CEO of AdTheorent, Jim Lawson. Jim is the CEO of AdTheorent, which is a programmatic digital advertising leader using advanced machine learning technology to deliver measurable value for advertisers. Jim has been a part of AdTheorent since the beginning, in 2012, serving in different executive capacities, including managing partner and general counsel, and chief operating officer. Before turning his focus to technology ventures and emerging businesses, Jim was a partner in the Washington, D.C. office of McDermott Will & Emery in D.C., where he represented Fortune 100 companies and growing businesses in both litigation and corporate counseling capacities. Okay, so Jim, let's talk about leadership first. So as you know, we keep track of, like, asset efficiency, revenue per employee, and you guys have really great numbers and great trends. You have about, round numbers, 300 employees. When you think about culture, how do you think about the culture you're trying to create for employees, which helps you attract and retain the best employees to generate more revenue per employee? Yeah, that's something that has been central to AdTheorent since 2012, when we started our business. We've been - not that awards are necessarily kind of really what you seek, but they are interesting metrics. We've been a best place to work in New York on Crain's list for 10 years. I think it's because of the fact that we have treated our employees like adults. We treat them like professionals. We give them incentives. I think most of the kind of best employees that you get in business are people who want to be there and have a reason to believe in the cause, and they're invested in the outcome, and the volunteers are the ones that provide the most commitment, and by that I mean, you can tell someone, for example, to be in the office, but it's much better if they just want to be in the office. It's much better if somebody. You can tell someone to do something, but if they know that they're gonna make a bonus or they're gonna make certain additional incentives based on company success, and you can provide a career path for people so that they can join your organization and stay in your organization for a long time in a competitive environment. Our leadership team has been with AdTheorent for 7.5 years on average. Our general kind of tenure is almost four years. In 2024, in ad tech, with a lot of young employees, that's kind of rare. Our kind of involuntary attrition every year is about 9%, which is much lower than, I think, what you see in many orgs, and it's because the people believe in what they're doing, and we give them kind of a franchise, you know, franchisee kind of mentality in terms of building a career, building a financial upside for themselves within the context of our broader mission. Okay, well, let's pull that into future of work. So what's your work from home policy right now? We don't mandate that people come into the office, and it's always in our back pocket. I firmly believe that you're never gonna be successful in a world where everybody's remote. I just don't think it's possible. I think the collaboration, especially when you're building new technology and you know you need to be with clients, you need to be in the real world, young people need to learn. I firmly believe that. I also recognize that when you tell people to do something, it's much less impactful than making them want to do something. We have a beautiful office in Soho with great balconies, and we have happy hours. It's a great way to get people into the office. We've had about - we have a 45% in-office rate, and that includes, you know, holidays, Mondays and Fridays. So, you know, we can always revisit the kind of decisioning around, are people required to be in or not, but I think at this point, like, we have a lot of people that really care and really want to succeed. So it's been working for us. You have no required in-office days, and 45% of the staff comes in five days a week, three days a week, once a week? The kind of like in-office rate is 45%. So of the employees. Yeah. Half of the time, our office is full. Half of the time. The guys could be coming in two and a half days a week out of the 5, and that would get you to the 45%? Yes. Okay, and I assume it's the same people. But that includes, again, Fridays, Mondays, and holidays. Okay. And vacation. So... Have people moved away? I assume it's the same people that come in and not... Like, it's not that everybody comes in at least once a week, it's that 50% of the employees live nearby, and they're the ones that come in. The people that we really care about coming in are client-facing... Okay ... strategy, product. If our finance team is more comfortable working from home, it's not something I lose a lot of sleep about. I mean, I think you need to have managers that manage. If you have managers that hold their people accountable and that things get done on time, I think in certain groups, I mean, you know, we have certain data scientists that can do work remotely a couple days a week, and it's fine. If our sales, strategy, product, and tech people are remote, then they involve. So I think, like, that's kind of where we're focusing our effort rather than- I think last year, maybe it was the year before. I'm sort of mixing up my years. You were talking about these software engineers from the same, but they would move, you know, lunches, cars to work, crazy. Has that now changed so that people are coming into the office because they're more afraid for their jobs? Or are the best people still trying to regularly take it in terms of software? I mean, I think