The Trade Desk, Inc. (TTD)

Investor Day 2017

← View all transcripts

Oct 4, 2017

Good afternoon, everyone. I know we still have some people walking in, in the back, but if you can find a seat, we're going to get things kicked off. While people are coming in, if you have not seen a copy of the presentations that our management team will be presenting today, you can download them now on our Investor Relations website at investors. Thetradedesk.com. We'll have all five presentations plus supplemental non GAAP information there. We have a full agenda for today. We do have 2 breaks scheduled in. And in addition to the management team presentations, we're very excited about 2 panels that we have as well, an ad agency and brand panel hosted by Brian Skempek, our Chief Client Officer, and a client team panel hosted by Rob Purdue. Just a couple of housekeeping items. Already talked about the presentations. We do have some USB sticks in the back if you want to get the presentation that way. In terms of questions, we do have a management Q and A session following Jeff's closing remarks. So, we ask that you refrain from any questions during the program. And I'd mentioned we had the 2:10 minute breaks. And lastly, before we begin, the most important piece of information that we'll provide today is the forward looking statements that we're going to be providing information that is forward looking. We're asking you to beware, be cautious. All the risk factors are in our SEC filings, 10 ks, press release, etcetera. We do ask that you refer to them and be fully aware of the risks. They're all posted on our website as well. So with that, I'm going to announce our Founder and CEO, Jeff Green. Jeff? I'm delighted to see the room full. So I'm going to I tend to pace a lot, so I'm going to try to stay out of the way as much as I can, but definitely want to be able to move around. So I want to start by just contextualizing what's happening in advertising today so that you can understand more about our business. But I think it's really important before we jump into the numbers and the strategy of our business to understand what's happening in the space. So advertising because of the Internet will never be the same again. And right now, the problems that the Internet have created are at a tipping point where things have to change. You cannot be doing business as usual. So the advertising that was happening 10, 20, 30, 40 years ago, where you essentially place orders will never work again. And it's not working today, because brands are spending 1,000,000,000 of dollars to make consumers hate them. And the reason that this is happening is just because of the fragmentation of media, right? As more and more media divides into smaller pieces because of more websites, more TV channels, all of that fragmentation, brands are not being as coordinated as they can be. And that's even true inside of digital and things like Google and Facebook and others, those are built in sort of walled gardens, where people have to buy in a silo without referencing what's happening in the rest of it. So when you see an ad on a commercial break and then you see the same ad the next commercial break and then you see the same ad on your phone, that's the reason why consumers are starting to feel like they're being bombarded, not respected. Their data is not being used to actually make it more relevant. And instead they get annoyed by the brands that are advertising to them. So I've spent a lot of times with CMOs lately, and I just want to relay from them to you some of the anxieties that they feel. I'll do this in a couple of places today, but I wanted to start with the biggest one. And that is that because media is moving online, more and more of them are having trouble getting in front of especially the younger generation. So Gen Zs and Millennials are not consuming content the way they once did. And especially in television, where, in my opinion, that's where the heavy lifting of advertising is done, which is where you use picture and sound to really win hearts and minds. You can't get in front of a 6 year old, a 10 year old, a 15 year old today if you're doing traditional advertising the way you did 10 years ago. So while the media universe is getting bigger, the coefficient on advertising is going up. And sometimes, I think we I'm going to get closer so I can see my notes. Sometimes, I think that we're forgetting the fact that the Internet is actually growing because we tend to focus on some of the sites that are doing better than other sites, and we think of it as consolidating, when in reality, GoDaddy continues to sell more domains. YouTube continues to get more videos posted to it, but that's not just about YouTube, that's also about the content creators. And then of course in TV, more and more content is going from traditional television so that you see more commercials, more full episode players. What was you take somebody like Crackle, who didn't have much content on their sort of Roku and Amazon apps not that long ago, Now they have real content inside of Crackle. It's just one example of many that we're adding tons and tons of new content to this ever expanding media universe. And that expanding media universe makes that problem we were talking about earlier worse. So this is becoming more and more consensus in the world of media. And I know that this has been a controversial this would have been controversial to make this statement 5 years ago, but it's less and less controversial as time marches on. And that's partly because many of the traditional players in television didn't take strong positions on the statement. In fact, their positions were more everything is fine. And then when you hear a vision from people like Randy Stevenson about the future of television and that they're going to go all in on 5 gs and on demand and a totally new experience for traditional TV owners. And you hear Disney, who didn't have a programmatic strategy 2 years ago, talk about it a lot now. You're seeing an evolution where they recognize this reality, which is things typically do not move in a linear fashion and that television as we know it will be irrelevant in approximately 10 years. And when I say television, I don't mean the content. So those that are producing amazing content are going to continue to do well, but they're going to do well with a better monetization engine. So I think this is a happy ending. This isn't doom and gloom. This is great news for almost everybody. So let's get to some controversial stuff, all right? So I've had a number of conversations with the companies that are considered walled gardens, right? So you take a company like Google or Facebook, and Google really is a walled garden inside of YouTube more specifically, where they're saying the only way that you can access that demand is through them. And those 2 have set a playbook that other people have tried to copy. And basically everybody inside of social have tried to copy that playbook. And what I think if you spend a lot of time studying the people that are lower down the list, the Snapchats, the Twitters, the Pinterest of the world, if you spend time looking at their businesses, that playbook does not apply to them. And the reason why it doesn't apply to them is because as media fragments, advertisers and agencies do not want to log in to 100 platforms or 200 platforms. In fact, they don't want to log in to 7. And especially if you're a global brand, like if you take somebody like Procter and Gamble, who has over 100 brands and advertises in over 100 and 50 countries, multiply those two numbers together. Now multiply that by 7. Do you want to log into that many platforms? Of course, you don't want to. And so what we're predicting is that as all of them are trying to create more effectiveness, their and when I say them, I mean the publishers themselves. When social networks are trying to figure out how do I make more money, how do I Snapchat become profitable. It's our belief that what you have to do is open up to programmatic and get as much demand as you possibly can, because it's the optimal way for you to monetize. I've even made that assertion to YouTube and to Facebook. And even in those four walls, there are people that believe that the optimal way to monetize is to get as much demand as you possibly can, even with the strategic losses that would come from opening that up. So I believe that, that is inevitable. It's simply a matter of time. And we'll talk more about some of our predictions around this front. So I just want to highlight just a really quick case study, which is Dropbox. And when Dropbox came out, there were hundreds of companies trying to do something similar, all of it different slightly different brands, slightly different approaches to it. But basically, you take your files, you put them in the cloud, nearly all of them were based on S3, by the way, including Dropbox, I mean Amazon's cloud product, and you create a relatively simple software that makes it so you sync everything to the cloud. Dropbox won, the other 100 lost. And I would argue that the reason for that is because of advertising. And so what I'm arguing is that because the barriers to entry on nearly every business and part of the reason why I reference somebody who's built on-site on top of Amazon is because Amazon's cloud product lowered the barrier to entry on that type of business, which is why there were so many trying to do it. But the difference between Dropbox succeeding and others not was their marketing. They did an amazing job of that. And so what I would argue is that success the equation of success, the coefficient on advertising is going up, meaning you have to be better at advertising or you're going to be one of the other 99 companies that doesn't win. So to just take a look at the size of what we're going after, as we talk about this massive opportunity and differentiating for those companies like Dropbox, I just want to point out that we're about 10 years away. So if you assume 5% growth, which you notice over the last few years, it's been above 5%, in fact the lowest year was last year I'm sorry, the lowest year and this prediction is in 2021. But if you assume 5% growth from thereon in 10 years, we are a $1,000,000,000,000 industry in advertising. I've just taken the time to compare that to GDP on the bottom. So we have an amazing tailwind at our back, and we are pointed at a massive TAM. So as you look at GoPro saying how big is this market really? You look at Yelp and GoPro saying how big is this market really. You look at Yelp and say how big can you really be. You're already a market leader and how much bigger can you get. And I just want to highlight that we're pointed at a $1,000,000,000,000 industry with massive amounts of room ahead of us. So now instead of talking about the macro our biggest day ever in May of 2011 was our first day. So our biggest day ever in May of 2011 was our first day when we launched with our first $0.08 day. So that represented less than 1,000 ads. It was a big day. It was really a big day for us. Hope they go away soon. And then, of course, a $1,000,000 month, which, by the way was right about the time that we got profitable. It was so exciting for us because we did something that was different than most early stage companies and different than most of the so called unicorns, which is instead of raising as much money as we possibly could, we had this insane idea of what if we raised to profitability. And so we got profitable. And then as soon as we became profitable, we changed everything about the incentives in our own four walls to reinvest every single dollar that we possibly could. So the culture that you'll hear us talk about throughout this presentation, which is we make money and we reinvest it as fast as we possibly can so that we can grow our business, is something that we started doing in 2012. You'll see that we say we've been profitable since 2013. It's just because we deliberately went in the negative at the end of 2012. And so the lawyers make us say from 2013 on. So one of the questions that's been asked of us a lot lately is, are you a SaaS company or are you an ad tech company? And I want to be super clear that we are a SaaS company. We are a software company that just happens to be pointed at advertising. We've at times toyed with saying we're a SaaS data company because of the fact that we get all this insight from looking at every ad available on the Internet. It makes it so that we're in a really great position to deploy data. So yes, we are a technology company in advertising. But I think that the mistake in putting us in that bucket is that most of the ad tech companies have very different business models. And I would argue none of them have been SaaS companies. And SaaS, of course, does not necessarily mean that you charge on a subscription basis. To me, the great earmarks of a SaaS company are those that have recurring revenue that is predictable and, of course, has massive amounts of operating leverage and the fact that we operate with higher EBITDA percentages, then I think every publicly traded SaaS company puts us in a great spot. So I'm going to very quickly, so those of you that heard from us on the roadshow, many of you we've met with before, you've heard some of the basic beliefs that made us start our business 10 years ago. So these were things we believed 10 years ago that we still believe today that instead of spending a lot of time on them, I just wanted to spend a little bit of time to go through them all. So first, if there's anything we learned from the financial markets in 2,008 2009, it's that price discovery is really important to the health of a marketplace. And so one of the things that our business does is it enables price discovery in a market we think can be better at it. So we think we're doing one of the very things that is essential to healthy and sustainable markets lasting and because we think we're doing that at an optimal level, that's why we in the very first deck we ever pitched at The Trade Desk, we said we are building a business that will be around for 100 years. So data driven is better than guessing. So most of advertising has been spray and pray. You just guess, it's order placing. That has been the history of advertising. We don't do that. We actually use data to make decisions. We think that will forever win. So this is not old media against new media. That's not the right paradigm. This is data driven versus guessing. Which one is better advertising is really the question. We believe that because advertising has gone from 600 since we started to 675 this year and will be a $1,000,000,000,000 industry within 10 years. And of course, that doesn't factor for any additional quantitative easing. We'll most certainly continue to have a big opportunity in front of us to continue to win market share and grow our business. The buy side is forever in the power position, and that's because you can it's easy to add another 32nd spot to a commercial break. So if you study what's happened in advertising in the last year even, just what happens during the Olympics, what happens during an election, you see more ads in a commercial break. It's not hard to put another tag on the page to put 5 ads per page instead of 4. So anytime you get close to equilibrium, they will add to the supply and that makes it so that it is again a buyer's market because there's more supply than demand. So anytime you get close, somebody will add supply in this super fragmented market, which means it's forever a buyer's market. Decisioning in Wall Street is where the money is, the people who make the decisions about what to buy and sell are the people that make all the money. That's exactly true in our world as well. So as markets mature, really the money extraction, the value add comes from those that discern between what to buy and what not to buy. That's the business we created. That's the business that we're in. We think the sell side will continue to fragment, which makes it so it's hard to be a power player on the sell side. The sum of tech take rate, we think is 25% to 30%. So that's buy side, that's exchange side, that's sell side. We'll talk more about ours towards the end. And then we think the buy side aggregates. So as the sell side fragments and there's more and more sell side players doing things on their own and there will be hundreds of effective exchanges the way that we think about exchanges today. We think there's 3 to 10 max and 10 is being generous players on the buy side. And we think at least one of them has to be an independent and objective player as a contrast to the rest of those. And we think right now, we own that position. So I probably since going public a year 2 weeks ago, have probably answered a question about what about Google and Facebook at least 500 times, right? It's a legitimate question. I wanted to spend a little bit of time on it. But I especially want to explain why I don't think we're competing with the core of Google and Facebook. So most of their businesses, I think, come from new media. And by new media, what I mean is, they're monetizing google.comandfacebook.com. And the people that they're competing with are all of these. And in Google's case, they also monetize youtube.com. But they're competing against all these players that are creating, in my mind, really premium content. These are the people that are competing against 95% of their P and L. So it is true that we are competing with some speculative part of their businesses, which are not significant parts of their business from a financial standpoint today. And I personally think there's strategic question as to whether or not those initiatives in those two companies in particular will last, because they don't move the needle and it's super hard for them to be objective. We are not a media company. We don't own any media. We don't plan to own any media. And so as such, we're not competing with them. In fact, we go to the biggest brands and biggest agencies in the world and say, we're going to help you objectively figure out which of all these you should buy, including among Google and Facebook. Now we don't do much with Facebook today because of the walled garden, but we spend tens of 1,000,000 of dollars a month with Google. So they're an amazing partner to us and we expect them to continue to be because I think we're their largest independent buyer. So let's talk about TV for just a little bit. So a lot of numbers here. 1 in 6 American households are already cord cutters. That's 20% essentially of the 125,000,000 households. 