn't propagate and become an even bigger issue.
Mm-hmm.
And if there's an issue with, you know, kinda leaked credentials or accounts being taken over, it's in the customer's best interest as well to understand this and stop it. And they're seeing it from a lot of different vectors, so we always feel that there's a lot of, great collaboration with the client at that point, right? They understand the problem, they wanna solve it. So we haven't had to have too many hard discussions around, you know, what levers to push.
Yeah. I mean, do these end up being kind of catalysts for better conversations, productive conversations on the back end around, you know, doing more, expanding the relationship?
That's a great point. It's probably a bit. Those are usually the best conversations to expanding the scope and saying, "Hey, this is where the fraud vector happened. This is why it was unprotected: because of, you know, account security issues, because of data issues, because of integrity issues, because of a segment that we're not looking at right now that potentially enabled the fraud to come in." So yeah, that's a good point.
Got it. So how about competitive dynamics then? You've... So you have standalone players, like Riskified, you've got merchant acquirers. There are some risk scoring models that are, you know, less on the guarantee side. So, you know, where do you most often run into competitors? And when you don't win a competitive deal, what's the most common reason for why that is?
Look, let me answer it this way. We've shared that over the past few quarters, competitive win rates have been in the range of kind of 80% to above 60%, right? Depending on the quarter. So when we sit around the room as a management team or as a board, you know, obviously, every competitive deal, we wanna make sure that we win and get it up to a 100%. But we feel that the biggest focus we need to have is: how do we make sure that we generate more deals and more pipeline, right, in a way that also diversifies the client base that we have right now, so that we're less impacted by macro, you know, and kind of in future cycles and don't have that type of headwind, and that's really the focus for us, right?
Having said that, you know, when we do lose, it could be related to, "Hey, I'm not sold on this guarantee model. I wanna continue managing my own kinda fiefdom and team, and I'm afraid of letting go of some of the decisioning capabilities." That's probably the number one reason.
Got it. That makes sense. Then just how do you think about the dynamics between sort of acquire-led models versus the standalone models? I know there are multiple types of models out there.
I think that what we've seen so far is that in our segment of the enterprise, people are not looking for, "Hey, let me take a simple package solution from one provider." They tend to run multiple gateways, so they need something that's, you know, usually gateway agnostic, and they tend to choose the best-of-breed solutions and test them out, right? And that's what the category we put ourselves in.
Got it. So switching gears to international, I think in 2022, you added a lot of go-to-market resources, I think you referenced this earlier, to establish some beachheads in new markets. How are you feeling about the international growth runway today?
Look, some of the fastest growth that we've had over the past few quarters has come from those regions, whether it's, you know, kinda that 30-40% in APAC and in LATAM prior few quarters. So we definitely are seeing, you know, kind of positive contribution there. I think we highlighted a very meaningful to us win in Japan to open up that market. And we think it's a market that really referrals are probably more important than others there. So we're very excited about that. I think we're seeing good progress there.
I mean, one theme that we talk about a lot in payments is just sort of the increasing kind of commoditization of the U.S. payments market relative to international, which remains super complex, and I'm wondering if that kind of benefits your business internationally. Like, do you see bigger problems to solve for customers in international markets just because of the inherent complexity?
It's an interesting question. I haven't thought about it like that. I think from our perspective, our platform and decisioning, we're already actually operating in a lot of these regions from a risk perspective, right? So we work with very large merchants, and even if they're headquartered in the U.S. or in Europe, they have business in Japan, they have business in China, they have business in Brazil. So our decisioning science is already set up to, you know, for the nuances of some of these local markets. It's true that everyone has a slightly different competitive makeup, right? Or kinda go-to-market strategy, maybe is the way to phrase it.
Got it. Okay, and then on, on new verticals, I think ticketing and live events, I think, has been particularly notable. I think this has been one of the newer verticals that's been added since the IPO. You know, it was a big success story, and I think it's now roughly a third of the business today. So can you kind of talk about, you know, how you got into that space, and then, you know, the story of how it became, you know, so large in such a short period of time?
Sure. So I think success helps drive future success for us in some of these categories, right? I think the kinda standard story would be we added a mid-tier merchant in that space, had an opportunity to prove the value with one of the largest merchants in that space, and as we added that merchant, we focused in and said, "Hey, how do we make sure that we're the best of breed, best-in-class solution for, you know, kinda the ticketing industry?" And for us, that means, you know, slightly customized modeling with slightly customized features. So for example, we developed an algorithm. You know, these marketplaces, they're selling a lot of different tickets. They're selling NFL tickets, NBA, Broadway shows.