it's been a very competitive market for those types of jobs. I think our job is to make it worth it for them to be here. Like, you know, we have to be able to show that working at AdTheorent is going to be very good for their career. They're gonna be able to do exciting work. They're gonna be able to succeed. They're gonna be able to make a lot of money. And at the end of the day, that's kinda what we focus on. Interesting. Okay, let's move to AdTheorent specifically. When you think about, we're gonna sit here a year from now- Yeah. What do you want to achieve between today and a year from now? What are your key goals that I'm gonna write down and hold you to? So when we're on this stage a year from now, I'm gonna say, okay, Jim, last year, your goals were... Yeah. ... what are you willing to let me know? Well, we're back to growth mode. Yep. You know, I think we wanna continue to have strong, consistent growth. The work that we've done to build new innovative products that I think will sustain us through different macro, I think it have been time well spent and money well spent. So growth, obviously, we wanna be able to consistently grow strong quarter-over-quarter performance. Okay. So that's something that, you know, we're certainly signing up for. We saw- I can go do that. Yeah, we saw a return to growth in the Q3. We signaled that we anticipated that growth continuing. I can't talk about the Q4, but I can't wait to talk about it, you know, at the end of February or early March. I think that, the health product that we brought to market is, is an incredibly valuable addition to the programmatic marketplace. I think, our algorithm audiences. So let me just unpack that for a second 'cause ad tech, sometimes, you know, you hear a lot of, like, buzzwords. Most DSPs target ads based on activating audiences. Audiences are lists of IDs. That's what all the big platforms are targeting. We have a system that is an impression scoring system rather than an ID targeting system. We, this year, rolled out two core products that will reshape our business, and in our view, the industry, and they are our health audience builder solution and our general audience builder solution. And what these are, are tools that allow marketers to access primary data sets, such as health claims, pharmacy claims, hospital claims, electronic medical records is one example in health. And to take that data and build audience quality algorithms that allow, in this case, a health marketer, to build a targetable segment using an algorithm rather than a list of IDs. In other words, an algorithm asking the question: of the billions of ad impressions that are coming into our system, is this impression in front of a person who has you know, arthritis of the hip? And how would you know that? Well, if you have a lot of data about... a lot of health data, and then you can look at the different signals and the attributes from that data, and then you can correlate those signals, those high-indexing signals of people who have that condition, with media exposure data, and you can kind of understand the, those patterns, then the algorithms can predict that an impression that comes into a programmatic, you know, DSP like us, can predict that that impression is valuable, and that that impression is likely someone who has this condition. As opposed to licensing a list of IDs from a health audience kind of creation system of, "These IDs all have arthritis." Well, where'd that come from? And to make that HIPAA compliant, they had to take 90%, a certain, you know, number of IDs and put them into the list that don't have the condition to make it privacy-compliant. It's just not scientific. So what we built is a way to take primary data, allow the signals to kinda tell the story, and then have algorithms predict media that will correspond to those people. One of the things you've always talked about at AdTheorent is, since it's algorithmically based and you're scoring impressions, which means you're figuring out what should be targeted, you're not using cookies. Right. Cookies is finally gonna get deprecated... Yes ...this year after two extensions... Yes ... and the announcement back in 2019. So finally, 2024- Yeah We're gonna get cookies out. Does that mean more business moves to you as people sort of scramble, looking for ways to target without using cookies? Absolutely. We're already seeing it. Do you see an impact with that? We're already seeing it. We actually did some research in the market, where we asked advertisers how they feel about the post-cookie world, what they're gonna do about it. The answer is, they're gonna double down on things like contextual. They're gonna double down on things like predictive, using solutions like ours... Okay ... machine learning... Yep ... scoring impressions. There is no reason why a cookie-based system needs to be the future. The days of ID-based targeting are coming to an end. The next kind of thing to fall will be IP address. So I think we're doing a lot of work to understand. We do not use. Because we're scoring impressions based on the likelihood that an impression is, for example, going to drive an airline sale or a hotel booking or a prescription of a drug or an insurance quote, we're not. We don't care what your user ID is. We don't care who you are. We wanna understand the patterns in the data. That allows us to efficiently deliver media, and then, you know, drive better ROI for our customers. So, you know, cookies go away, those things go away. The main thing for us is attribution. So IDs are useful and valuable in attribution. If you don't have a way to correlate a