67% of TV watchers recently were surveyed and they said they're not willing to pay more for a service without ads. We'll talk more about that concept in a second. When you think of the 67% and I personally think that 75% to 80% is basically market saturation. And the reason for that is there are about 10% of American households that have never had cable and perhaps never will given that it's the cost is going up. And I think the quality of service is going down. And then just to talk about digital audio is just a frame of reference for there, 61% use digital audio in some regard. So you think about how ubiquitous Spotify and Pandora are, as we talk about that 75 ish percent sort of saturation mark, we're not too far away from it in that world. So really, all of the things that are happening in TV start with one amazing thing, and that is that TV content is at the top of its game. So I imagine all of you as consumers can agree that we're at the best place in our lifetime in terms of how much great content is created. It's impossible to go to a dinner party or to go to dinner with a friend where you're not talking about some amazing TV show that you recently watched that you're saying something like, it may be my favorite ever. And you forget that you said that about something else like 3 years ago. And the fact that M*A*S*H season finale had 110,000,000 people watch it and Breaking Bad had 6,000,000 people watch it, that disparity is a commentary on fragmentation, but it's also a commentary on how great content is that it's that fragmented. And at those dinner parties, every time we all walk away with a list of like, oh, yes, I've got to go check that out. I've never even heard of that, right? We all have that experience. But this, not only is it a beautiful thing, it's also a problem. It is what is the forcing function of every problem that currently exists in TV. And here are the 5 biggest that are making change. So these are vectors for change. The first is that content costs are going up. So in order to make that amazing content, it costs money, right? And so if you look at the spending that's going on, particularly between HBO, between Amazon and between Netflix, You it's like the Cold War, where the spending on defense is just through the roof. And that creates trickle down challenges. But it doesn't just affect those 3. Everybody else is trying to keep up with those 3. And imagine doing that at Disney, while cord shaving, which we haven't talked about, is perhaps the biggest thing affecting ESPN. So subscriptions are prices are capping. So some important stats here. 67%, so we talked about those thresholds being at like 75%. So we were above that in cable and we've been dropping below it. So 67% subscribed to cable, 67 or 68% of adults above 12 subscribed to Netflix. So we're both at those saturation points there. What I think that represents and what we're going what we'll spend a little bit of time talking about is that because most people are subscribing for cable and most people are subscribing to Netflix and then a lot are also subscribing to Amazon and a bunch are subscribing to Hulu. Consumers are at a place where they're saying, fact, 67% of the previous are saying, if you offer me something, give me an ad funded model so that I don't have to pay another $12 or $15 because I'm tapped out. So this is happening while content costs are going up. And then table subscription costs have doubled. So what that's doubled in less than 10 years. So it was around $80 We're now with taxes and everything, if you include all that stuff in it, we're at now over $150 And the number of ads per pay or number of ads per commercial break have also gone up. So I think traditional television is a ticking time bomb as we continue to add all these vectors. But perhaps the biggest one is that connected TV is just better. So we've created an experience, while this one is under pressure, while the old one, the traditional one is under pressure, companies like Netflix have created an experience that is just better. On demand is better. And that's the reason why companies like AT and T are making existential bets. They're making moves that I think affect the outcome of their company in the future. So this is perhaps the most bullish statement or number that I can share all day. And we talked about this in our Q2 earnings report. But and this I think this reinforces my point about subscriptions capping out. Connected TV ads available to The Trade Desk increased by 10x Q2 2017 over Q2 2016. Now cord cutting didn't 10x or anything like that. It's just that we have a small amount of ads available in places like Hulu and Crackle and Roku, who I'm excited that they're doing really well. We have a small amount of ads available in Connected TV. And because CMOs are worried about that missing out on that generation, they are paying a massive premium to get in front of them. And because that creates a pricing disparity, where in many cases, the prices that are willing to pay for connected TV ads are more than double what they pay in traditional television. What that does is it makes it so other content owners are, let's go tap into this. So it's accelerating the move. So to just talk about our big picture ambitions, we think that we fund media. It's not just that we are trying to monetize our website. We're actually funding all of media and creating a better experience for that. So whether that's on Spotify or whether that's on The New York Times, we are creating a better economic engine for media. In our S1, we said we power the marketplace of ideas. We make it possible for ideas that never could have seen the light of day before actually see the light of day. And media does not exist without ads. There is no world of media without ads to some extent. It doesn't mean that there aren't successful paywalls, but we all know that those have been few and far between. Even companies like The Wall Street Journal struggled initially with paywalls, and they arguably have the richest subscriber base of any journalistic outlet ever. But companies like Dropbox don't exist without ads and the coefficient on in the equation of success on advertise sorry, the coefficient on advertising in the equation of success is going up. So there's a lot of negativity in the world about global problems. I just want to point out for a second that there is a growing middle class around the world that we're experiencing that has never been experienced before. I would argue never in the history of the world, especially on a total numbers from a total numbers view. And I think sometimes we think about the pressure on the middle class with a very American centric view, because if you look at it from a global view, it's actually extremely positive. And this picture was taken in Indonesia. Indonesia is one that I love to talk about because it has one of the fastest growing middle classes in the world. And as we talk for the next minute or 2 about the growing opportunity among the middle class around the world, I just want you to remember them because in the next 15 years, they're going to make decisions about which detergent they love. They're going to make decisions that their parents never made about buying a motorbike, about buying an air conditioner, about buying a refrigerator for the very first time. And if you are a global multinational company, you want to win their hearts and minds, because that can have an effect for generations. So there's this really beautiful thing that is happening around the world, and it's this, which is the slope of the line over the next few years as as steep as it's ever been as the global middle class is growing. And so to break this down just a little bit, well, actually before we do that, I just want to remind you that maps lie and actually the Internet in some cases makes us dumber, right? And it does when it accounts to this, which is that at the top of the globe, you actually are disproportionately making those things look bigger than they actually are. So Russia is not as big as it is if you compare it to something in the middle. So Indonesia is actually much bigger than you think it is, which is why it can contain a population that's the same as the United States. So many people discount the size of the opportunity just because of geography, which is just not fair. So just a quick math lesson before we talk about the actual opportunity in Asia. Yes, you're welcome. But here's the actual opportunity, and we'll look at the actual numbers in just a second. But look at where all the growth is coming from. Look at Asia. Look at how much opportunity is coming to the middle class in Asia. So North America is seeing roughly the same, the green all the way up. Europe is actually about the same, maybe shrinking. In fact, we'll look some more precise numbers in just a second. But all of that growth is coming in Asia. So here are the actual numbers. So North America, in terms of percentage of the middle of the world's middle class, we go from 18% in 2,009, which were of course is in the rearview mirror to 7% in 2,030. Europe is going to go from 36% of the world's middle class to 14%. But across Asia, we go from 28% to 66%. So when we talk about the offices that we've opened in Korea, in Japan, in Shanghai, in Hong Kong, in Indonesia, in Singapore and of course in Australia, a huge bet that we've been making since the inception of this company is that multinationals want to advertise to the fastest growing middle class in the history of the world and most of that's going to happen inside of Asia. Of course, Indonesia is only the 3rd largest growing and China and India are massively and disproportionately representing the growth of that middle class. To dive into China for just a second, the 2nd largest middle class in absolute terms with 157,000,000 consumers, but that is only 12% of the population today. It could grow to 70% by 2,030, which represents a massive opportunity. So our strategy in China, just to touch on this for a second, is to grow from the outside in, Because many of you have asked the question, we get asked a lot, why do you think you can be successful in China when so many American headquartered companies have struggled to go into China? And it's really for 2 major reasons. First, the government likes us to be there. And second, the big corporations, which have as much influence as the government, also want us to be there. And the reason why they want us to be there, man, what's happening today? President's in town. The reason why the big corporations want us to be there as well is because we're writing them checks. So we do not come into the China market there we go. We don't come into the China market trying to win Chinese consumers to our business or to our brand or to our website. Instead, what we're doing is we're taking brands that they already know and love from multinational companies who are extremely successful and want to advertise what will soon be the largest economy in the world. And we write checks to companies like Baidu, Alibaba and Tencent, which are some of the most powerful companies in China. So because we're bringing money from outside in, we think we're in a much different position than if we were trying to win Chinese consumers to Uber or something else like that. So let me talk a little bit about the state of the industry and then how we're doing compared to that. So digital is growing by 12% over the last 4 years or 3 years, and we've grown with a CAGR of 90% relative to digital. Of course, if you want to break that down by channel, we're happy to. We're super excited about audio. Wow, 2, I think that's not quite represented at to scale, 200,000 percent growth in audio. But even in video and connected TV, which massive amounts of that have come online, and of course, display and native and display in particular has been just small amounts of growth. We've grown at almost 10 times the pace of the industry. And then we've done, of course, 3 times that of mobile. And mobile, in terms of a new channel, is the fastest growing channel, I think, we'll ever see in the lifetime of our company. So I think what we've experienced in mobile is something that we won't ever see again, just because the world has just moved mobile so fast. So if we break that down by countries, and I apologize for a couple of them not being represented here, it's just because of the way we've tracked historically, where things like Indonesia were included in Southeast Asia. But if you look across geography, we've also outpaced the industry in every possible way. Super excited about the disparities, especially in Asia. So if you believe like I do that the growing middle class is a huge opportunity for our business, and then you look at the fact that we're outpacing the growth by even more in Asia than we are anywhere else in the world, perhaps next to that connected TV number of 10x, these are the most bullish numbers that I can show you today. Because where the most growth is, we're actually outpacing the growth by the most. So another question that I get asked a lot is the barriers to entry. So okay, got it. You guys have historically been winning. Why? And how defensible is that? So I want to take a little bit of time to go into more detail than I probably have ever been able to do that, certainly publicly. And with nearly all of you in the room, I haven't had a chance to go in this level of detail. So in order to do that, we have to talk about the landscape. So you have to understand a little bit with a little bit more detail the nuanced and esoteric nature of digital ads. So for those of you that never wanted to understand this, bear with me for 10 minutes. It will help you better assess how we're doing and why we're doing as well as we are. So first, let me just say the objectivity that we think that those new media companies don't have matters more and more as time goes on. And part of the reason for that is if you are a CPG company or an automotive company or a finance company and you're trying to differentiate from your competitors. You're trying to separate from the pack in the same way that Dropbox did. You have got to be good at advertising and marketing. If you're not, you're going to die. And the way that you get better is you deputize your data. So if you're the largest advertiser in the world, Procter and Gamble, who spends $8,000,000,000 a year in advertising, if you're not learning from that, you're doing it wrong. And you'll lose to somebody else who is learning from the spend that they do every single year. So if you're going to not lose, you deputize your data, now who do you trust with it? It's because in order to stay competitive that you have to use that data, the objectivity of the technology and your business partner that helps you buy in this crazy esoteric world that I'm about to describe matters more. So there's a bunch of businesses that I would say the historical ad tech businesses have been those that they ride waves instead of accelerating change. So what they instead say is, hey, I don't have to be transparent. I can be on the buy side and on the sell side. I can ARB media. I'll buy low, sell high. I'll operate at 60% margins. Like it's the same things over and over again, which is they're not transparent and they have margins that are not sustainable. They have businesses that often don't have operating leverage and therefore aren't scalable. We decided to build a different type of business, again, a SaaS business that instead of riding the waves is accelerating the change. So I say that to just give you caution as you're making comparison between our businesses and or our business and any of those businesses that we have really built a fundamentally different business. So header bidding has changed everything. And this if you don't know what header bidding is, let me just take 1 minute to describe it, okay? It used to be that when somebody landed on a page, an auction would be run-in oneten of one second and a publisher had to decide who was going to run that auction for them. You had to pick 1. And if that company that you were picking took a higher margin than you thought they did, then you would not take as much money home as you would if you'd pick somebody else. And so what publishers figured out a way to do, in fact, other tech companies help them figure this out, is what if we ran multiple auctions on the same impression at the same time. So that the same impression, the very same ad in oneten of one second can be considered from one exchange and another exchange, so that I could bid on it from either of them. And that creates a level of competition that we didn't have before. But it does mean that impressions are duplicated. So if you were looking at 3,000,000 auctions a second 2 years ago, even without any growth in the ecosystem at all, you're going to be at 6,000,000. That has happened. And then that means that if you have a P and L like ours, which is really simple, right, 60% of our costs go to people, 30% go to machines, 10% go to airline tickets and hotel rooms. That's basically our P and L. So the 30% that goes to machines, if I double that cost, I'm in trouble. So we'll talk more about that in just a second. That means that SSPs have to run more auctions and monetize fewer of them. We'll talk about that in a second. DSPs have to participate in more. Operating leverage, if you didn't have operating leverage before, now you have to participate in more auctions, you're in trouble. And that means that the profitability finish line has moved. So if your growth rates, if you're not outpacing the industry with a meaningful clip, like the way that we just described in our business, then you're already losing because of the fact that that 30% and that's only in our case, which I think our tech is really efficient and our team is really lean. But in another business, that could be more devastating. So let's talk about them. So on the buy side, if you used to look at 1,000,000 auctions per second, and let's say you monetize 20,000 of them. And to be super clear, every time I look at an auction, it costs me money. That's true of, I think, every DSP's business model in the world. Every time they look at an ad opportunity, it costs them money. And the only way that you monetize that is to win. So you want to win as many as you can. So that means that if there's somebody whose tech is just as good as mine, their business model is just as good as mine, But they're half the size of us. They're really far away from profitability because of the fact that they only monetize half as often, because they're incurring the same level of expense, but not making as much money. So you have a 2% win rate in that scenario. But if the number of options go up to $6,000,000 and you win the exact same amount, now your win to look ratio has gone down meaningfully and you're struggling to be profitable. So if you're like most DSPs that over the last year have stayed flat, so the average is flat, That means that you're farther away from profitability even though you've had another year of water under the bridge. It's even worse or more pressure on the sell side. So if you're on the sell side, you used to run 200,000 auctions and a sell through rate of 20% was not uncommon 4 years ago. So that meant that you run the auction, the publisher sets the floor price, you monetize it 20% of the time. Their business model is the same in the sense that they only get paid in the case where they're the ones that monetize it. So in that 20% of the time, they were even back then struggling to be profitable. And you can see this on many of the publicly traded companies that were sell side focused like Rubicon. And then what happens over time is as competition happens, that by itself put pressure on the sell side margins, because the whole reason publishers did it was to create competition among them. But it also made it so that you're now having to run 600,000 auctions instead of 200,000 auctions and you're monetizing now just below 7% of them if you're using if you're monetizing exactly the same amount. And that makes it so your tech costs may have tripled or even more, because there aren't many economies of scale, especially on the sell side. So what do SSPs do as a result of this? It means that they're under a lot of cost pressure. So if you look at the curve, it basically so if you're an SSP, you are going to look at the curve. And what I mean by the curve is the demand curve, where you say, okay, what percentage of the wins come from The Trade Desk? Then after that, what percent come from Google's DBM? And then what percent come from Criteo? And you go down the list to see what percentage of the wins come from the top 5, top 6, top 7, because I don't want to send it to 100 potential bidders, because every time I federate to somebody new, it costs me more money. And you may say, oh, it's just a little request out to ask for a request for bid. But if they never monetize it, you're leaking data and you're repeating this process potentially millions of times a second. So when you do it that many times, the costs do add up, which is why our P and L has 30% of its costs inside machines. It is expensive to look at 7,000,000 ads per second. So as a result, SSPs make a choice to not send it to everybody. What that means is that if we have the ability to spend, let's say, dollars 100,000 for a given campaign in re messaging, if our competitor, who is half the size of us, who has the exact same tech as us, they copied it, it's not possible, but they have the exact same tech as us, Everything else is the same. It means that they will get fewer auctions. And therefore, instead of spending $100,000 in highly targeted data driven remessaging, they'll spend $50,000 So to be more concrete about it, right now in Ventura, California, there's where we're headquartered, there's 237 people in market for pizza, and they're all within a 7 mile radius of the pizza shop that's trying to do the advertising. And you shouldn't advertise outside of that 10 mile, 15 mile radius because nobody is going to drive 25 miles to go get pizza. I mean, our pizza inventor is not anything like pizza in New York. Nobody is going to drive for that. It's good pizza, but it's not New York pizza. So if you're the pizza owner, you want to advertise to all I forget what I said, 250 ish people. If you can only advertise to 125 of them, you're not going to sell as many pizzas. You're not going to stick with that tech for long. So what I would argue is that just by the nature of the landscape, it is nearly impossible for a new up and coming DSP to go raise money and be competitive against us. And we haven't even talked about the fact that we have a nearly a 10 year head start. And we also didn't talk about the fact that you have to march down Sand Hill Road to a bunch of people, excuse me, who lost money on most of the DSPs that they invested in and say, I'm a DSP, I'd like to raise $100,000,000 $200,000,000 $500,000,000 to go compete with companies like The Trade Desk, who are 10 years ahead of me. I know you lost money last time, but I can catch up and our business model is better. I don't believe there's a massive improvement on the business model that we've created nor on the technology that we've created, And I don't believe that company is fundable. Just need a drink. All right. So we're long term focused. I hope that we have made that clear in our interactions in the past. We're not optimizing to a quarter. That's not what we do. We think that the big opportunity is over the long term. I think that's important to point out because we have really desirable EBITDA margins and growth rates. But I don't want anybody here to think that we're optimizing to those EBITDA margins in the short term. We're not. We're trying to grow as fast as we can. That's a demonstration of our operating leverage. So we've been trying to win trust. So any of you that met with us on the IPO, some of you said to us, you're a technology company in advertising, are you crazy? Because ad tech was so hated and there were so many of those that did not do well. And what we said was, we believe we have a business model that will last for 100 years. We think we have the one business model that can really differentiate and separate from the