So we developed something fairly complex that could, you know, take the text of the ticket, based on that, do an analysis of what type of event it is. Is it a Broadway show? Is it an NFL ticket? Is it a preseason game? Is it the Super Bowl? And based on the category, assess a risk score. That's something that's, like, hard to do.
Mm-hmm.
Okay? It means that you have better models for everyone in this category and industry, and potentially we can use kind of a similar solution in, like, the fashion industry and, you know, kinda distinguish between Gucci and Uniqlo for if anyone is selling both of those products.
Mm-hmm.
But I think that was the approach, right? So you have some initial traction. You make sure that the product is able to provide a superior performance for that industry. With that, it's easier to go to all the other merchants in that category and have more success there. So I think that's what we saw in the ticketing industry, and hopefully that's obviously something we're trying to replicate in some of the newer categories that we're trying to expand our presence in, whether it's, you know, kind of groceries or food delivery, or also from an international perspective.
Yep, that makes sense. And so I guess, how does that work from the boots-on-the-ground perspective? Is the sales force verticalized? And, you know, how do you think about opportunities for new vertical expansion, such as the groceries and food delivery space?
Sales force is really built based on tiers. We have, you know, kind of our strategic sellers that are selling to clients with over $3 billion in GMV sales, and then it kind of works down from there, right? So really based on tier. You do have some global practices. Airlines is a great example. They're just so unique in how they operate it and how the payments and fraud systems are set up, that it makes sense to have it on a globalized nature. And we, through the compensation plans, is where we try to drive kind of the strategic behavior of, hey, you know, we wanna make sure that we're getting more new products sold because we have a better margin profile, and we think it's gonna help the platform longer term.
Yeah.
We wanna make sure that we get those first few deals in some of these newer categories. That might be a bit more challenging because you don't have that, you know, kinda, massive testimonial background, but that's gonna lead for the future successes there. So that's how we approach it from a go-to-market perspective.
I guess I'm always surprised at just the number of different variations that sort of, like, the risk profile of your customer relationships can take. So, you know, whether that's, you know, risky or versus less risky volume, I'm curious how that differs across verticals, and just how do you see the risk profiles and financial profiles of the different verticals?
I think one of the most important things we do is we have risk-adjusted pricing, right? So to your point, we can work with a... You know, we mentioned the home category in the U.S., someone selling furniture domestically, not super high risk.
Mm-hmm.
Okay, and at the same breath, we can work with someone selling, you know, digital gift cards globally. Okay, this is like cash on a credit card, extremely high risk. The commonality between them is we can outperform the internal team and system, right? So one merchant might have, you know, a 98% approval rate with a 15 basis point chargeback rate, but we can do that for, you know, 10 basis points with a 0.5% higher. The gift card merchant, you know, would have a 70 basis point chargeback rate with an 85% approval rate. So you have these very large variances, but for us, really what's important is to be able to outperform the internal kinda metrics and KPIs in those cases.
Yeah, that makes sense. You've referenced new products several times today. You've been expanding into new and adjacent products, and I think Chargeback Guarantee, I think as you mentioned, is still the bulk of the business today. How would you frame the opportunity and kind of additional services, and what's the trend been in kind of attaching these to kind of existing and new deals?
Policy product, our Policy Protect product, which helps merchants both manage their. It helps them reduce abuse from non-fraud chargebacks. For example, I can call a merchant and say, "Hey, I never received my package and I want a refund," and they would provide me a refund. But I actually did, right? And I just figured out that I don't need to steal a credit card, I can just do that. That's a form of abuse that our Policy Protect product solves, which is not credit card fraud, but is using similar technology. Our Policy Protect product also helps merchants provide different experiences to some of their end consumers.
So one way that our client uses it is they configure our system to set it up so that the top 10% of international shoppers receive free shipping and free returns, but the others don't. Right? So this isn't really abuse, it's more leveraging our kinda network and technology and understanding lifetime value, to set up a different experience for their consumers to optimize their internal kinda margin. Another use case for policy is, for example, they might want to create a first-time customer 10% discount code. But then, you know, you have these returning customers abusing that code, creating fake emails. So that policy product has been going well, and I'm a big believer in that. And I think we're seeing in the zone of 10%-20% of the revenue of the chargeback deal-
Mm-hmm
... when you're talking about kinda same merchant. And that's incrementally higher on the gross margin, because that doesn't have a chargeback component to it.
Yeah.