conversion with ad exposure, then it's hard to prove performance. Can you slow down attribution? Yep. Because maybe some people... Sure ... slow that down. Sure. So we're doing a campaign for United Airlines, and their goal, their KPI, is online bookings, purchasing tickets. We serve an ad to a user, we serve an ad to a user, and then in order for our model to know that the data attributes in that bid request were valuable and likely to yield to a ticket purchase, we need to know which impressions led to sales. If you know which impressions lead to sales, then you can say, "Well, the impressions that all look like this, the impressions that have these attributes, those are the impressions that are buying the tickets." So if you don't have that feedback loop, the way of saying, "All of these impressions purchased tickets," okay, what do all these impressions have in common? That's what machine learning systems do. So in order for the machine to say. It starts, this isn't easy to do. It starts with a significant investment in the data that you get from publishers. Publishers send you a lot of crap. They send you data that's broken. They send you data that's inconsistent. Can you tell me what a publisher is? The publishers are the media organizations that are seeking to place ads on their properties. Like websites, or content. Correct. Like ESPN, New York Times. And in our world, those. And thank you for that, 'cause I think I sometimes... It's our baseball. Yeah. So, ESPN, New York Times, you know, you name it, they wanna sell ads on their properties. And they don't just do it individually, they plug into SSPs, which are supply-side platforms, which essentially aggregate all of those publishers, all those media companies that wanna serve ads on their properties. And those SSPs communicate with companies like AdTheorent, or a DSP, which is a demand-side platform. So the supply aggregates all of the ESPNs and New York Timeses of the world, and they say, "We have these billions of ad impressions. We need ads on these." And then the DSPs, their job is, "Okay, well, we have all these clients, all these advertisers. They wanna buy media, they wanna serve ads, but which ones do you buy? There's billions every second." And that's the job of the DSP. And the way most DSPs have done it from the beginning has been, "Find the user ID that's in your list and target an ad." It's a very simple logic, very simple logic. And what we have attempted to do, and what we are doing, and what we're scaling in market right now, is a system that's much more holistic. We're not dependent on any one data source. We can take all types of data, and the more data we have about every bid request. And when I say bid request, I mean the code that we get from the SSP or from the publisher. And that code is a string of code saying, "This is a web page that was just accessed by a person at this location, on this device, at this time, with this operating system and these other attributes, and there's an opportunity to serve an ad on a page on the New York Times gardening section." I'm mentioning a couple examples, but there's really thousands of data attributes because of the investment that we've made on the front end. A thousand attributes that we consider. And when you have those attributes and the models can connect the dots, it's very powerful. This all happens in milliseconds, 'cause he delivers... The SSP chooses AdTheorent's ad, and then it gets delivered as the... Yeah ... consumer's downloading the page. So it's all... Exactly ... in milliseconds. In that slight second when you kind of are on your, on your screen, and your, the content renders on your screen, and there's that, like, slight second before the ad pops up, that's when all this is happening. Yeah. It's actually unbelievable. It's unbelievably fast. Okay, so going back to goals. Health always made a lot of sense because I remember last year on this stage, you said in order to be HIPAA-compliant... Yeah ... they would have to, like, dilute the... Yeah ... actual sufferer of a particular malady... Right ... with 50% or more of people who weren't in it... Yes ... 'cause then you couldn't tell who actually... And you said, so this was one of the areas that your predictive scoring was really gonna come in handy, 'cause there was a lot less waste. Exactly. If you could predict, even if you were, you know, 20% wasted, that was a lot better than the HIPAA-compliant current version. Right. My question is, when you think about goals, we're on this issue of goals a year from now, when you think about the health product, do you have specific goals for health, or are your goals more like, "Okay, we've done it in health, now let's do it in other silos," like financials, that also have very strict regulations and sort of dilutive current tactics, where you can bring a lot of value? Great question. We have much more to do in health. Okay. We have incredible amount of opportunity in health. Okay. We grew 28% in the Q3. We talked about how we saw that growth only accelerating. I'm excited to talk about the Q4 when we get a chance to do that at the end of February or early March. But what health has done for us or shown for us is it's validated the need for an alternative way to efficiently target audiences. And let me unpack that for a second. So in our world, when you're dealing with the airlines, or you're dealing with an insurance company or some client that wants to just 'cause we tend to be more of a performance engine. Awareness, in advertising, there's generally awareness advertising, and then there's performance advertising. Performance is more