pack. We're excited to come tell our story, and we want to make certain that we win trust. And so I just want to remind you of what we've done over the last year. So we launched our successful IPO. Our revenue increased 84% year over year. We were excited in Q3 to report that mobile video was up 3 50% year over year, which I said that mobile is the fastest growing category we've ever had. If we narrow that to something a little bit more fine grain, mobile video is the fastest we've ever had, ever will have probably. Our cohort growth was 81%. We recorded record revenue for the fiscal year. Our international grew 2x faster than the U. S. For the year. So just a reminder that in that global pie, about 2 thirds of the dollars are international. Mobile video grew by over 2 20%. Our data usage grew by 77% year over year in our Q1 report. We had forecasted that our EBITDA would potentially go negative and we came back and surprised everybody that we were at $6,300,000 positive. We also revealed for the first time publicly that we power 7 of the 10 largest brands in the world, and we opened offices in Paris and Madrid. In Q2, we talked about the most bullish number that we've arguably ever shared, which is connected TV inventory went up by 10x. And the reason why I think that's such a bullish number is because I again believe that TV is the most effective advertising ever in the sense that motion picture and images are what wins hearts and minds. And that that is real advertising, which is winning hearts and minds to your brand, to your product. It's not navigational clicks or anything like that. It's winning hearts and minds. And because TV can do that in 30 seconds, 60 second spots in a way that no other medium can, we think that's the most effective. Now if you can add to that both scale and targeting like it's never been seen, we think that's the biggest opportunity perhaps ever in advertising. And that 10x growth is just such an optimistic and exciting number for us. In Q2, we posted a 35% adjusted EBITDA margins. Our Hamburg office grew by 150%, Tokyo grew by 300% year over year and then, of course, mobile video continued, and we opened our Shanghai office. So what else have we done? Not financial results, but what else have we done since the IPO? I want to just take a minute to talk about some of the victories because there just haven't there hasn't been a forum for us to talk about it. So I just I want to take a few of the minutes that I have left to talk through those. So this year, we spent 30% of our engineering resources on a product that won't ship until next year. So all the things we just did, we did while siloing 30% of our engineering resources to work on something that will move the needle next year. And just to give you a little bit of a preview of what that is, some of you saw a demo before. We think we can give time back to our customers as well as make it more powerful. So a concept that we've talked a lot about in the past is something called expressiveness, which is you want to be able to express with as much grain and detail the consumer that you're after. So it used to be the very best you could do is say 25 to 45 year old females, those are my target or whatever demographic segment that was the best you could do in advertising, whether that's in print, whether that's in TV, whatever, radio, and then you pick the channel or the show based on which demographic it appealed to. You can do that with so much more grain now, but most technologies make it prohibitive to actually type that in. They make it too hard to enter that and that becomes the barrier. We think we've done a better job at that than anybody else, which is one of the many reasons why we've distanced ourselves from the competition. We did that with something called bid factors. But the separation between line items or the status quo of programmatic and bid factors is something like this. And the distance we think between bid factors and what we'll release next year is greater than what we've ever done before. So we're super excited to ship that product and give them half their time back and increase users' expressiveness. We also happen to change the business model of 1 of the largest cybersecurity firms pointed at digital advertising, which is a company called WideOps. I'm super proud of the partnership that we announced a few months ago. If you didn't see the press releases, I would encourage you to take a look and read through them to better understand what we've done with them. But basically, White Ops has pointed at stopping all fraud in digital ads. And as the Russian impact on our election and what's recently happening with Facebook points out, we need to do as an industry a better job of fighting that. The thing that's made that difficult is that many of the cybersecurity firms have had to invest so much money that they've struggled to figure out a revenue model that all the rest of us can afford. So most people haven't leveraged those assets the way that they should. And so what we convinced and this was only possible because we're the largest independent buyer in the world. We convinced the largest cyber security firm or we think at least the most credible to change their business model so that we can afford to layer their technology on every single thing we buy. And we've launched that this year, and we think that is a game changer. And we absolutely believe that we can in a meaningful way put companies and individuals who are committing fraud against advertisers and against publishers. So this isn't one side or the other against both of them that we can actually put them out of business. So we're excited about what we did there. We also there's a fair amount of misrepresentation in some of the auctions where somebody says there's something that they're not. So we launched a program where we're actually eligible to sell their product. So in other words, if you're The New York Times and you SSP 1 is eligible to sell your product and SSP 2 is not, we know that if we see an ad from New York Times on SSP 2, we can ignore it. So by going to publishers and encouraging them to just list all the authorized sellers of their product, we've convinced over 40% and by the way, we've done this in less than 3 months. We've convinced over 40% of the top 500 publishers that we spend on in programmatic to adopt this. We think we can get much closer to 100%, because there is no good reason for them not to do it. And as those that as the 40% rises to 60% 70%, the amount that publishers will lose by not doing it goes up. So it makes it so it's impossible for them not to do it without losing money. So we also rolled out a gold standard for SSPs on the total. And this was a byproduct of really SSPs coming to us saying, because you're the largest independent buyer and because header bidding is kicking my ass, I would love it if you just because this auction is available in 3 different I'm sorry, this impression is available in 3 different auctions. If you just always buy it from me, I'm back to 20% sell through rate. So then I'm in business, the other guys are out of business, just make me your favorite. And so instead of just picking arbitrarily or saying, oh, this is the nicer guy or these are the people that bought us that nice dinner, Instead of playing that way, we said, okay, we are willing to win all economics and performance is the same. We're willing to choose one over another. In fact, we have to. So given that we have to choose which one we're going to buy it through, let's create a standard by which we'll make that choice instead of picking arbitrarily or by some what we see as irrational means of doing it. So we created a list of all the things we want them to see. We want to see them do better. And then we gave them the list and said, this is the criteria by which you'll be graded on transparency, what are your fees, how do you pass data, do you enable us to be more effective? Like all the things that they can do to make us better, as well as to be more transparent are the things that we ask for. It makes the ecosystem better and it's going to make some companies win and some companies lose. We're accelerating that pace of change simply because we could. We partnered with more and more companies as ad tech consolidations rolled out. There's a bunch of them that I would love to talk about that I can't. And it's often because they don't want to talk about the fact that they have pivoted or like their tech efforts weren't as successful as they wanted them to be. But as the pressure has come on many of the players in tech, you can imagine that when you're spending as much money to operate your tech as we are, that other companies look at their tech costs and they say, what if we just leverage The Trade Desk? Isn't it more economically rational for us to use them instead of doing it on our own? And so we've had a number of companies who used to be competitors now be partners that we power today and super excited about that. And then, of course, just weeks ago, we launched a new product in connected TV, which we'll talk more about throughout the rest of the day. But given the size of opportunity here or in that space, we're super excited. So our goals for the next 5 years. So this is where I get bold. I make Chris nervous. So if you have any questions about these goals, please send them to chris. Toast. Just kidding. So here are our 5 biggest goals for the next 5 years. So the first is that connected TV and video will become our largest channel. This means that this would grow faster than any channel that we've seen to date. Because mobile has been so great, this is a really bold statement. Perhaps this is like manufacturing Teslas, which is if we come even close, this is a win. This is a win. Few Tesla shareholders in the audience. So but this will surpass mobile. And I think this is the biggest opportunity in front of our business. So the second is that China will become a top 3 market for The Trade Desk. This also is a really bold statement, because we've had some really aggressive growth in other parts of the world. So we think that the size of the opportunity or the amount of money that exists in China will make it so that even if we're just getting smaller percentages on the whole, we're still outpacing the growth that we've seen in other places. And never before have we gone to a market that big, right? Because we basically said, we're starting with number 1, then we're going to number 3, 4, 5. We skipped over somewhere along there, Russia and Brazil. But basically, we've gone down to number 14 ish. And even in Indonesia's case, at the time we entered, it was the 17th largest, but because its growth rate was 3 times faster than anybody else in the top 15, we went there earlier. But we never came back to any of those we skipped over that are so big, which is part of what makes this such a bold statement as well as a massive opportunity. So the third one, I have to contextualize for a second, because this one is a little bit technologically nuanced. So one of the things that people often ask us about is, well, what about the login advantage that Google and Facebook have? And I don't think it's quite the advantage that you might think it is. Now it is today, there is no question, it is an advantage, which is that Facebook has 1,300,000,000 people log in to Facebook every single day. Now that doesn't mean they're logging in to multiple devices. In fact, most of the traffic comes through on mobile. But they do over time tend to go on multiple devices. You've checked Facebook most likely on more than just your mobile phone, even though that's probably where you spend most of your time doing it, if you're the average consumer. But Snapchat, incidentally, not even making the top 10, because even though there are some really big players in there and those numbers are humongous. WhatsApp numbers are humongous. They look small relative to Facebook, but those are massive numbers. They all, to some extent, pale in comparison to Facebook and Google. So with that as context, our third goal for the next 5 years is that our ID footprint will be larger than any single company's login footprint. And you may say, how is that possible? You don't have that many people logging in to thetradedesk.com. And you're absolutely right. I never expect to have 1,300,000,000 people log in to thetradedesk.com. But keep in mind that 4,000,000,000 people have Internet access and that's growing super fast. So $1,300,000,000 out of $4,000,000,000 There's a big rest of the Internet. And a huge part of the argument that I've been making is that media is fragmenting. So we're super excited that we recently inked a deal with 1 of the largest telcos in the world to share their login data to help us put together a better ID footprint. We're super excited about the prospects in digital audio because lots of them have massive amounts of login data. Some of the largest dating sites in the world have massive amounts of data that rivals many of those social networks. In fact, many of those social networks are willing to monetize and some of them do. So what our plan is, is not to have 1,000,000,000 log into The Trade Desk. Our plan is to continue to tell an objective story the same way we have in every other aspect of our business and say that if we want to help monetize the Internet better, we have to work with the rest of media to do that. So we're super excited about the over dozen top 500 publishers that we now leverage parts of their login data to make ours better. So when you stitch all that together, it's actually not as hard as you might think to develop a footprint that represents more than $1,300,000,000 which again is only about 25% of the connected Internet. So number 4, we'll grow our data usage at twice the pace of our media spend. So I said earlier, this is not about new media versus old media. This is about data driven decisions versus guessing. I still think we're in our infancy in terms of the way that we use data and perhaps the best proof of that is the fact that I think we can grow data usage at a much faster clip than we have media. And our media growth has been pretty remarkable. So that has lots of positive impacts on our business. And then the last one, really the boldness in this one is in the implications. What happens today in most planning is you sit down and you talk to a media mix modeling company or you use sort of thumb in the wind technology to figure out how do I divide between digital, how I divide between TV. The level of sophistication that goes into this is not that good. And it doesn't mean that the mathematicians behind it aren't good, they are. The data they're working with is not good. What we do is we start with a data driven approach where we're look alike modeling against what's actually happening. There is no reason why everybody shouldn't be planning using digital first, because that's where all the data lives, including the fact that they buy media. In fact, the majority of their media is still elsewhere. About 65% of it is still bought not in digital formats. So there is an opportunity for us to change the way that planning is done by them starting with us. And when they start with us, it makes it easier for us to make the case why data driven decisioning is so much better, which accelerates not just digital and not just programmatic, but accelerates us. So because again, we you could describe us as a SaaS data company, this is where we're starting to exercise or leverage those muscles as a data company more than we are a media company. Of course, we're leveraging both, but the data is an asset that we have not spent enough time talking about. So a long term target, our adjusted EBITDA will be at about 40%. We think that we can be higher than 40%. And incidentally, to get there, our take rates will remain at our historical ranges, which are the 15% to 20%. So we as I mentioned, we started a business that was not riding the waves. We instead built our business towards end state. So our take rates have stayed in that range since inception. In fact, the overall trend has been up, although we don't expect that to continue to go up. We expect it to stay in that range, because we're deliberately making the choice to trade off some of the historic take rate in order to get loyalty and give discounts for quantity purchases. And so as that 95% client retention has gone up, We've managed to offset that with value added services. But those two things together, meaning the 95% retention rate often comes with quantity discounts because they keep growing over time. So as they keep growing over time, we're giving them more quantity discounts, but we're also giving them access to more and more products, which they're using, which keeps it at the level that it is, and we're confident that we can continue to do that. So there are so many things that I could say about our culture. We've never once asked anybody to fill out a Glassdoor rating. If you haven't been on Glassdoor, you can go read what people say about our company, people that work there. We're not perfect, but I think we're really, really good at developing a culture where people want to stay and where they want to build their careers. And it's really hard to talk to this in any adequate way. But maybe one number that I can share is just that every C level that was here when we filed our RS-one is here today. They've all made life changing amounts of money. They all could go do whatever they want. And we've all made the choice to stay together and keep doing what we're doing. And that's because we believe that at a $700,000,000,000 industry, we're 2% done. So just to summarize, we're a self-service omnichannel platform. We're in the business of decisioning. 95% of our spend comes through MSAs and is predictable. It's recurring and predictable. Our tech is measurably better at decisioning. We're objective because we align our interest with buyers. We have a massive TAM. In less than 10 years, I think it will be $1,000,000,000,000 We're best positioned to capture the massive connected TV and the international growth opportunities. And we've delivered on both growth and profitability. So please don't discount either of those. As you think of us as a growth company, you don't often have the level of profitability that we have. But just remember that we're still growing at a clip exponentially faster than our industry. And we've been profitable since 2013, I'll say 2012, really, but 2013. And we're not trying to monetize thetradedesk.com. We're not a destination. We're not Facebook. We're not Google. We're a play on the whole Internet. And if you believe like I do that media is going to continue to fragment, then advertisers need to get out of spending 1,000,000,000 to make consumers hate them. They have to be more deliberate and more data driven about their decisions, and that's the only way that they'll win. And as they do that, we'll continue to do better. Thanks. Now is our Founder and Chief Technology Officer, proud to introduce Dave Pickles. All right, is my mic on? Good. Okay. So Jeff talked a lot about the future. I'm going to start by talking about the past. So to understand our tech, you really have to understand where we came from in our history. So 2,008, I was at Microsoft. I met Jeff. We're coming up on our 10 year anniversary. And we I built an SSP or an exchange first. So I wasn't into ad tech, I was into scalable software. And while I was building this SSP, I started getting excited about all the possibilities we're creating. We can make it so you can pay whatever you want for every impression. And how powerful could it be, like if you could actually get that right. And we're kind of fantasizing what we're coding and everything's cool. And then it turned out, since everyone hated us at Microsoft because we were disrupting the business, I also had to build the first two bidders. Well, my team did. And so we built a couple of little really simple bidders and started getting more excited. And we decided we needed to get out of there and get into the space. But I'll get back to that later. But so when we built the exchange and we shipped it, I got the great opportunity in this context of seeing people bid. And so we plugged in all the existing bidders in the space, are a lot of the companies that like you know about now that are DSPs. And we were kind of shocked. We saw that most of the time they had one bid for an advertiser, sometimes they had 2. And I came to think that the 2 bids were like, I have data, I have no data. Like it's $2 or $0.65 And we were like we're almost like offended that like we made it we did I did a lot of work personally to make it so you could bid whatever you want on every single impression. And you're like, I'll take whatever you got to $2 And so I started trying to figure out why it was that way. And I got to know DSP 1.0 is what I described that whole class of companies. And there were a lot of companies that existed before RTB happened and kind of pivoted into it and a few new entrants too. But they all use this thing called decision tree, basically, variance of this. And so it's going to get a little data science y. It's not that bad. I'll make it as easy as possible. So the decision tree basically takes the world, this note here would be like the whole universe of ads, every possible ad you could ever see, and it starts breaking it up into finer and finer pieces. Because