All right, so that's the, the policy product, and I think that's applicable to, I don't wanna say all, but most of our merchants. Different forms of policy abuse, whether it's a digital merchant, a physical goods merchant. The dispute management, which is us helping take the chargebacks and, and work with the banks to resolve them in favor of our merchant in an automated way with high win rates, that's probably sub-10% incremental revenue increase. It's a lot around automation and workflow, but also seeing good traction there. Overall, they've been growing around 3x year over year, but like we mentioned, still kind of on the smaller side, about 0.5%, incremental uplift to kind of our gross margin.
Feel good about kinda the pipeline, what's ahead from them, both from a product and traction perspective and, you know, I think if we can able to maintain some of these growth rates, they'll become meaningful pretty soon.
I guess, have you seen these be the tipping point in sort of competitive deals where, you know, this is kind of a one main point of differentiation versus your competitors?
I think policy has some wow factor for merchants. It's really a complex system that combines a lot of our machine learning and network, and the ability for the merchant to configure their own policies and manage it on an ongoing basis. So we definitely see excitement around that, and when we kinda reference how we feel are kind of the relatively high win rates, I'm sure that's a component of it.
Yeah, that makes sense. All right, so I'm not the guy to ask a bunch of questions on AI, but I think you're perhaps the most AI-focused name in my coverage, and it's interesting as the investor conversation around AI has shifted to be more focused on return on AI, as opposed to just how much you spend on it. You know, I think you have some of the most quantifiable returns on investment when talking to customers. So can you just talk about just the AI that powers your model, how the customer conversations have been impacted by this explosion in the use of AI?
So you're right, we're an AI company. We were an AI company when we started in 2013. We said, like, the original pitch deck of Riskified, it was machine learning back then, and it's like, we believe that we can leverage machine learning to do smarter fraud detection for e-commerce merchants.
Right.
The type of machine learning or AI that we did is called supervised machine learning. We've actually spent the first few years of the Riskified looking at high-risk transactions that other systems were declining, and merchants would submit them to us and say, "Hey, you know what, Riskified, if you can guarantee this transaction, that's great. You know, my existing team and system thinks it's a fraudulent transaction." We would look at a lot of data and features and say, "Oh, you know what? I found a way to explain why this isn't super risky. It's actually an okay transaction," and approve it. We've actually had an incredibly powerful feedback loop in the form of chargebacks, right?
If a chargeback comes in, you've made a mistake, so you can tag it out to the system, "Whoops, we were wrong." If it doesn't come in, you know that you were right, so that's really important over time. It's created a very, very, powerful effect for us. And that type of supervised learning, where we own the taggings, is what we train our models on, right? So that's very proprietary. And when we talk about the models that we train, I think I gave you the example in the ticketing space, it's, it's stuff that we engineer, we build, right?
We look at a transaction and we say, "You know, how do we as humans, as people, as analysts, understand that this is a good or bad transaction?" And someone says, "Well, you know, I know it's a Super Bowl, you know, ticket, so I know that it's higher risk than a Broadway show." And we say, "Okay, so how do we help the machine understand that?" And that's when we build that feature that encapsulates that information that's fed into the machine learning model. And we have hundreds of those, right? That we've built and engineered, and we deploy different features on different models. So that's all been an incredibly meaningful and significant part of our stack. And we've invested in ten years ago, and we continue to invest in it today.
And the value that we get is in the superior performance for our merchants, right? The better performance for our merchants, you know, the more sticky we are, the more challenging it is to move away from us or go to our competitor. And also what we see is that the better we get, we also take some of that increased value and gain to improve our margins. So we released kind of this annual chart that shows even if CTB starts, you know, slightly higher at different cohort levels, over time, it improves, right? And we see improvements in existing cohorts, sometimes offset by, you know, going into newer geographies, or newer categories where we have slightly less experience.
Yeah, makes sense. And I guess from a customer conversation perspective, have there been instances where customers say, "I need to have an AI strategy? This it makes sense." Has there been kind of an easier sales cycle or just increased kind of intent when you're talking to customers about kind of pulling the trigger?
Maybe people are just more aware of AI, so I don't know, maybe relative to three years ago when you were talking about it, and they would gloss over, and now it's like a bit more interest and hype and, you know, going internally and trying to sell solutions, saying, "No, we don't have a real AI solution. Now, this is an AI solution." It's probably gaining a bit more... It's, it's helping in that sense, I would say.
Yeah. Yeah, that makes sense. I wanted to hit on just the gross margin and model performance. You were touching on it just a second ago. I think, you know, today, I think a little less than half of the revenue goes towards covering the cost of the fraud guarantee and chargebacks. You've seen some improvement since the IPO, even, in form of higher gross margins. But how do you think about the potential for further enhancements, and kind of where gross margins can go over time?