tied back to, like, an actual action, a sale, some sort of tangible action, visitation or what have you. But, we believe that... What was your specific question? Health. Health. Health versus moving on to other verticals. Yeah. You're saying you have more to do in health... Yeah, we... ...that's your topic sentence. Yeah, we have a lot, we have a lot more to do in health. I think that the ability to drive those outcomes is critical. And I think that we've shown that when you can invest in the data, and that you can have more of a kind of primary sourced method for analyzing the data and driving performance, that it really works. So, I think we're looking to do. We have another number of ways that we can bring that to fruition in travel. I think, you know, I think when you talk about health, I think that's more of a privacy kind of oriented vertical, and I think because of the privacy aspects, I think there's a lot of value in algorithmic-based targeting. Yeah. So doubling down on that, both on the patient side and on the HCP side, the healthcare practitioner side, we haven't even, you know, really talked about that much. And bringing our tool to self-service users, because I think, you know, there are different ways that you can transact in programmatic. You can utilize a managed service delivery, or you can utilize a self-service kind of version of healthcare marketing. So I think scaling health through self-service is gonna be a huge part of our work in 2024. Yeah, typically. So last year, I had a couple companies that really transitioned from managed service to self-service. It kills your revenue growth, 'cause usually it's half the take rate. Self-service is much lower, like I said, half. So they had sort of a year of really increased demand, but since the take rate was half, and so after that, years goes by, okay, you go back to growth. But it's sort of a really expensive... It's... ... it's an expensive strategic move to go from managed service to self-service. So why do that in a product that's so valuable, it's saving your customers so much money already? Yeah, it's, no, it's an excellent question. We're already, we're already kind of living that. Okay. It's not that we're going from one to the other, we just want to do both. Okay. But the point you're making is a very smart point, and I think it's relevant when you consider, like, our financial results. Because we report managed services revenue on a gross basis, which means it includes the cost of the media. When we report self-service, it's recorded... It's net ... on a net basis. Which means that you're only recording as revenue, your fees. So when you take a customer that's a gross customer, a managed customer, and you bring them to a self-service, essentially it looks like the revenue is going backwards. Yeah, it does. And we just had to be strong enough to be okay with that. And we have been, because at the end of the day, our mission is bigger than one or two quarters in a row. I think we worked through that to some extent. We're gonna be about 10% of our revenue. We're not pivoting... 10% of revenue will be what? Will be self. Will be self. Yeah. Oh, okay. I think we're not pivoting or changing our business from a managed business to a self business. It's not what we're doing. We just want to not care. We want to be able to go to a customer and say: How do you want to buy media, that you can do it however you want to do it here? If you want strategy and support and optimization and data and analytics and data science consultation, if you want all those things, creative, fantastic. We have a business model for that, and we do it better than anybody. If you just want the best machine learning impression scoring system in the industry, and you want to make a business for yourself, and you want that tool in your house, so you can hire people to run it and make money on it, perfect, we can do that too. And because of the fact that we can do that now, I think we have much bigger opportunities with holding companies than we've ever had. I see. So this is targeted at the ad agencies? 100%. Okay, so those are new clients. So those should be new dollars, not cannibalistic dollars. Exactly, 'cause it's a new market that we were boxed out of. Okay. Because if you go into a holding company, and the holding company... He's talking about ad agencies. Yeah, large... Yeah ... you know. He's not talking about Karetsu in Japan. Right. Exactly. You know, holding company ad agencies, like, you know, the very big media buying organizations, Publicis, you know... GroupM ...GroupM. They have their own kind of, call them trading desks, where they have technology, where they're gonna buy media themselves. So we wanna be in their tech stack. We want AdTheorent to be in their tech stack. They're gonna have the Trade Desk in their tech stack, obviously. Yeah. It's a great company. They do a lot of great things, but they do things differently than we do. They don't necessarily have the health capabilities that we have. They don't have the kind of, you know, predictive advertising, the algorithm audiences. There's just a lot of reasons why you would wanna use AdTheorent as a part of your portfolio. And we just... Especially for health clients. Exactly. Makes a ton of sense, 'cause it's so specialized. Exactly. Okay, so more to do in health. Then, you mentioned travel. I thought where you were going after health was insurance or financial services, 'cause our business is so locked down on privacy stuff. Exactly. Why travel? Well, I mentioned travel 'cause it's new. BFSI, banking, financial services, and insurance, is something that