obviously everything is not worth the same price, I'm going to break it up into pieces. And so I might say, okay, let's have a different price for USA, Europe, Asia. That's a good start. Okay. Then the next tier down, well, obviously, like little rectangles have a different value than like full screen takeovers. So we're going to keep breaking it down as you go through variables. It seems like a very sensible way to attack this problem. And then once you get a few 1,000 impressions in one of these buckets, you can make a decision about it. Like you have to kind of fill this thing with data. At the beginning of the campaign, it doesn't know anything. And you basically like pour money into the top of it by buying at random. You just buy impressions and see what happens. And you start to see how this flushes out and what all the different values are. And then 7,000,000 times a second, you go look at every tree for every customer and it gives you the answer. That was the other reason it was real attractive. It's because like all I got to do is a look up. Everybody is afraid of too much processing power in real time. I've got 100 milliseconds to get back to the auction. I need to make this quick. So let's figure out everything ahead of time. And then in real time, I'm just looking it up, sending it back, looking up, sending it back. Turns out this has some major, major flaws that massively outweigh all of that. So if you look at the tree the other way, a tree is a set of line items. So I'm introducing you to our new friend for line item. He's very important in our world, which basically says that for this set of attributes for athletic apparel audience in the U. S. On those ad sizes, the bid is $4 the budget is $50,000 it's very explicit. And but then you find out that other formats have different prices. And it turns out you shouldn't have been paying $400 for the smaller ads, you should only be paying I'm sorry, dollars 4. You should only be paying $4 for the bigger ads. And so you start to discriminate and everything kind of still seems fine. But this is a silly example in terms of the variables that actually exist. So what if the Midwest is performing better, then you get into regional. And but I so I discovered one new fact in between the last slide and this slide, and I doubled the number of lines. So this design has a really unfortunate fact, is that every time I make a decision, the number of lines doubles. And that series gets very big, very fast. So if you look at a more real example, instead of like the elementary school example we had in the previous slide, like this is these are actually the cardinalities of these variables. This is how many of these things there are. There's tens of thousands of sites, there's actually way more than that. We're being generous here. We're keeping this small. Five frequencies per user, 4 different ad formats, 10 different creatives, all of the towns and geos, and that's even not if you're doing hyper local, that's just if you're doing kind of local ish. 500 site category, 24 hours a day, 7 days a week on and on. This is a massive understatement, but even just this math is 52 quadrillion permutations. These numbers get super big, super fast. And this ignores the biggest variable, which is audience. And the cardinality on audience is basically infinite. It's the infinite variety of human behavior. People do things you wouldn't believe and we see a lot of it because we're out there on the Internet. And we're so this is a gross oversimplification of the fact that a big difference. If you think about like the number of lines that you're going to have in an average campaign, that's going to be, I mean, max 10000, 50,000, say it's 100,000 lines. If it's 100,000 lines in a normal campaign, 52 quadrillion is 13 orders of magnitude more, 13 zeros more. I mean, it's just let's look at the number. This was going to come up. Yes, it's a big ass number. And that's for every single campaign. Hopefully, you've got more than one campaign running. Hopefully, you've got like thousands of customers and everybody's got this big set. And so this is like how specific you would be ideally, if you could actually value every impression individually and you knew every fact and you could put it all together, you'd want to have 52 quadrillion different bids. So if I'm a line item platform, I obviously am not going to store 52 quadrillion lines. Because as I told you before, what I've done is I've gone and figured out ahead of time what the bid is going to be and then I've stuck it in RAM somewhere in a server. So that when an impression arrives, I can look it up. But there is the lines are going to get real big, real fast. And if you've ever built a computer on dell.com, you know that like the basic price is the basic price. And as you start adding RAM, that machine, the PC gets line item platform has this vertical line item platform has this vertical scaling problem, where you actually can't store all of these lines in memory. Like if you wanted to do 52 quad Drilling lines, it actually isn't possible. It would be massive, massive amounts of data. So what you do is you limit the line count. So I said before, I got a little out of order. A lot of these systems have like a 10,000 line limit. So on any given campaign, they'll have no more than 10,000 individual bids. So they can only price differently for 10,000 specific subsets of the inventory. And what they do the rest of the time is bid on averages. And so the lines instead of being very, very specific kind of go over a swath of inventory and you bid average through it. And this leads to what I lovingly refer to as the spiral of mediocrity, which is you can't afford the 52 quadrillion combinations, even if you could find them. So even if you had this decision tree that could pound through and find all the permutations, even if you had enough money to spend $1,000 impressions in every one of those buckets of that learn budget that you're dumping into the front of this thing, Even if you could find it, you still can't action on it because you can't afford all the RAM. So you put in a line limit, you cut corners. And what cutting corners means is that you can't kind of all ties together that it just gets really, really hard to break out of the cycle, So it kind of all ties together that it just gets really, really hard to break out of the cycle of delivering average performance with the line item system. And this is why I only saw 2 bids at that time apparently they only had 2 lines and they hadn't started building new lines yet. If I waited around the rest of the week they'd probably had 40, but I didn't, I got out of there. So, yes, that was really our moment was watching those poor bid landscapes come through. And we just realized nobody else was thinking about it the way we were. So as we've been sitting there building and and building and thinking, everybody else had been operating a company in an entirely different business model. So even though they were existing companies when we started, we had an intellectual head start that was massive. And so we went about trying to solve it. And so there's a lot of differences between what we did and what came before. A lot of it was basically second mover advantage that like everybody else had their platforms. We got to look at all of it, start over and reimagine the whole thing, and like what would we really do now. And so it actually ended up being a really great thing. But I said, this tree, I get it, like it's nice to have all of the variables together and to be able to make that decision with all the variables in one place. But I think we could do better by instead of modeling as a tree, we'll model this as a linear equation with independent variables. Ad format, or I've got a slide here. The 300x250, the rectangles are worth twice as much as other sizes. And I know that fitness sites are worth 25% more. Midwest is a sweet spot for me. I can pay 50% more in the Midwest and still hit my ROI goals. And I don't do real well from 8 to 11 am for whatever reason. So those are basic facts that I can go and learn very quickly about a campaign. And to represent that same amount of logic, I need 4 lines effectively or 4 facts. You have to have 24 lines to represent those same 4 facts just because of the common combinatorial math. So it says 3 things. So we're able to describe the bit landscape that we want to create in a very compact memory space. So they I need 4, they need 24, that expands geometrically as you increase it, as you can imagine. So I'm able to have a compact memory that will be cost effective. I'm able to learn much more quickly, because I only need a 1,000 impressions on 300x250 to know whether or not that entire size is going to work for me instead of having that sort of facts sprinkled throughout a tree and all fragmented. And then this also produces insights that are meaningful to brands. Another huge problem with the line item is like you have a tree and you go back to the brand and you're like, ta da, here's a tree. And it means nothing to know anybody, especially if there's hundreds of thousands of nodes in there or something like that. And you can color them green and yellow and red for good and average and bad. Problem is most of the trees yellow. And so then you're kind of trying to talk your way through it. This we can go back and say like oftentimes we'll go back to a brand and say like, I know you thought that was your demographic, but this is your demographic or this is your audience or did you know that like nobody's driving more than 10 miles to your brick and mortar locations or whatever it is, those individual discrete facts just pop out of this thing. This is the toughest slide, I apologize. So I'm going to try. So the effect this is the effect of what we do versus what a line item system does as seen from the sell side. So on the X axis, imagine just a bunch of impressions flying by random impressions. This is fake data. This is just to kind of illustrate the point. And say the gray line, the gray line there is the absolute perfect price to pay, the platonic ideal price to pay for that impression, which is that I'm taking advantage of every opportunity I can without ever overpaying. So if I bid too low, I'm going to miss opportunities. Somebody else is going to buy it, and I'm going to not scale and all the things that go with that. But if I overpay, I'm wasting money and that is basically just going to come straight out of the ROI that I'm delivering to the client. I could have got it for a buck, I paid a buck.50 dollars Simple math, simple math. So what the line item system does, it does a very good job of occasionally, because it can only have 10,000 different combinations of occasionally being very accurate. And it did a fantastic job here, it found that ideal. Like they found the 16 variables, the magical 16 variables that are going to produce that one result. But then on the rest of this, you're sitting up in a higher level line that was just sort of like average. And so the distance between the gray and the red, the sum of the distance between the gray and the red, it's pretty large. The big factor will I never in this chart gave it credit for being exact. It isn't. There's a little bit of imprecision built into not having the dependence between the variables. But what it does is follow the curve. It's very close. And so if you take that same sum of the distance between the blue and the gray, you're going to get something much smaller. And that basically is ROI. So this is something that Brian is going to give you lots of data about, because this is just us driving performance. That's how you see this. This just drives fantastic CPAs and great results. All right. So the second impact other than campaign performance, campaign performance is super important, but so is cost efficiency. So like from day 1, I had a theory that because you can only take so many if a dollar is leaving an advertiser and ending up at a publisher, there's only so many dollars we can all take out of that and have make any sense. So I always thought we needed to be very cost efficient. And I knew every dollar that I spent, I had to create at least that much value. So the more cost you have, the higher the bar is to hit that same ROI at the end of it. So we wanted to be very cost efficient. So that was very much in my mind. And so I looked at it and I said, okay, the line item systems made this big bet that they can scale RAM. They said, I'm going to scale RAM as high and as far as I can. And it's a very tragic bet in a lot of ways, because it's just it's difficult and it's very expensive. I said, let's go scale CPUs. Let's push hard on CPUs. Let's do math in real time. Let's just not be afraid of like, I think, I got 8 milliseconds and I got a good processor, I can do a fair amount of thinking in real time. And everybody else is really afraid of that. So in the way that RAM terrified me, CPU complexity terrified our competitors, and they didn't want to touch it. But we said, let's go for it. And so just the scaling ramp thing, if you're looking at 7,000,000 requests a second, you need basically a fixed number of servers, unless you're doing something really different with the CPU. So you need a server can handle, say, 20,000 queries per second. So I need to just buy a big pharma servers. If you need a normal amount of RAM, a regular server is like $2 If you need to max out the RAM in that same server, you're talking $10,000 $12,000 It's a really, really big difference in price. But you still need just as many, unless you're creating a massive amount of efficiency. So the cost is just really, really different between needing a sort of a custom high end scary high memory stack versus a very commodity simple stack. Like we have we get hardware from multiple vendors from like anybody. We're like, you got some servers, we'll throw a bidder on it. It's not a big deal. Like it's very, very helpful. And so those two things got us our dream, which was being able to de average and be profitable. Those are the 2 things we were thinking about. We wanted to be able de averaging being very specific about what we're paying for the impressions, while being able to turn a profit, which we were focused on very, very early based on all of our prior experiences of running out of money. Just as important, we got something that was flexible and extensible. So this is almost just like a side benefit. It wasn't that it's super planned on this, but if you have the line item system or anything like that, that's built around a very specific kind of machine learning, It's a tightly coupled architecture, all the parts just fit together. There's a learning piece, there's a reporting piece, there's a bidding piece and they all kind of speak the same language and it goes around and around and around. And so if you want to change anything, you have to change all these different parts in lockstep, and it's pretty difficult. Everything's got to be matching. If you want to add a new feature that you're evaluating in your tree, even though that will explode your tree and you probably never want to do it, it's going to take a lot of work versus us, like we're doing math in real time. If I want to try different logic, I code it up and I ship it. And even under socks, we code it up and we ship it. We got a nice system built for that. And I can put it on to 1 bidder, and I can see what it does. And I can iterate really, really quickly. And so we've been able to try lots of experiments over time. And you really can't overstate the value of consistently shipping improvements to software over a long period of time. So every single week we're shipping iteration, iteration and iteration and we've developed a lot of really cool technology in the process. So you might say, sounds great, Dave. This is very simple. Why doesn't everybody do it? And the two reasons are, it's really hard and it's really expensive. Even with our cost savings, it's very expensive. And getting that CPU efficiency where it is today was a very long and windy road. If you look at this is the growth of QPS over time from 2015, so this doesn't go back all the way, but use your imagination. When we started the company, there were about 40,000 requests a second in the whole universe. And so like 5 guys $1,000 in an AWS account, you can start doing some bidding. It wasn't that big of a deal. And so since we started at that time, we've been able to build and test and iterate and refine as the growth occurred. And along the way, at a bunch of different points, we've had massive stair step improvements in our cost to serve through what I call being clever. And as this grew, I always told our team, okay, guys, I know it's growing. We have to buy servers. Let's buy servers twice and be clever once. For every two times I buy servers, you got to get me a reduction. And we've had major improvements, things I would have never thought of. So it's a huge barrier to entry to even go through that process at all. So even if I give you credit for going through that process and getting everything figured out, the amount of money you're going to earn in the time is gigantic, which Jeff touched on too. Power and scale. So this is something Jeff talked about. A couple of my next slides are a little redundant with what Jeff talked about, but it's a different view of it. So this is a visual representation that I drew for Chris one time and he really liked about what happens with the QPS. And so you've got suppliers and direct inventory on top. You've got the Trade Desk and other big DSPs. Every SSP sends us every impression, every single time. Because to an SSP, a DSP is 2 things. It's a bunch of cost when I have to send something to you and it's maybe some revenue if you happen to buy it. And the ratio of those two things depends determines whether or not I like you or hate you as a partner. Simple. So what they've done is the top 2 DSPs get everything. Next 2 DSPs, they started putting some throttling. In these cases, they're probably being pretty smart about it. They're probably saying like, okay, you don't buy anything in Germany because you're not in that market yet, let's turn it off. But there is also just sort of a little bit of a randomization. And then as you get farther down, they're basically like, I don't know anything about your business and I don't care. You don't spend enough with me. And so they just basically say, every 5th impression you get to see in order to keep those costs in line. And at the bottom of it, that's they're literally sending a trickle just in case something changes and that DSP starts providing them some yield. Their yield curves are just like super steep. So they see almost nothing at the end. Or another way, or you said, why does it matter? So the good stuff for your campaign, the good impressions are randomly sprinkled out through all of the inventory. You're looking for needles and haystacks. And so visually, you can see like the say the blue is the best impressions, the yellows are the good ones and the grays are like bad impressions from my campaign. We see everything, we can go pick all the good stuff. If I see half as much, I have basically 2 choices. I can either spend less money and make it just as good, so I can't hit the budget or I can spend the same amount, hit the budget that they wanted to hit and my performance is going to tank because I'm going to have to include a bunch of great dots. Brian has the slide too, I'm stealing a little of your thunder, sorry. But it is a different point. To the point where it gets real, real thin down at the bottom. And this is one of the reasons like we've proven this again and again. We just win and win and win against smaller DSPs, a smaller DSP can't compete. So this all ties together because in order to compete, you have to look at everything. So if you're a startup DSP, day 1, assuming you could trick SSP into sending it to you, maybe you've got some leverage, like you get them to send you everything. You're paying this gigantic brick of sunk cost from day 1 to look at $7,000,000 QPS or if it's USA only, it's 2.75 or whatever it is. I mean, it's still huge QPS even if you do some kind of geo thing. And over the course of your build, you're going to burn a tremendous amount of cash. You have no revenue and you have to look at everything in order to compete. Or we just take every campaign from you. All right. So this is a vintage slide that we had to go dig up, because this goes all the way back, all the way back to the beginning of our company. So at the same time we founded the company, I went around and looked for a suite of APIs that would do what I wanted. I looked for some way to adopt a platform that would let us grow our business quickly. A lot of people were like, we really wanted to get it going and we just knew we wanted to be at that point, we knew we wanted to build a business on the buy side and decisioning. We didn't know what it was going to look like yet in terms of exactly what we're going to do. And I found that the APIs that were available were there was no way for me to create proprietary Vantage on top of what that API was providing me. There were APIs where I could say, here is $100,000 tell me what happens. But that was about the best you could do in terms of being able to build on somebody else's platform. There was no way to add our special sauce, especially at that time. But even now, it really is true. And