I think we'll see an incremental step-up in gross margins like we have been over the past few quarters, and I think that's gonna be fueled by a combination of improvements to the core machine learning algorithms, you know, somewhat offset by the addition of new clients and whatnot. But also as we continue to grow some of the newer products that have a better gross margin profile, inherently, that's gonna help improve things as well.
Yeah. No, that makes sense. I think you had some pretty significant step-ups in margin over the past year or so. Can you talk about kind of what drove that, maybe as a case study, and you know, how you were able to kind of identify these improvements?
I mean, broadly, it's the exact same categories that we said, right? The combination of the newer products, which we size that kind of 0.5% from that overall improvement-
Yeah
... and the other from the core machine learning. When you're saying from that core machine learning, it's a lot of things, right? It's like, you know, we do dozens of activities throughout a year that would help incrementally. And one example I gave you is around feature engineering and that feature that helps the ticketing industry. That also helps approval rates, but also helps our margins in that industry. We've done a lot of work on automating the build and deployment of, you know, kind of customized machine learning models. One of the unique things about Riskified, you know, we don't, we don't work with hundreds of thousands of SMBs, and then we just need to have a generalized model for an industry. We actually really can tailor a model for our enterprise clients.
You know, that's something that's hard to do at a scale. We work with over 50 publicly traded companies, and you wanna have a model deployment every few months to quarters to take into account new data, changes in the population, things like that, to run in a continuously optimized way. So we've actually built a very complex and great platform to automate a lot of that process. And if it used to take, you know, two to three data scientists would take them one to two weeks to train and deploy this new model. It's now someone is supervising the process, but it's basically, you know, with the click of a button.
Yeah.
So I think all those different things, and again, a lot of things that we do across our tech stack, we focus a lot on. You know, data science is probably one of the largest teams that we have in Riskified, always has been.
Mm-hmm.
Those are the type of things that we do in order to help facilitate both better performance for our clients and a better margin profile for us.
Got it. Yeah. So I think you mentioned after the IPO, you made a big investment in go-to-market. Since then, I think costs have actually been sort of flat to down over time. There's been a big focus on expanding margins. You've put out some targets for twenty twenty-six. The company became free cash flow positive, and you're kind of trending in the right direction on margins. So, you know, can you talk about what you know, how it's been kind of maintaining that flat cost structure internally, and just how do you think about the sustainability and how you kind of continue on that path over the next couple of years?
Look, it hasn't been easy, but we're disciplined. Okay? I think that we're saying in various macro environments, we're gonna drive to meaningful EBITDA improvements. That's been our story for the past eighteen months, and I'm really proud of the progress that we've made. I think in twenty twenty-two, I think it was above a thousand basis points improvement on kinda the bottom line. I think last year, you know, slightly over nine hundred basis points or so. So really meaningful improvements there, and it's a combination of that cost discipline together with better margins, together, you know, obviously some good top line growth or kinda mid-teen's top line growth.
Yeah
... over the previous two years. So that is a focus for us, looking at it holistically and making sure that we're improving the bottom line. I think that's slightly broader. We're trying to learn what's happening over, you know, in this current macro cycle and saying, "Hey, how do we position ourselves as a better company as we come out of this cycle?" Right? Whether it's with a more diversified and broader customer base across more geographies, with a more diversified product platform that has more touch points and, you know, kinda more revenue from different features in that. And I'm just really proud of what we've done there, and I think we've created a lot of leverage in the business to be able to continue to scale and grow.
And hopefully, and I believe that at some point, the things that are now a headwind would also become a tailwind-
Yeah
... and the growth would become also more meaningful, but the bottom line improvements will, you know, continue to outpace that significantly.
Yeah, that makes sense. So we've got about a minute left. Just maybe we can end on capital returns. So the business is now generating positive free cash flow. You know, how much cash do you kind of need to run and invest in the business? And then, how do you think about capital allocation going forward?
I mean, look, we're a profitable company that's generating cash flow. There's no, you know, "Hey, we have so and so capital requirements." Obviously, we would wanna feel comfortable with some amount in the bank. But looking at, you know, kind of our current enterprise value, which I think for the past few weeks has been somewhere between 2-3 times kinda gross profit, we think that the best use of proceeds is to continue to buy back our shares and decrease the overall amount outstanding and available. As, you know, probably the largest shareholder, that's definitely something I believe in. We are interested in looking for interesting technologies that we can, you know, kind of further cross-sell to our, you know, great blue chip brands that have a great relationship with us, and I think that's the overall strategy.
Got it. Makes sense. Look, you know, I think we're just about out of time, so thanks for joining me today. Really great to have you here. I hope to have you back next week... next year.
Next year.
Next week would be too soon.