we've been working on for a long time. I see. Okay. The growth in that vertical for us was depressed a bit in the last year or so, because a couple of our largest customers were auto finance companies, and they just had different kind of macro dynamics. But it doesn't change the fact that the privacy considerations in financial services, like credit extension, under the Equal Credit Opportunity Act and the you know, the FLA, fair lending, there are a number of rules that limit, prohibited basis variables or targeting based on certain, criteria. And most programmatic advertisers don't even know what that is. And they, I guarantee you, they break those rules... All the time. ...on a regular basis. But smart money, smart advertisers don't wanna do that. They wanna understand it. So we've made a lot of good relationships with those types of customers by educating them about it and showing them how algorithms can be the way forward, rather than, targeting a, for example, a geographic area. A hyperlocal targeting for a credit card or for a mortgage product, can, can violate federal law because it is... You're using location as a proxy for... ZIP codes ...for discrimination. Yeah, you are. And you're gonna get under represent... Facebook paid a big fee about this. Exactly, and that's a big, big problem. And so if you can use algorithms to say, we're not gonna target our credit card based on these kind of proxies, like how for household income and stuff, we're gonna use real-time data about engagement with ads that's not tied back to an individual, and follow those signals. And then use like, use data for, like, lifetime value. And where, you know, where's the greatest lifetime value? Like, which ad impressions are yielding really good, long lifetime value customers, which aren't? And then optimizing towards those in kind of a more of a longer tail optimization. So there's a lot that we're doing in pharma... Okay, banking. In banking, financial services. Banking, financial services. For sure. Okay. But I mentioned travel 'cause it's net new, and we're quite excited about it. Okay. Well, that one I don't get, but it sounds like we can talk about it next year. So right now, we've got health, and we've got banking, all of which I understand as having big competitive advantage. Plus, my opinion is a lot of business is gonna come to you with cookies deprecation. Yes. People are gonna be throwing up their hands, panicked about how to target. They're gonna be worried about their return on invested capital when cookies, as cookies deprecate. And I think you're gonna get business from places probably you're not expecting, because you don't use cookies of any kind. Exactly. People will be looking for substitutes. Yep. Does the health vertical... If you go into, is the health vertical like a camel under the nose of a tent, where you can go into the big ad agencies now and say, "Hey, we've got this healthcare-specific problem," and that gets you into their. But you have to use self-service. I get that that's a complementary. Yeah. You have to have healthcare and self-service... Yep ... 'cause they do it themselves. Does that then let you make it more likely that the next product is easier to sell for you? Absolutely. For 2025, maybe some, okay, whatever the next product is. Yeah, no question about it. I mean, we always viewed healthcare as a beachhead for... Okay ... expansion. Yes. You know, one could say: Why not just be a healthcare business? Yeah. Um... That is a good question. Yeah. What's the answer to that? I think it's because there's so much opportunity outside of health. I mean, we do believe that our health business will grow at a rapid rate. And that, you know, it is in large part because of the fact that we're highly, highly differentiated and adding a lot of value. The results from these campaigns are excellent. But that same infrastructure and that same framework that we've built in our platform works in a lot of other verticals, and the tech is built. And when, you know, you can bring in to travel five people in travel, who can kind of monetize that and provide great results for hotels and cruise lines, why not do that? Like, I think we have the tools, why not sell them in different markets? Having said that, we do have specific points of view in health that are different. Okay. I mean, and my answer to that would be, t here's no way you should be able to get the pricing on travel, where there's a lot of substitutes, than you do in healthcare. And if you are, then you're undercharging for healthcare, 'cause healthcare really is differentiated. Travel, there's a lot of guys that do travel because it's sort of low-hanging fruit, right? Well, that's a good point. But I think the reason why travel works so well for us is because of the performance nature of our executions, and that those customers really care about performance. In other words... Okay ... bookings... Yes ... and visitation. Yep. And then proof that you delivered value. So one way that we do that is working with airlines, we can prove, we can prove that we're driving a lower cost per action on their bookings for their airlines, and their cruises, and their hotels. And then for, like, travel boards and, like, you know, Visit Colorado, Visit Florida, you know, we can show through some partnerships that we have with credit card companies, that not only are we driving, visitation to locations, but we can then report out on the incremental spend that they made in those... So it's about, like, some of the additional partnerships and services that we're providing that other, you know, pure DSPs are not. And