so what we did is from early on, we built the stack in layers. So we said that and this is all the functions that we thought of at that time. This isn't even current. There's a lot more stuff you have to do now. There are some things the platform is going to need to do, like looking at 7,000,000 QPS and responding, doing ad serving, basic log aggregation. There's these nuts and bolts, big data operations that they're going to do. But as you go up the stack, there's more and more things like attribution that an advanced customer might want to do or doing their own optimization or doing their DCO or insights or all these other things. There's all these other functions that somebody else could do. So we started building APIs that would let other companies do this, because I felt like DSP number 18 that I was picking on earlier for being in a tough spot, because he can't compete with us unless he looks at all the QPS. He can't get all the QPS and he doesn't have enough money. It's like it's a really tough situation. But if there's some way that you can take whatever that part of your business was viable, and it probably is, they probably some kind of a good idea and implement it over APIs, You can make it work. If I take all that cost out and I take more important than the cost, I get rid of all the work of rebuilding everything that we've already built. So whatever engineering resources you can afford are 100% working on special sauce, creating proprietary advantage full time. We have a lot of really successful clients on top of our APIs now. And that's great. So takeaways, expressive bidding, valuation, you can tell I'm obsessed with it. It drives ROI. We've proven it over and over again. We've basically built this whole company by winning campaigns on a CPA head to head. Every step of the way through is a lot of work. The cost effectiveness ensures profitability and increasingly ensures leverage. You've seen that in a lot of numbers recently with the accidental EBITDA, like we have a lot of fantastic operating leverage now. We have the best feature set, the best APIs. We ship product every week even under socks. Actually more often than weekly, we sometimes ship intra week. But we're on a regular cadence of shipping real product, not just break fix, it's real product. And really important one, there's 8 years of trust between our customers and our product team. We're in a really good cycle of listening and reacting to the customers. They really feel like if they talk to us, they're going to get what they want. We're not going to go take their customers. We're a true partner to these companies and they rely on us and know that we're going to come through for them. So even if somebody else has And then only a massive scale platform can win. I And then only a massive scale platform can win and the players are already in the game, just for the financial reasons that we talked about, just the ability to raise that money. So we're in a great spot. Thanks for listening. Thank you very much. Dave Pickles, Chief Technology Officer. We're going to take we're running just over 5 minutes late, so we're going to take about a 5 minute, maybe 6 minute break and come back for Rob Perdue, our Chief Operating Officer. Thanks. Thanks, everyone, for coming today. And I know it's a long day, and hopefully we're continuing to make this informational and sort of show what we believe in and what we're trying to build here. So what I'm going to focus on is our path to future growth. And where I'm going to focus really is on where some of the key highlights and where we think about investing in terms of our operations and then also where we invest internationally to give us the highest chance to continue that the growth that we've seen today. And I guess my bold statement to start the presentation is, I believe we believe that we will grow faster than our industry for as far as I can see, as far as we can see into the future. So I hope that this part of the presentation will put some ballast underneath that statement and then we'll come back to that towards the end. So just to level set, Jeff touched on it briefly, but just to go one click down, reminder what we're after, right? So there's a $650,000,000,000 global advertising industry heading to $1,000,000,000,000 over the next 10 years or so. And we think and we believe that only about $15,000,000,000 of that $650,000,000,000 is currently in decision programmatic. So that's just about to a little bit over 2%. So sometimes people say, hey, you guys have been at this rate years, you've gone public, like isn't it haven't you had your run? We say, hey, there's 98% of the transition that we started this business still to go. So 8 years in, we still have 98% of our work still to go. And that's the thing that drives all of us. So I hope that you guys take that away. We do believe it all shifts. And just a quote from a little over a year ago, Brian Lesser ran Group M. He said all media will ultimately become programmatic, which we believe you may know he recently left WPP to take the job at AT and T to help that AT and T Time Warner when it consummates make that a programmatic business. So he's sitting in as key So before So before we jump in, let's just take a quick sidebar into where are we today? Where are we as a business? For some of you, this is an update. For some of you in the room that are new or this may be new. So 600 people, almost 30% of them are outside the U. S. Jeff talked about the international or Asian opportunity. We're definitely investing ahead in terms of people and integrations and all those things. Over 600 clients running nearly 30,000 advertisers on our platform today with over 100,000 campaigns. I checked when I came in this morning, it's meaningfully over 100,000 individual campaigns on our platform today. And our agency partners and their brands, they're consolidating more and more on our platform. So an average client on our platform today is spending on 5.2 of the 9 channels that we offer. And so a year ago when we went public, that number was under 4. So we've had almost a 20% increase in terms of the channels people are spending on, say, nothing of just the dollar growth. So people are consolidating on a fewer platforms and making bigger investments. So where are we in terms of driving that omnichannel product adoption? Clients spend more with us. Right now, today, video and mobile are the majority of the spend on our platform. 3, 4 years ago, we were primarily a display or desktop display business. That's not true anymore. Desktop display is less than 35% of the impressions and spend on our platform, even though desktop display on our platform is outgrowing the industry average. So even that slower growth channel continues to grow on our platform faster than the industry. But we're really a video and mobile business at this point. Other emerging channels like audio and native are growing at 100 of tens of thousands of percentage points. And of course, that's off of a low base. But the transition into those channels is faster than, frankly, video and mobile when they got started. So we're seeing a lot of growth everywhere we look, particularly in mobile and video. We run-in every industry vertical you think of. So our platform works with any advertising strategy, any kind of campaign, any kind of goal. We really work for any model in the world. So just a quick update year over year, roughly the same. Our business delivers strong operating leverage. You've heard Jeff talk about it. You heard Dave talk about how much we're focused on from a technology perspective to make that happen. And then our business model is also very focused on driving that operating leverage. We think of our we're a platform and platform businesses are not people intensive. So when you compare and for the first time, I think, in doing these, I haven't put up a revenue per employee slide. But if you were to compare us to any of the public SaaS companies of similar size and growth rates, I think our revenue per employee stacks up, I think at the top of the list. So 2 of the core tenants to that is 95% of our business comes from MSAs or what we think of as long term contracts that are auto renewed. So we're not out fighting for insertion orders, right, that campaign by campaign basis. We sign a contract or a deal with an agency, and then we're essentially always on. It doesn't mean we're not talking to them about what they're going to do next quarter or next year, but we're not fighting for IOs. And then the second is 95% of the campaigns on our platform are self-service. And I'll talk a little bit later in this presentation about how we help our clients get to self-service. And that's important because in the world of planning and buying media, right, that's what media agencies do. The only way they can do that is to learn how to push the buttons, optimize in platforms like ours. So that's the core of what they do. If they don't make this a core part of their business to be self-service, they can't serve their clients well. And so we think we've done a good job at helping them do that. So you tie all that together, this is sort of the punch line slide, 95% retention rate for 14 straight quarters. We continue to grow our clients at a very high rate. And so out of those things in the platform business, we've generated very high operating leverage on a very consistent basis. And we continue to think that the business model, the operating model that I tried to just describe through statistics is the way that we'll continue to grow in the future. So now I just want to focus a little bit on where we're focused to drive growth as we think about the future. So first thing is, Trade Desk platform, we have a marketplace, right? So there if any of you have ever looked at a LumiScape and I assume you've suffered some brain damage at some point in looking at those, hard to figure out. There are lots and lots, hundreds in fact of point solutions in our industry. And so we think about it as our job as being a platform. Our job is to go out and roll up all of those, integrate all of those point solutions so that our clients can then use the best of them, use the ones that work for them. So we offer all of that. You could think of us as sort of an online supermarket. And we offer all of those things. So as our clients walk down the aisle, they put take things off the shelf, put it in their shopping cart, if you will, and that those things make campaigns work better. And then in our reporting, we're aligned, clients can measure how those point solutions worked. And if they didn't work, they turn them off and try something else. So that's a very big part of our marketplace approach. And as we add more things, that adds more value to campaigns and to marketing performance. And then, of course, that makes clients come back and spend more with us. And so here's some examples of the kind of point solutions I'm talking about. Lots of categories here. I think there are 15 categories. These are things that we sell. They're all in our platform. We sell them all day, every day. These add value to clients' campaigns. You'll see in green, in almost 50% of these, we have a Trade Desk version of that product. So it may be a little bit lighter feature set, all the core functionality is there and it adds value, right? So they if it works, clients will spend with it. They will use it. And we use it as a one, as a way to keep our point solutions honest, like keep them competitive. Our pricing for our own stuff is a little bit cheaper often. Maybe you can think of it as sort of like Amazon Basics, some of the product set that Amazon rolls out that's a little bit cheaper. Maybe you call it the generic brand. We do a lot of that as well and our clients put those things in their shopping cart all day every day. But again, to be clear, we're a platform, we're a marketplace, we're not pushing anything other than what works best for the client. So we want to offer everything that's available in the industry and then let the clients decide what works for them. So we innovate fast. So this industry is moving super fast. If you heard took away from nothing else from Jeff's just the amount of connected TV, the amount of change in mobile, the amount of international opportunity, it's really massive. And so we've got to innovate fast, not only to keep up, but also to continue to win. So we believe that in a rapidly changing market, any market, frankly, innovation speed matters. And Dave called it out. We do 45 product releases a year. We literally are burying our competition with our innovation speed. So there are times when competitor rolls something out, they're working with us and they come to us and say, hey, that's actually a cool feature. I wish you guys had that. Because we innovate so fast and because we've rolled out so much product over the course of 8 years, we can go to them and say, yes, you're right, we don't have that, but give me 2 months. If you give me 2 months, we'll go get this built and you'll have what you need in addition to everything you love about The Trade Desk platform. And because we've earned that trust, because we've actually built product, we'll say, yes, I can wait 2 months, I can wait a quarter. I'm not going to change my business with The Trade Desk because you guys have done so much for me. So we actually win business all the time on our ability to innovate, on our ability to listen to clients and then build to them. And that really speaks to the next slide. How we think about product priorities, channel expansion, global expansion platform and then API extensibility. But underneath that, I think the other thing that doesn't come out often in rooms like this is, I think the feedback loop from our clients back to product and engineering to building in that fast innovation cycle and getting it back out to clients, I think is the best I've seen in certainly in Ad Tech and maybe in software. Our feedback loop is super tight. We listen really well and we build the right stuff And we do that in a way that clients feel and know that we're building on their behalf or that they're being listened to. And so that's been a really big differentiator for us as well. So thinking about going to market, so from a biz dev more from a business development perspective. So agencies are our core clients. Paul, I think I'm getting sick. I think I got it from you. Agencies are our core clients. But inside of agencies, even today, as much as we've grown, as much as you've heard about programmatic growing fast, there are still literally 1,000,000,000 and 1,000,000,000 of digital dollars. So digital advertising spend that's not in programmatic today. And a lot of those dollars are either direct sold, so they'll go directly to a publisher. They won't use the data driven decision making power that both Jeff and Dave described. They'll just hope, right? They're paying a fixed price and they hope they get the good stuff. There are 1,000,000,000 of dollars that does not perform as well and isn't tied together with the rest of the media plan. So a big part of what our BD team is doing today is helping to break down those silos. So some of those dollars sit with the traditional agency or digital team. Some of them sit with some of those decisions are made up higher at the senior levels. So we're trying to break down the silos within the agency and help them bring all this together. And what we are seeing more and more is that programmatic budgets are growing. So coming out of there are even mobile teams, believe it or not, inside of some agencies. Mobile was coming into just a programmatic buy. There are TV or connected TV teams. Those are coming into programmatic teams. So the budgets that we core go after are growing all the time, but we've got to go talk to those other folks as well and help them understand the value of programmatic and what they do and help break those things down. And so we spent a lot of time in agency trying to connect those dots and help, frankly, agencies do better by their brands. Another thing we're doing is we are moving up the stack. I talked about that briefly, but we're going up to different parts of the agency world. But they care about different things, right? So when they think about what do they care about from a marketing perspective, they have different goals, They have differing value propositions. And so we have to talk differently, frankly. We've talked a lot about some of the things that Dave described when we're going to talk to agency trading desks. We're talking about our feature superiority, why the architecture produces better ROI, differentiation, our business team support, all that stuff matters a lot, right? But when you start to go up to the digital agency, they may not be putting their hands on keyboards, but they may be making spending decisions. So they care about things like KPIs and analytics. They want more insights to bring back to their brands. And so we're spending a lot more time honing and talking about those things to digital agencies. And at the same time, the agencies are starting to walk us into brands to help them sell through the value of programmatic, the ROI that we can drive, the ability to get better insights in who their customers are and who their prospective customers are, perhaps even more importantly. So and they care about completely different things as well. They'll probably never touch the platform, but they want to understand how we can help them reach their marketing objectives in a more efficient way and more effective way. And that leads to I know people ask this all the time. We are getting closer to advertisers. I know Brian will go way more in-depth in the next session on this. But we're going alongside our agency partner. So they're asking us to come in oftentimes as the programmatic expert, come in to explain the value and educate how things work and how we can help brands do marketing better. So we are doing this. But what the model, it's not that brands are bringing this in house. There's not an agency team working on a global brand. They might have 200 people just working on that one brand around the world. There are very, very few, if any, brands that are bringing or hiring 200 people just to run programmatic, right? So think about a large global organization's marketing department, it might be 200 people. They're not bringing in 200 people to run programmatic. What they are doing is hiring a programmatic lead. As programmatic becomes a bigger part of their marketing budgets, they need to understand it more. They need someone to educate them how it works. They need to understand how to put the data driven insights, the power of optimization and reporting, all the things that we do in one place, how they can make that work through all the business units, all the brand owners inside of a big company, that's what they're doing. So they do hire 1 person or 3 people, but their job is really to get closer to us and understand how this can help them make marketing better, and then also frankly manage their agencies and sometimes in a different way. So when we win a client, this is our typical support model. Every one of our clients gets a version of this and I won't go into it in deep at all. Frankly, it's really just a preview. We've got a great client panel where some of our business facing team are going to be on stage, and I hope that they'll really bring this to life later this afternoon. But really what they're focusing on is when they go through this client launch process, it's really about empowering those clients. So it's going back to the agency and saying, hey, if you don't figure out how to do this in this new world where programmatic is just the way you buy media, it's the only way, it's the most effective way possible to run advertising. You don't want to outsource. You actually don't want to give us that money. You don't want us to push the buttons. You don't want us to get the insights. You don't want us to go talk about all the stuff we've learned to brands. That's your job. That's what you've been good at for 50 years. Now it's just a matter of doing that same thing inside of machines, right, and making use of automation and the ability to use data to make campaigns work better. So our team focuses on this over and over again and in a core in one word, it's really about empowering the agencies to make their job better. And if they do their job better, which means better marketing performance and more insights, of course, more spend flows to us. And that really is the cycle that happens over and over again. I think I covered this on the last slide. But in the end, better results mean more ad spend on our channel. So I'm going to try to do a case study. I'm not going to do it justice because we don't have a ton of time. But this is an example of a real client that we started working with almost 4 years ago in 2013. And I've got a design of all the sort of high level marketing funnel branding, prospecting, retargeting, all of the channels that we work on. And so they were trying to get prospective customers, new customers onto their product set online, right? So Q1 of 2013, we got our first test, dollars 15,000 tiny little budget in desktop display and it was prospecting. They said, hey, we work with Google, but if you can help us find new customers, hey, we'll give you a chance. And that's usually how we start our relationships. So $15,000 prospecting, we found them a bunch of new customers. We talked them things about customers they didn't have that they could find. And by Q2, we've done a great job of finding new customers for them and helping them really target in a more specific way that Dave described to get more customers than they expected. And so they expanded that those budgets into the mobile optimized web. And you can see as spend starts to grow just a little bit. By Q3 of 2014, so a year later, we had rolled out our version 2 of our video product, really robust and saw the wave of video spend coming into programmatic. We actually didn't have to sell them. They trusted us enough by then. They said we want to be one of the early early adopters. Please deal us in. We want to get on video and want you to help us prospect through video. So that's a great example of where we build trust and where they start to adopt more product for us. So that started working. And again, it was around prospecting, help us find new customers, just expand the reach where we go prospect for new customers. Then in Q1 2015, things started to get really interesting. We expanded into branding. So we'd always had their prospecting budget, but then there's all that awareness, right? There's all those customers that aren't even thinking about signing up for this product or service. There's a big amount of advertising budgets in the world. They dealt us into that. And they said, hey, that video product that we tested with you guys in Q4, it's phenomenal. Let's do branding with you guys now. So that moved us into the next part of their spend. So then we rolled out Cross Device in middle summer 2016 and they said, hey, we're bought in. We know everything you guys roll out is great. Let's test that. Again, more prospecting and we got more budget. Then you roll into Native. We rolled out Native, at least a bigger version of native in late 2016. So continuing to scale, they rolled on to yet another channel and continue to spend more with us. And then you go into 2017, we have big audio and connected TV releases, and they said, hey, we want to add prospecting in those channels because we know there are prospective customers that consume media and need to be advertised to there. So let's prospect there. And then the latest and big win just in Q3 of this year, they finally came to us and said, okay, we'll give you all the retargeting. You do everything else for us. You've proven that you can bring awareness. You can prospect new customers for us. And then let's go retarget. And there are big budgets there. And so now we've really captured a really large part of their programmatic spend, almost $2,000,000 a month. But to be clear, they spend tens of 1,000,000 of dollars in digital advertising. So it's still early for those guys, but that's what a process looks like when we just march quarter after quarter rollout new product, win trust, deliver marketing performance and clients move spend over. So I hope that brings that to life. So let's turn to the international opportunity. So these are the top 20 ad markets worldwide in green, the ones that will be in 2019. So I've included Indonesia because every projection says we'll be in the top 20. Well, today, 88% of the Trade Desk spend is in North America, only 12% outside the U. S. It was about 8% last year, so growing fast. But there's only 33% of all global advertising, dollars 650,000,000,000 Only a third of that is in North America. Two thirds of that is outside the U. S. And most of the top 20 markets outside of North America are in the international markets. And the international markets are growing 2 or 3 times faster for us than our North American markets. So when we talk about going global and why we've invested internationally, you can see where we've invested and why we've invested. We've been very rigorous about picking the markets where we think are going to be the biggest spenders, the biggest adopters of programmatic, the biggest opportunities. And we think we've gotten in ahead of a lot of the growth. We're on the ground. We've built trust. And we're ready as those markets continue to explode in terms of programmatic spend. The one thing I want to call out is we've been in Singapore since 2013. And you may say, great, that's a 1st world market, but small. But within Singapore's orbit is Thailand, top 20 market India, top 20 market Indonesia, top 20 market Philippines, if it's not a top 20 market, it will be soon. We do we get spend significant spend out of all of those markets out of Vietnam from Singapore. So we'll never be in 100 countries and we don't need to be. Where we need to be is in the key media markets where spend media spend is aggregated and we'll go get all the spend from the countries around that. And we've done that in Singapore, in London, in Australia are some of the markets where we get the most leverage. In the U. S, you'll also notice Mexico and Canada. We've been going up to Canada for 5 years. We've got a very high 8 figure business sitting in Canada today, even though we've never had one person on the ground. We've been going into Mexico for about a year now, and we're developing business every week. So again, we don't have to be on the ground. You shouldn't look at where the flags are of The Trade Desk to say that's where the spend is. On an average day, we look at bids or look at potential ad impressions from 195 countries around the world. Of course, not all of them are as valuable as the other, but we buy from almost 90 countries around the world. So we've been Dave's been globally deployed for 5 years. So as we move into new markets, there's nothing really new to build. We add servers and occasionally a new server farm. So let me talk about how we build a market or how we build a spend. It really is just a process. So as I just described, we often earn revenue before we ever enter a market. So France and Spain are 2 markets we entered this year. We've been generating revenue and earning trust from the local teams on the ground for 3 years before we ever entered either one of those countries. We had a good base of spend. We had committed clients who understand our platform and understand our value proposition for them. And then over time, it gets to a place where it just makes sense to go invest. And so what does it mean to invest? We go on the ground, it's a pretty small team, 3 or 4 people. We've already got the server infrastructure, the network is already operating, so there's nothing big to do. And it's like a WeWorks office. So we get asked a lot of times in rooms like this like, gosh, isn't it aren't you going to sink a ton of money into all of these new markets? It's really not that big of a deal from an initial investment standpoint. And then we have a philosophy of pay as you go. So as we grow and as we earn more spend, we'll invest more in the markets. And so here's just a couple of examples. This is the growth of North America, real live but not to scale. So at about 36 months, we hit a real inflection point and spend really, really took off. Here's what London looked like. So it took slightly longer than 36 months, but you actually notice the bend of the curve is going up much more aggressively. And that's because so much spend in Europe is aggregated in London. And as clients move to The Trade Desk, we really got a lot of leverage and a lot more clients out of that. Here's what Singapore looks like. So they not only do they not only did the inflection point happen earlier, but it's actually bending up way more aggressively. And it does speak to that comment that they get spend from Thailand and India and Vietnam and Philippines, a lot of the fastest growing middle class markets in the world. And so when we look at Singapore, I think that that's the harbinger for the markets we entered in France and in Spain and in Indonesia earlier this year. We know how to do this. We've done this every time. But as you think about us when we enter a new market, it takes about 3 years, right, before we really get spending. We get spend from the beginning. It's not like we're going in and earning or starting from 0. But it does take a little while to build momentum and it's around that 36 month time where things really take off. So that's the way I would think about as we enter new markets. Now China is a little bit different, right? So Jeff talked a little bit about this, 2nd biggest market in the world, it's massive, massive opportunity, a little bit different strategy. In some ways, it has its own ecosystem, in that a lot of the publishers are different. But Dave put up part of this chart early. We're now at about 7,500,000 ads per second that we're looking at globally. Well, even today, China is only 100,000 of that. So when we were at 100,000 ads per second in the U. S, I think it was late 2010, maybe early 2011. So when we look at China's programmatic industry today, we think it's 6 years behind where the U. S. Was in terms of scale and adoption. And so when we say, hey, we're there, it's huge, but it's early. And so sometimes people say, how fast is that going to go? Well, we think it's going to take a few years. We don't think it's going to take 7 years like it took the U. S. We actually think programmatic will go faster in China, but it's probably a 4 or 5 year growth rate than it is 7 years simply because we're further ahead. We've got teams on the ground today. We've got product and engineering teams who are doing all the hard work of integrating all the big publishers over there, building all the sort of ballast, and we're building trust with our agency partners that we work with everywhere else in the world. We're just doing it again in China. So it's really just a process, and it's just a larger market. So now I'm going to try to tie all these various things up. So how we look at our business from a financial perspective, and I know Paul will put a much finer point on this, but from an operational perspective. I want to talk about future growth. We think again, I think we'll grow faster than the rest of the market. And so I've talked about a fair number of business metrics during this presentation. I just want to touch on 1 or 2 more, and I'll tie it all together. So first, when you look forward, programmatic globally is expected to grow about 21% CAGR over the next 4 years. So we think of that as the base. So we're going to grow faster than that. I've said that a few times. And how do we get there? Well, 95% client retention, which I touched on earlier, we didn't talk about this, but our gross spend expansion, so our existing clients by themselves are growing over 50% this year. They grew 71% last year, 50% this year. Our new business wins are particularly strong this year. And then you think about all these emerging channels, there's almost $300,000,000,000 of advertising spend that's flowing into or towards programmatic today. That's barely programmatic. We're super early in all of those, and I think every single one of us has touched on these at some point this today. So when I think personally with Paul, with Jeff about how we're going to grow faster than the market, what are the vectors, where are the places that we invest, it really comes down to a formula like this. It's our base business, programmatic industry growth is about 21%. Our retention rate is 95%, so much higher than everyone else. Our same store sales growth is over 50%. Our new business is somewhere between 6% 10% of our total spend every year. So you can do the math and get a sense of how big that is. We've got tens, if not 100 of 1,000,000,000 of dollars of new advertising dollars coming in through new channels. That's really just fresh opportunity for us. And then you look at international growth, it's growing 2 or 3 times faster than our U. S. Business. So those are all the vectors that we think if I'm wrong on any of those or any one of those changes slightly, there's just no way that programmatic industry growth is where we end up. We have so many opportunities across so many vectors that we think will grow faster than the industry as far as I can see it in the future. And frankly, given all the insight that I have into these things and some of the things you'll learn from Brian in the next presentation, I personally have never been more bullish on our business than I am today. So Yes, actually let me introduce Brian Stempek. He's our Chief Client Officer and I think he'll bring a lot of this really to life and really look forward to it. All right. So you might be asking, as I often get asked, what does a Chief Client Officer do exactly? So I'll cover that with a very exciting org chart here at the beginning. I oversee a lot of our global agency relationships. So we've got relationships in place with Omnicom, WPP, Publicis, largest agencies in the world. And a very part of my job is to figure out what do they want to do next on our platform. So when we go to build a text to TV product, when we go to build a digital radio product, what do their buyers want to do and how we go structure the product and partnerships behind that. So I oversee some teams in data partnerships, in inventory, so new channels and markets where we can actually run ads, our marketing team to get the message out there. And then we've touched a bit on our enterprise customers, So clients who are building on top of our APIs and using our bidders rather than their own. A bit more about our clients. So in the past 3 years, we've gone from 389 up to 614 currently. I think what's important to note is that those 6 14 clients represent about 30,000 advertisers. So we're powering some of the largest brands of the world in many different markets through aggregators like agencies and other enterprise companies. Large agencies are about 58% of our spend. The midsize and smaller agencies and the enterprise API clients are 42%. Enterprise, in particular, is a fast growing category for a lot of the reasons that Jeff and Dave mentioned. They're facing cost pressures. They're more willing in the past 18 months to say, well, should I just use Trade Desk bidders and white label that rather than using my own? So we're actually powering other DSPs, other ad tech companies, capturing their spend as well. We talked a lot of it a lot today about programmatic is growing. Rob just mentioned it's growing at 21% per year. I'm going to focus for a few slides on why that's happening, so we don't just kind of take for granted like programmatic is so amazing. Like, why is it actually taking so much share? Why is it growing so much faster than digital? So left hand chart here, this is an example of a large global advertiser. So if you look at some of the large CPG advertisers, for example, they'll spend something like $8,000,000,000 in advertising. And of that, digital is still minority. So digital is about 25%, if you look at a large a lot of large brands. And that, of course, is changing. More of the dollars from the $8,000,000,000 are flowing into the $2,000,000,000 So digital is a rising tide. If you look at a large advertiser on the right hand side, the green chart here, you'll see kind of the annual budget for a large advertiser globally in programmatic going from $30,000,000 to $70,000,000 to $120,000,000 coming close to doubling every year. And that's largely coming out of the share of digital here. So digital, on one hand, is growing, taking budget from broadcast TV and print and other more legacy channels. And within digital, programmatic is taking more and more share. Now that's a bowling alley we're putting in upstairs. What's going on there? So advertisers keep on putting more budgets in programmatic, and they have been this has been a trend for the last 8 or 9 years. But why are they doing that? So Dave stole my slide earlier, but mine is slightly different. I think it's really important to understand how media buyers used to buy media, actually for the most part still buy media. So think of a magazine like Vanity Fair, each one of these dots is a single ad impression. In the way it used to be and the way it still works, if you go and do an insertion order and say, I want to spend $1,000,000 of Vanity Fair, give me the front page, give me the premium ads, is that Vanity Fair is the one who decides which ads you get. So you're the automotive company, you pay the highest CPM, you cut an order with Vanity Fair, and they give you the first look at all those ads. They give you the first 10 ads that they show. The shampoo company, Vanity Fair says, okay, well, they paid a slightly lower CPM. Let's give them the next tranche of inventory, the next 10 ads. And they go on down their list of who did direct deals with them. And for a long time, that's been how the world has worked. The publisher decides who gets what ad. Programmatic is just bringing the choice to the buy side. It's saying, well, of all those 100 ads that you could buy, who should buy what? The automotive company, of course, is looking for people in market for a new car. Maybe they're loyal to that current car. Maybe their lease is up next month. They have data to say, those are the exact impressions we want. We have better data to decide that than the publisher does. The shampoo brand maybe is looking for women 18 to 24 trying to build the brands. They know other products that those types of people buy with them. And they're picking different ads. There's a lot of types of people who buy or go to Vanity Fair. I think we actually understate this, the level of this change as an industry. This is an order of magnitude better way to buy media. In the previous way, you're letting someone else decide which ads you get. You know that Vanity Fair is a good place. It's relevant to your brand. That's somewhere you want to be. That's a good start. But if you're getting ads at random, are you getting ads chosen for you by someone who doesn't know what customers you're trying to reach, you're generally going to lose out versus choosing the ads yourself. Another way to think about that is that for an average consumer, an average user, we're seeing them about 1,000 times per week. Jeff put up the slide that there's a lot of activity that happens outside of Google and YouTube, outside of Facebook. They're going to Hulu. They're going to Spotify. They're going to a lot of other places. It's a big Internet. And a lot of campaigns that we see, the brand is trying to say, well, I want to show 1 32nd spot once per week to a consumer, maybe twice per week. With 1,000 ad opportunities, that's an easy way to lose your shirt. It's easy to buy media poorly. We help them decide with a lot more data behind them, what's the 1 or 2 ads you should buy from that 1,000. So a bit more tactical on a case study. So this is what you look at here, this is a large insurance company. And the y axis is the cost for acquiring a new customer. So someone going to an insurance company website, filling out a quote form for auto insurance or for homeowners insurance. And it's over a 12 week period. And the blue dots, that's where our you heard Dave reference CPA. That's a common metric we're using for performance, an ROI goal for the advertiser. And we see this dramatic change over time where when the campaign started, it was costing them $2.50 to acquire a new customer. That's way too expensive, and that's not the goal they had. The goal they had was the red line of $100 CPA. And we're gradually bringing that down over a relatively short period of time. In the old days, the way an agency would do this is they would cut insertion orders with publishers and say, I've got an IO with Vanity Fair, I've got an IO with ESPN, I've got a bunch of publishers in the plan. Every month or every few weeks, they'd look at it and say, well, ESPN is doing well, Let's give them more budget. Let's take budget away from Vanity Fair. And they're essentially kind of manually allocating and reallocating that budget. The difference is here is we can do this much more real time. So we can do this on a given site every hour, every day, and we can do with humans doing it, and we can do with algorithms doing it. So what are changes that we make? So here's one that happened in this campaign. There was a recency change. Now that means is sometimes when you get retargeted, you go to a hotel website and you see ads for the next 2 weeks, even if it's long after you've already booked the hotel. You're already out of market. We realized that for this client, for insurance, the period of decision making was about 3 days. So there's no point to retarget them for 7 days or 14 days. We cut it down to 3 days and stopped wasting money in places we didn't need to spend. And when I say we, I'm talking about the Trade Desk and the agency working together. The next step, we have a lot of tools that buyers can use themselves. They're going to say, well, I want to change recency, I want to bid more here, bid less here. And we also have algorithms they can use. So we have auto optimization. Some buyers come in and say, you know what, I actually don't have a great thought today of how we should change dayparting. Should we pay more on Saturdays than Sundays? I'll get to that later, but for now, I want to auto optimize. And we have machine learning in the background that's going and saying, well, what's your goal? It's $100 CPA. Great. We're going to optimize the sites, the day parts to try to hit that. And that's happening behind the scenes. So they gradually add strategies. Later on, they might add something like keyword targeting. They find out from some of the site analytics we look at that people were going to automotive sites and comparing policies or going to recent real estate sites and comparing, well, how does Liberty Mutual compare to State Farm looking at different brands. We have partners, as Rob described, in our marketplace. We can look at keywords related to that. So these kinds of things happen every day and every week to drive that performance. But getting to a level where you're coming in at a third or half of a long standing goal. This is a KPI that Afristers had in place for 5 years. And you're crushing performance, not on 1 or 2 conversions, not for a week or 2, but at scale in a global campaign, that's the reason that programmatic is winning. That's the reason more dollars are flowing here than anywhere else. So what we also see is that agencies are responding to this. They're staffing up and they're realizing because programmatic is powering more and more of the media plan, because it's the most effective vehicle they have, they need to do more. This story is actually from 2 years ago. Ad Age wrote a story about Vivicky, which is the centralized agency trading desk at Publicis, overhauling its trading desk. And what Vivicky did, which I think was ahead of its time and this article describes, is that they had to centralized trading team where there'd be a floor at the Publicis building in New York, Chicago, where all the traders sat. And so if you worked at Starcom or Zenith or any of the agencies within Publicis, you would send media over to that centralized team and they would execute it for you. What Publicis realized and what many agencies have realized is that's not enough. As programmatic goes from being kind of one line in the plan or one small portion of budget to growing to be a lot more, you actually need most people at the agencies thinking about programmatic and executing programmatic. So they took apart the Vivici Trading Desk, they took all the experts they had, and they put some of them in each agency. And that's generally what you see from a lot of big agencies now is they're moving from a centralized model with programmatic to a decentralized model. In the same way that no one talks about electronic stock trading, no one talks about, well, are you doing that or are you waving piece of paper on the stock market floor? The same thing is happening in agencies now. In the next few years, I think we'll move to a phase where you stop talking about programmatic versus digital. And programmatic starts to power most of the things that you buy and most of the buying you're doing is data powered. It underlies the entire media plan. Here's another way to look at that. So this is a large agency client of ours, and we're looking at the new users we have by year. So how many log ins did we give out to employees of the agency? And you can see in 2013, the agency had just started. So they had 68 people, they're testing, they're running a handful of small campaigns, they're kind of dabbling. 