then, moreover, those audience algorithm solutions that we have are allowing customers to build custom, targetable audiences for travel that have been quite valuable. So I think... Okay ... those things together are making that... Okay. Well, the travel thing sort of surprises me, but okay. Yeah. That's a new one. A year from now, you'll have that more. Yeah, you... Let me go to questions from the audience. Yes, sir? It seems like earnings are down from where they were in 2021 and 2022. I guess you have a lot of growth opportunities with just the, the actual incremental margins as we grow, and how you kind of look at the opportunity for earnings power? Earnings meaning? The earnings per share. Okay. Yeah, so I mean, we - so we went public at the end of 2021. We took on a number of new costs. So I think that's been part of it. But having said that, I think our, like, our adjusted gross profit, our AGP margins, our EBITDA margins are really strong and getting stronger. One of the reasons why our audience products are valuable is because with... We can replace to a large extent certain third-party costs with our own products. And so I think our AGP margins are gonna continue to increase. AGP, adjusted gross profit. Adjusted Gross Profit. Okay. I think that our platform has been operating more efficiently than ever, not less efficiently. Our kind of pullback, if you will, on earnings, was mostly tied to the entry into the public market, some of those incremental costs. We made a number of investments. There was a reason why we went public. We made a number of investments in data, to make our health business possible. We made a number of investments that were kind of one-time investments in the tech, and the development of the products and the tech, that we won't have to make again. I think we can obviously continue to iterate and improve, but, you know, we think that when we get to 15%, 16%, 17%, and more% growth, our operating leverage is incredible. Like, we have, you know, +30% EBITDA margins, as long as we have that growth. We had a couple quarters since becoming public, where our growth went back 5% Q1 of 2023, 12% Q2 of 2023. There were a number of reasons for that. They weren't because we don't have a great product. There were just we're small, there were some supply chain concerns, there were some, there were some macro concerns. But we just kind of doubled down on what we're doing, and I think, you know, we could be easily a Rule of 40, Rule of 50 company. When you talk about revenue growth, EBITDA margins, that's what we were right before we went public. We've had to make some investments. We've had to kind of adjust some of the new costs. But we're as lean as we've been ever, and the efficiency that we're delivering is continuing to get better. So, the answer to his question is: if we see the revenue growth you're projecting, we should see faster earnings... Exactly ... earnings expansion... It all... ...and higher net income margins or EPS... Exactly .....margins. That's why my first answer to your question... That's what you're getting at. ... was just return to sustained, kind of repeated growth. I think you were thinking top line. I think he's asking about EPS. Right. You're saying they're linked... As long as we have... Because you don't have more costs. ... as long as we have fuel in the plane, we can fly, and it'll be beautiful. We just need to have enough fuel in the plane, and we had some challenges in that regard a couple quarters. Again, you know, we try not to beat ourselves up about that. We have a bigger view. But as long as we have that, which we're, that's why we've been laser focused on it, we believe the earnings are gonna follow. Okay. Other questions? Okay, let's talk about political real quick. Sure. Every three months there's a higher... Today, GroupM is out saying it's gonna be $17 billion, which is up 34% from two years ago, and up 24% from the last presidential. Do you guys benefit at all from political? How does political affect your top line? Yeah. Well, with our audience, with our audience-building products, we have a number of ways that marketers can use our data and our tools to build great political targeting products. Okay. So we're excited about that. It has not been kind of a main focus for us, you know, in terms of just kind of focus. Historically? Yeah, it just has not been. But we think we're gonna still benefit from it. I don't think it has to be, like, our primary focus area for us to benefit from it. I think it's also kind of a... It's about who you have on your team, it's about, you know, who are the kind of the people that control those dollars, and do you hire those people? And it hasn't been a primary focus for us. I think health, and some of these other have been- Yeah. Yeah. They're more systemic... Yeah ... whereas political's really... Yeah, it's... ...volatile. Yeah. Generative AI. This is a good question for you. Yeah. One of the things, your stock was running the other day, and I was getting incoming calls just asking. I've said for a long time that if somebody wanted to just buy talent, you guys are at the forefront of algorithms and AI, and that's a good place to start. It's hard to find generative AI engineers. Talk to me about how you're integrating generative AI into your products. Well, a couple things about generative AI. This is maybe a little bit counterintuitive to your question... Okay ... but generative AI is gonna generate a lot of crappy content online. Yeah. It's gonna be its primary focus, actually. Exactly. That's its primary contribution: crappy... There's just gonna be... ....content... ... an incredible