2014, they doubled. So they have more and more people logging in, running campaigns and more channels with us. 2015, they give out way more logins. So they give 1,000 people, 1,000 new users logins to our platform. That's all over the world. That's new people at the agencies. That's not just a centralized programmatic team anymore. This is now most people on the media side of the agency starting to log in. Same thing in 2016. At this agency, you've now got 2,500 people logging in where it used to be 60. That's a theme we're seeing across the board. We expect that if your core job is media buying, generally, we're offering tools for you, whether you're a core buyer, you're a media planner, you're an analytics person, all that happens in our platform. That's a big part of our strategy. So when we think about what to build next, it's to get deeper with agencies. Every product we roll out has that in mind. So here's a few examples. A lot of media buyers want to spend with premium publishers like Hearst, like Vanity Fair. So we have a private marketplace product where they can go and do direct deals with publishers. We have a text TV product you've heard a lot about today. TV buyers can use that. We have digital audio, which radio buyers use. We have a planning tool that the media planning and strategy team can use. We have log level data feed and APIs to pull down data sets that the analytics team, the data scientists at the agency use. So more and more, our goal is to say we want everyone at the agency logging in and using a lot of these products. But of course, we never expect to get 100% of the media plan. If you're at a media agency and you've got a large advertiser client, a big part of your job is allocation. And so you give some money to publishers, you give some money to the DSP, some money to your trading desk if you have 1, you allocate across and see who does the best. That's still part of the role of an agency. What happens is though, we power a lot of that. So the Agency Trading Desk is likely using The Trade Desk. Goodway Group, they're a digital solutions company that focuses on automotive clients. They're really good at automotive. So this is an automotive advertiser. They might give some budget to the Goodway Group. And Goodway Group has built its own thing. They're built on top of our APIs. So that money comes through us even though the agency, the media planner might not realize that. But they don't need to know that. What they care about is the performance Goodway is driving. SteelHouse, they're a dynamic creative company. So they compete with Criteo, the kind of U. S. Version of Criteo. A lot of retail level, SKU level creatives doing retargeting. They're built on top of our APIs. So they're retargeting customers. They don't own their own bidders. They do a great job with creative. They have their own user interface and they power e commerce. We don't talk to any of those customers, SteelHouse does. And then other DSPs, some of whom we just can't name, who actually use our platform. So we ended up capturing more of the media plan that way than if we went just directly to the agency. I want to talk a little bit about how even beyond our platform, we're thoroughly ingrained with all the different functions at the agency. So kind of day in the life for a media buyer. You've got a buyer comes in, has their coffee. 9 is probably a little optimistic for a media buyer in New York, it's probably 945. They check their reporting. Live agencies have their own centralized reporting dashboard, and it's pulling and reporting for everywhere. We're linked into that. So they're seeing real time insights flow in from what they buy on the trade Desk. They go to lunch, in their desk, they're loading an ad creative, the actual image of the video that's going to be run. And they're loading that in the ad server, they're pushing that to the Trade Desk. We're integrated with their ad server. So we're integrated with most of the major ad servers, so it's one click. Again, more time in their day, they can spend on valuable activities, not doing the same kind of repetitive task over and over again. 130, they log in and say, of course, I want to double budget for The Trade Desk. That's the case. We'll do that. 3 they log into a data management platform. So they go to Crocs, so they go to Adobe, they do some audience modeling, they build a new audience, again, one click to push that to The Trade Desk. We want to make that seamless. 5 they're getting tough questions from an advertiser about, hey, how does this cross device technology work? Can you explain it to me? Can you tell me more about it? We're actually connected to their learning management platforms. So at the agency, a lot of them have deals with us to learn, have video content to teach them more about new technology and programmatic. So throughout the day, we're thinking about all the different touch points to make the agency smarter to serve their clients. It's also important to us. Dave covered this in a slide showing you, okay, here are the core things that DSP has to build. We want to leave room at the top of the stack. We want to leave room for others to innovate. We don't think we're the be all, end all of everything that can be done in programmatic. And a big part of that is a quote here from Steve Kettleman from Omnicom, quoted in that exchange, you're talking about how our system is extensible. You can build on top of the APIs, you can build your own proprietary advantage. We have our own algorithms, but we encourage clients to build their own. That's what a lot of agencies and enterprise clients do. They'll go and say, well, I want to bid more for a certain site. I want to bid more for this time of day or for this ad format. Again, we have our proprietary techs to do that, but we give them the tools to say, you're an agency, you've spent 1,000,000,000 of dollars in the last 30 years in media, you probably have a pretty good idea of what you're doing. Let's help you turn that into an algorithm. Let's help you build that. We have a team that actually enables them to do that. That's unique. Many DSPs are more black box. Click a button to optimize, it doesn't really tell you what it does. You don't have influence. We're trying to give them controls to do that and do it in a scaled way. So it's there's people having insights, but there's also machines that are powering a lot of this. I want to address this pretty directly. Every year in the media, there's like in the ad tech or media press, there's kind of a trend. And the trend this year has been, our marketers taking things in house. It seems like they're taking media in house. It seems like they're taking programmatic in house. And I think that's a little bit too black and white, where what we see is a lot of advertisers are showing more interest in programmatic. They want to learn more when they're doubling spend. If you're seeing $100,000,000 in programmatic, whereas you used to spend $5,000,000 or $10,000,000 of course, you're going to pay more attention to it. You want to meet the tech company. You want to understand what you're doing. But I think too often, it's presented in this binary way. Either take it in house, do it all yourself with 100 person media team or just let the agency do it and never learn anything more. What we're seeing is this middle path, and there's actually a great piece in AdExchanger this week where the Head of Media for Electronic Arts that talks about the different kind of flavors of taking things in house. Sometimes brands are getting logins. They want to get reporting logins. Other times they want a direct contract with the DSP, so they get the insights. So they have more of a long term SaaS type deal. But in almost every case, the agency is still the one managing campaigns and providing the insights. In house doesn't always mean you're fully taking media in house, it means you're getting closer to the technology. And so we see a lot more of this middle path where it's The Trade Desk, the agency and the brand all sitting at the same table talking about, okay, you're spending $100,000,000 in programmatic, how do we hit your results? That's a lot more common than a brand saying, okay, we don't need the agency anymore that's serving us in 110 countries. That's actually still relatively rare. So what does it look like when the agency marketer and trade desk sit down at the same table? Here's an example of the types of things that we share together. This is a large bank that's trying to get people to sign up for brokerage accounts. They're trying to hit again a target cost per acquisition. They're using campaign data to figure out how to drive better results. And overall, this decreased their cost per acquisition by 44%. So here's what they learned over the course of this campaign. So first thing, we looked at the data with the agency and what we found was that 78% of this bank's customers who were signing up for a brokerage account signed up the same day that they hit the website. So again, there's no need to retarget them for 2, 3 weeks. Programmatic isn't just retargeting. We said, okay, your site is actually really good at converting people. What you need to focus on is not retargeting, it's finding more new customers. So how do we drive more people to your landing page, which is doing really well? Where do we find those people? What are they doing before they come to your site? Well, we look at 8,000,000 ads a second. Let's look at that population. Let's see what other websites they go to. And those are insights we get without spending any money. Turns out they're going to finance sites. They're going to the NASDAQ. They're going to CNN Money. They're going to news sites. That's all in the hour before they sign up for a brokerage account. So as you expect, they're in a financial mindset before they sign up. The day before, they're going to LinkedIn a lot, they're going to ESPN, they're going to Seeking Alpha, kind of free trading sites. These are day trader type customers. These are the sites that you want to buy. So we're providing these insights to inform the media plan, to tell them, here are the exact places that people go the hour before they convert. This is how a better way for you to spend money. One thing we love to do is heat mapping. And so we'll work on analytics showing state by state, country by country, even county by county or town by town, where is your brand strong and where is it weak? So in Virginia, there are actually counties and towns where people are hitting your website a lot and they know your brand well. Sometimes there's an urban rural divide, sometimes there's a coastal Midwest divide. But it kind of provokes this conversation with the advertisers to say, what do you want to do where your brand is weak? Because this is just about performance and acquiring new customers. We would tell you double down in the blue the places where your brand is really strong. You should spend more money there. But that long term, that's probably not a good strategy. Long term, you actually need to do branding and awareness campaigns in the red areas, in places where people don't know your brand. But it provokes this more interesting long term conversation about this isn't just about direct response. It's not just about cost per acquisition. You have real issues with the way your brand is not strong today, and we can help you understand that and sort of split out which creatives you show and what message you show. Last type of insight, weekdays, work hours are best. People actually sign up for this brokerage account during the middle of the day. They're browsing during lunchtime at work. They're most definitely not going on a weekend and sign up for a brokerage account. They're not in that frame of mind. You shouldn't be wasting money there. And we'll show you the exact hours of the day and exact days where you should bid a lot more. These are the types of things that along with hitting performance goals for the brand, the other important thing to note is this informs the rest of their media plan. They're still buying broadcast TV. They're still running in print. They're running with publishers. Those are still a huge part of the reason that there's $600,000,000,000 in advertising and most of it isn't in the programmatic yet. These insights can help them get smarter on everything else they do. So I want to talk a little bit about our partnerships team. So a big question we often get from investors is, do you guys have enough ad inventory? Are you looking at enough ads in every channel and every market? How do you think about that? And so we have a 20 person inventory team all over the world. We have a team in Asia, in Europe and the U. S. That goes out to sign up new inventory to the platform. In many cases, we're working with SSPs. In other cases, it's with large direct publishers. And we focus in places we want to add scale. So some deals we've done in the past year that are great examples. You've heard a lot about Connected TV today, Hulu, Sling, DIRECTV, Roku. We have a couple of people just focused on getting connected TV inventory at scale in the platform. We have a team going after inventory in China. So partners like Baidu, where we want to run programmatically in China. Spotify, we launched more than a year ago. We're adding more and more programmatic audio inventory. But constantly adding these new partners, as Rob said, we're operating this grocery store, of all these shelves where people buy things, having inventory in every channel at scale is a huge part of what we do. Jeff and Dave both touched on header bidding and all the various impacts it has on cost structure, putting pressure on other DSPs. There's one other trend that's worth calling out. Before header bidding, back to this Vanity Fair example, the way things worked is, again, they would sell a bunch of the ads premium and direct where Vanity Fair picks where they go to the automotive company. They give a nice tranche of inventory to the shampoo company. And they'd sell all the stuff they could directly, and they'd say, well, this is all we could do ourselves. The rest of it, where that black dotted line is, let's take back the SSPs, NAT exchanges, NAT networks, see if they can do something with this since we couldn't sell ourselves. So that was the way DSPs and SSPs operated in 2010, 2011. With header bidding, that goes away. So when we see more inventory, yes, there are cost implications today. You have to look at more ads. But you're now bidding directly on the premium stuff. We're getting access to the good ads that were direct sold before and given to large advertisers. And we can now say, well, if I have data to indicate this is a premium front page ad on Vanity Fair and I know it's in the end market for a new car, that's good for everyone. That's good for the publisher. That's good for the advertisers. That's a better pairing than it was when the publishers, in a somewhat random way, allocated those ads. So our inventory pool has gotten bigger, which has cost structure implications, but also is a good thing in terms of available inventory. Rob talked about the data marketplace that we offer. Again, we have a full time team adding new things to the shelves and thinking about what products we build ourselves. I'll tell you a little bit more about what does that look like? What does it mean when somebody picks something off the shelf? Like why do they do it? How does that work? A few examples. This is a campaign for someone, an airline selling airline tickets. A few what they often do in a programmatic campaign is they'll try a few different strategies. So one might be they go to our data management platform and they say, let's go for an in market air travel segment. Who are people most likely to fly to Los Angeles this week? We will buy cookies and buy CDs for those users. It says the price of the dollar CPM to use that data. And that's one line in the plan, one strategy they'll try. Then there's contextual keyword targeting about websites and travel deals. They'll set up another line just for a strategy there. Then they're using our ad server to serve the video ads. There's a CPM cost to do that. They said, well, actually, your ad server is pretty good. Do I need to use Seismic? Do I need to use DoubleClick? No, let's use the Trade Desk ad server. They use pre bid brand safety to avoid sites and places they don't want to run, so negative websites about airlines or stories like that. There's a CPM charge to do that. That's kind of an insurance policy they take out. Then we have optimization features where they maximize viewability, often a KPI that advertisers care about, how viewable is the ad, how do we maximize that. These are all things that people check along the way as they're saying different strategies and they compare them. And they say, well, actually the behavioral targeted segment, that came in at a really good cost per acquisition, whereas contextual didn't work that well. Shift more budget to the DMP side of things. In the same way, the agency used to allocate money between publishers. Now they're just to allocate money between different strategies within the DMP and the DSP. They're figuring out what works, and they're doing that much more fluidly than they did before. By the way, one of the next marketplaces we're in the process of launching is a marketplace for algorithms. So actually giving some of our partners a chance to say, do you want to buy an algorithm based on weather data or based on financial data, other performance algorithms that you can license in the platform sold by outside parties that compete with our own. We believe that adding these new partnerships is a differentiator for us. There are some DSPs that only add 2 or 3 new partners per year. And with when you're dealing with advertisers trying to move into new markets, that just doesn't work. We've built this repeatable format where we can quickly add new SSPs, new data companies inside the platform. We've got an integrations team that does that. So our partnerships team goes out and says, let's sign up large Indonesian publishers. The integrations team quickly onboards them. We can do that in weeks, not months or years, with little impact to engineering. And so what that means is when we go into a new market like Jakarta and open an office there, where we have large international brand spending, we can go to these are the 3 top websites in Indonesia. We have deals with the top 10 sites all existing on our platform. So when that agency in Indonesia tries to go launch campaign, we're already there with the local and regional data inventory that they want. So that's a big part of our focus. Thing I'll talk about and then we're going to bring up a client panel for you guys to hear directly from a set of our clients is how we roll out a new product line. So you've heard tons about Connected TV today, why it matters, why we're investing there. I'm going to talk a little bit about how we built that product line and what we built. So when we think about a new product, there are 3 things. 