amount of crap on the internet. There already is. ...there already is. But it's gonna, it's gonna multiply at a level... Ten X ... that you've never seen. Yeah. So then, from the perspective of an advertiser, like, our ability to sift through that crap and find value is even more important. Because we call it AdTheorent IQ, in inventory quality. We do not operate on a kind of open web, serve on any impression just to get clicks. You can very easily game the programmatic media buying outcome, exercise. Like, in the early days of programmatic, those types of things happened a lot, and I think advertisers got smart. The industry is much smarter now. We operate on a curated set of properties that have to be approved before they're even in our network. We do not permit made-for-advertising content in our network. The... You mean you won't serve an ad to a property that isn't on a list? Yeah. Like, you have a list of who you'll serve ads to? Correct A nd therefore, you avoid made-for-advertising sites? Correct. We've excluded them from our... Oh, I'm a big user of those, just so you know. You know, look at all the dresses of the Oscars, it's 10 ads for - that's me. But a lot of - there's a lot of ways through MFA, where a lot of ads go to those made-for-advertising... Made-for-advertising ... sites, where there's just not a lot of ROI coming back to the brand. Okay. So, you know, we exclude that, we exclude cheap reach, we exclude multi-hop. Multi-hop meaning inventory that's being sold, resold, resold by all the different SSPs. That's why, and I know you've covered this a lot, there's a move towards Supply Path Optimization... Yeah ... and direct integrations. You disagree. disagree. Yeah. I disagree to the extent that I think there's a different way to get to the same outcome. Okay. Which is, you want to make sure that you're not overpaying for inventory. One way to make sure you're not overpaying for inventory is to go direct to the publisher... The seller. Yeah, the seller... The seller of the ad space. ... of the media, and get all the middlemen and the middleware out. Yep. But there are other ways to do that, which we think are more efficient, because the scale you get through programmatic and working with SSPs is the reason why it exists. So to integrate with every supplier, every publisher, is just prohibitive. So I think in our world, where we can kind of use algorithms to detect when inventory is being resold, we can avoid it. Right. So there's ways that we can do that. So I think AI, getting back to your question... Generative AI. Generative AI will be a threat to the quality of the supply in the media... Right ... I'm sorry, advertisers will need a solution for. We are that solution. And then in our UI, our new UI kind of work, there's ways that we can use generative AI for ad unit building. Okay. We have an in-house creative team. It's incredible what you can do. It's all about structured and controlled data sets. If you have your own data that you control and you know it's accurate, you can do wonderful things with generative AI. If you use unstructured data that's not of questionable source or reliability... Those large language models that are... Yeah ... sitting in AWS. If it's not correct, if it's not validated, then you're garbage in, garbage out. It's just you're just gonna kind of make the internet dumber. If you build a generative AI based on factually correct, curated, smart information, you're gonna get smart outputs. If you just take the entire internet and you use that, you're gonna have dumb ideas with good ideas mixed together, and it's just gonna kind of dumb down the internet. Does it start a sort of spending more, generative AI? I mean, I have guys using it, mostly in Israel that are saying it's gonna cut out 20% of their cost structure. They're gonna use it. They're tech guys, and they're gonna use it, first draft of code, all being written by generative AI, and then their guys are editing it for new products. So you're not using it in any of those ways, so it doesn't sound like in your business. We're looking into it. Our tech team is definitely looking into using it in different ways to make our process efficient. But like, the core of what we're doing in the machine learning scoring, it is not really relevant to that. Okay. All right. Yeah. It's mostly, your product is becomes more helpful the more bad content there is out there. Yeah, bring on the bad content, 'cause then you're gonna need... Right ... you're gonna need filters. Better scoring. Yeah, there's just gonna be... Better scoring ... you know, if you're, if you're L'Oréal, do you wanna serve ads on just, just impression after impression after impression of essentially computer-generated content that nobody reads? Okay. You know, you can, you can get clicks, but what is that? That's why, like, even in the upper funnel, it's meaningless. Like, even in the upper funnel, which is considered awareness and not performance... Yep ... it has, there has to be a performance component. Right. You need to know that you're actually engaging with people. Yeah. Whether they buy something or not is maybe, like, secondary to an awareness campaign. Uh-huh. But it can't just be mindless clicks that are generated because maybe the made-for-advertising content was smart enough to know that to X out of the ad, you actually have to engage, you have to engage with the ad. These games happen all over the place. Okay. I'm gonna call it there, 'cause we're up... Yeah ... against time. Thank you. Love seeing you. I appreciate it, Laura. Thank you, everybody. Yeah. Thanks for your questions. Thank you.