1st, where can we run ads? 2nd, what targeting can we add to make it better than just buying directly? And third, what can we report on that's a little bit different? Jeff and Rob touched on this. We see a huge impact of consumers spending so much time on connected TV, whereas that 1 in 5 hours of TV consumption today is streaming and soon to be 2 in 5 hours. So we think this is a must have. This is what the past 2 years looks like for what our inventory team has been doing. So constantly adding new inventory from Connect TV suppliers. We started with smaller players like Kraftl and Samba. And then you see this year, we've hit a tipping point. Large networks like AME, big offerings like Hulu, some of the skinny bundles like DirecTV and Dish Sling have been some of the largest inventory sources we've added. They represent huge amount of inventory we can buy. But every TV network, every cable operator, we're in conversations with to keep adding inventory to our platform. Same thing happens on the targeting side of things. So we've worked with our cross device partners to say, how do you take a cookie and extend that into a connected TV? If you know someone who bought your product online or you have a data point on them, how can you then find that same person on a Roku or on Apple TV? And cross device technology enables that. We do a lot of household targeting using IP address. Device type matters. There's a different audience watching content on an Xbox than watches on an Apple TV. It tends to be younger, more of a gamer audience as you'd expect. So targeting by device matters. And there's lots of good data that comes from cable companies. Who are heavy NBA watchers, who what people love dramas on TV, who's watching late night shows. They sell that data so we can target that and extend that audience. And then last, what can we measure? Nielsen is still key for us here. So we actually have partnered with Nielsen to measure incremental GRPs. So when you're missing those 18 to 34 year old millennials, are you finding them on Connected TV? Are we getting those GRPs back for you when we run this campaign? That's super important to TV buyers. Offline attribution or sales on other devices. This is a game changer for TV. Keep in mind, most broadcast TV is bought at a national or at a city level. There's no way to tie it back to a sale. You can't tie it back directly to a 1 to 1 consumer purchase. Now you can. You have a much better idea through our platform of the actual ROI of buying television than you ever did before. That has a huge impact for how people buy TV. And then things like viewability, player has the video completed, more typical digital metrics the buyers are looking at. And then we combine that with our marketing team. So our marketing team then tries to take this story out to our clients and to the press. And with Connected TV, I think we've been very successful in getting a lot of attention and being a leader in the space, building the products that agencies and marketers want to use. Thank you so much for your guys' time right now. We're going to bring up our client panel next, and we'll go from there. If our client panel can please make their way up to the stage or ad agency and brand panel please make their way up to the stage. Brian is going to be moderating, and I'll pass it off to Brian to introduce today's panel. All right. So I'll introduce the panels we have here. So my left is Barry Lowenthal, President of the Media Kitchen. We've got Gabriel Cohen is next, the Associate Media Director of Starcom and then Sean Healy, a Senior Brand Manager at U. S. Cellular, which is the 5th largest telco in the U. S. So we're trying to do here through their panel of people ranging from agency execs to people running programmatic teams at the agency to the actual advertisers themselves, so you guys can hear directly from all the different types of constituents that we work with. So I thought I'd kick it off just by asking you guys, we've been talking a lot about why we think programmatic is growing, why it continues to grow. I'd like to go down the line and get your guys' take on over the past decade. And Barry, I mean, you started the 1st agency trading desk ever, Barrick, inside of MDC Partners. Take us back to that day when you did that and help us understand where programmatic has come since then. Sure. So thank you for having us. Is this on? Yes. Yes. So for such a long time, and I think you did a good job with your Vanity Fair example, we used to use content as a proxy for an audience. So if I needed to reach women, I would buy Better Homes and Gardens. And if I needed to reach men, I would buy Sports Illustrated. And while that's probably not entirely true, that's what we did. And inherent in that model is waste, right? And it's extraordinary how advertisers for decades tolerated waste. We always used to talk about having a high waste tolerance. And the magic about programmatic is that waste just evaporates. Now you buy the people you want, when you want them, as soon as you see them online. And that's pretty magical. And that was a game changer for us. So now we can buy the audiences. We don't have to use content anymore as a proxy. And I think that's been sort of like what's been propelling programmatic for the last decade since its inception. And as we can bring 1st party data and 3rd party data, our targeting continues to get more and more precise and refined and efficient, allowing us to squeeze even more waste out of the model. Gabriel, you use platforms yourself. You actually built a trading desk and an agency. We talked about this at breakfast today where you were the first person kind of building that out at a smaller agency and then moved over to Starcom where you've run bigger teams. Can you talk a bit about the same, like how have you seen the model evolve with programmatic where it's gone from, okay, this is a one person team inside the agency to, and I mentioned Publicis has been pretty aggressive about doing this across the entire agency? Yes. Am I on here? I think so. Can you guys hear? Yes, I can talk loud. Okay. Yes, so from the very beginnings, you see smaller teams and brands and clients starting to kind of get on board, and it takes a little while. The adoption is slower in the very beginning. And as they talked about many times that trust gets built and then adoption moves faster. And I think that's what we're seeing is that adoption is just moving much faster now because the trust is built. The control is there. I think the big I would echo some of those things. Just said that the planning tools have created such a robust way for us to know and find people and just being more data driven about our media. I think also having hands on keyboard, growing the talent pool and growing people that can touch and be in these campaigns. And we're at the agencies, we're so close to the clients. We understand their business very well. So instead of passing that off to a 3rd party with an IO, now the people who are invested with their marketing teams, working in tandem with them every day, are hands on keyboard, controlling data, pulling levers, giving them access to information in a much faster way. What's that transition been like for both of you guys? I mean, so at the agency level, that is change management, right, to go from traditional media buyers to buying digital and then digital media buyers actually being able to get educated and programmatic. And that's been an ongoing process years. I'm curious as to how you guys have both navigated that. Yes. So I think it was 2014 we won programmatic and search agency the year from MediaPost. The reason why we did that is because how we very quickly integrated programmatic buying into our search group. So search was the first programmatic channel. If programmatic is talking about automation, search was probably the 1st programmatic channel. And we realized very quickly that those people had the skills needed to buy media in an auction based environment. And so we just retrained them. And we gave them a whole new skill set and a whole new career path. And then when paid social kind of blew up, we created what we call a biddable group. And so now we have people who know how to buy media in programmatic channels using DSPs, search and also paid social. And because these tools have become so easy to use and these guys have talked extensively about how they service agencies and how they train agencies. And I can't tell you how valuable that is as sort of like a real on a partnership level. We've been able to distribute our biddable team throughout our entire agency people. So people at the media kitchen know how to buy media across all different kinds of auctions, which is pretty great. So they know how to negotiate print and they know how to do a direct content deal with a publisher, but they also know how to buy biddable media across lots of different biddable channels. It's not easy, but when you're committed to training, and you're committed to bringing these we call people that work at the Media Kitchen Chefs, because you know we have to extend that metaphor. So when you have a commitment to training these chefs and you have a commitment to bringing the best insights to our clients, you have to be constantly retraining them with all of these new platforms. Yes. I would echo that from a person tactically executing standpoint. Those bidded media folks just kind of can make that leap and learn all these tactically. I think the slide you showed about our company and the dissolving of Vivici and putting those people into the agency, that was critical from a strategy and planning perspective and really bringing that expertise into the agencies, because you had people who could execute tactically. But those people didn't understand the world of media buying on a bigger level too sometimes because they had been tactically focused on digital media. So then learning the ins and outs for executing private marketplace deals and understanding the inventory and understanding how to negotiate and those type of things and then bringing their expertise in Bidded Media really helped accelerate the growth in learning and kind of powering those better strategies for clients. Yes. So it sounds like a lot of you guys are driving is like this becomes almost a kind of standardized part of the toolkit that you expect from everyone at the agency. Sean, to switch gears a little bit, I'd love to hear from your point of view at the advertiser level. U. S. Cellular, I know you've worked with agencies as well, but U. S. Cellular has been pretty aggressive in moving into programmatic. It will be interesting to hear why that is. Is it performance driven? Is it the insights? What is it that kind of moves the needle for you guys in programmatic overall? I mean, I think number 1, Barry, I already kind of touched on it. It's just simply a waste. Like when you think of my category, I'm not a QSR, I'm not a CPG. Our buying cycle might be one person, might not look at wireless for 12 months or 24 months or 36 months. I mean, I was spending so much money to message people who simply aren't even interested in the cell phone plan as an A, and then B, I was also spending a ton of money to message people who are already my customers, who aren't even eligible for the deals that I currently have. Like we all have that pain point, right? So from a programmatic standpoint, it allows us to start to find those customers who are in the cycle right now, not 2 years from now is A, and then also B, helping to eliminate targeting my current customers. Can you guys talk a little bit about how you think about allocation of media? So I think one interesting question is, 5 years ago, you would see, okay, there's a line on the media plan for retargeting or for some subset of programmatic. And now with private marketplaces, with other channels, it feels like that's underlying more of the media plan, but it's not done yet. There's still a lot that's done direct. I'd love to hear your guys' point of view on how you think that's changing. What's blocking it from changing more? So I think the concept of allocation is and no disrespect to you, is sort of like an old fashioned concept. And the reason why I think that is because it assumes that you have to make these sort of like arbitrary decisions based on some sort of strategic imperatives, which are already sort of defined by a person, which makes it a little bit wasteful to begin with, right, because these people are just inherently wasteful or the way that we decide is inherently wasteful. And so we kind of have been leading our clients into something that we call signal based planning. And so at the core, right, and I suspect this is true for US Cellular, the most valuable customers are your current clients who you can kind of upsell and cross sell and just sort of retain. And then if you think about clients a little bit outside of that are people that are shopping your product right now. And then a little bit outside of that are people that are shopping your category. And then a little bit outside of that are people that are, based on various life stage triggers are likely to enter your category. And we can use signals now, many of them data driven media signals. So shopping behaviors, search behaviors, people have gone to clients, people who are reading content, people who are people who have certain apps on their phone and so on and so forth. There are thousands of different signals now that you can identify that will place people very clearly where they are. And so rather than allocating people based on sort of like a purchase funnel, which doesn't really reflect the way that we make journeys anymore, or based on signals, you can find people as they're raising their hands. So it's not about allocating people based on strategies. It's about allocating people based on the right signal at that right moment in time and then very importantly serving them the right message to persuade them to buy. I mean, you talked about that, about wanting to do that. And I think you guys do that, right? You know, you want to be able to reach people at the right time so that they make the right kind of decision. And the way that we do that is by using signals. So it's almost more like always on like searches as opposed to the more top down upfront model of Completely and very much machine learning driven. It has to be because we want to do it at scale. Yes. Do you guys see that as well? I mean, it seems like some clients have moved that direction, but with other advertisers, there's still an upfront kind of allocation process. Programmatic hasn't quite won the day yet. Yes. I think TV is still a king when you're talking about global companies and larger advertisers. I think TV still is going to grab some of that line share. And so that's a little bit of a barrier to entry. There's education around the CTV and OTT space that's happening constantly to help make sure that we're balancing that. So there is that upfront aspect, but I do think that more and more advertisers are going to the plan that we start with audiences. We start with who we want to reach, not a demo SKU, not a panel, real online data, who that is and we want to reach them. And then that's where programmatic is really starting to win the day. Sean, what have you seen in terms of some of the insights? That's when we talk to marketers, what I find is most interesting to them is when they can model and say, okay, well, here's my current customer base. Tell me more about what they look like. How do I go find more people like them? Again, beyond 18 to 34 male or this income or this geo, but getting granular like, do they love bowling? What car do they drive? Is that something you guys have delved into? Yes, we actually do it both on the front end and we're working really hard to do on the back end as well. On the front end, we're starting to look at like our current CRM database. As we start to target, for instance, we don't want bad credit people coming in because we know they're just going to involuntary churn out 6 months after we gave them an $800 iPhone. Point being, we can start to find those key high value lifetime targeted segments and push a lot of advertising, which to the points already discussed, a customized promo, customized message to those people. And then on the back end, we're now starting to try to pull out insights to see who is converting online. And let's just say for instance, we know they're big basketball fans. So maybe that means then we can start to work in some type of basketball type terminology or basketball sponsorships into our other marketing activities, again seeing that's who's really kind of hand raising and who's ultimately buying our products. So it's definitely very cyclical and it just keeps feeding itself. What are you guys seeing on that front in terms of the consumer insights that are being passed back? I mean, you work in advertisers like Vanguard, you've been on Mars, Wrigley and Hallmark. These are all brands I suspect are hungry for more information about their customers. Yes. When programmatic first started, 3rd party data was a big sort of data charge. And because the price of media at the time was so cheap, data start if you started overlaying data, you wound up driving very inefficient CPAs. It was better to just let the algorithms just drive acquisition. But when we did that, we didn't have any insights because we weren't having any sort of audience overlay. Now, we have so much more audience insights using 1st party data and third party data. So we always say whoever has the best insights wins. And I think that's one of the reasons why programmatic has been able to dominate digital media because we get some of the best insights through programmatic data driven media buying. Yes. And it is the best, most robust data set. It is in real time. And it's very important for advertisers to start to leverage their data and learn from that. And I think you guys have done a great job about making it point and click. You make it so much easier to access a lot of different DMPs, a lot of different sort of data sets, which I think just will accelerate that kind of media buying. Yes, absolutely. One thing I touched on during the presentation was this notion of you hear a lot in Ad Age and Ad Week about our brands taking things in house. You guys are front and center on this, dealing with lots of different advertisers. You're directly at one where that could be a consideration. I'd love to hear your takes on how real is that trend? What are you actually seeing? So I mean, if I was a client, I would certainly want to take it in house. And I think the reason why I would want to do that is because I would want house and house and I would even consider maybe even paid social, I mean, all biddable media. The challenge has always been that clients are just not staffed up to attract that kind of talent. And I think that's sort of what's saved in a lot of ways, programmatic at agencies, because agencies do provide the best sort of career opportunities for that kind of talent. But I do think it's something that agencies have to really pay special attention to moving forward. Clients are going to want to control as much of their customer acquisition as they can. And so agencies really have to sort of make sure that they're providing a lot of value if they're going to keep that business. And I mean, I think you guys do a good job of being a friend to an agency. What do you guys say? I do think, one of the last speakers talked about a lot of clients are going down the middle road. It's funny, we had this exact conversation back in my office not even 12 months ago and it was a very short conversation. It was do we go in house or do we just have Starcom be hands on keyboard for us. The thing we really kind of landed on and it was just touched on is this space evolves so quickly, we would have to staff up so fast and we're just simply not that fluid of enough organization to where as the space changes, we can change our staffing that quick. Whereas if we have the agency do it, then we have our direct relationship with Trade Desk, you know, pass through to the agency is that as those needs change, the agency can just automatically change with it. And then also, you know, from where I sit on the brand side, I see a lot of value in having my team at the agency be robust across the entire media landscape and also all the analytics. So when they look at everything it is my one group of people, look at everything from a 100% holistic standpoint. Yes. I think the talent gap definitely from a tactical standpoint is important. Also operationally and just thinking about it at that level, that is a whole another different skill set that just doesn't exist. And then if you think about taking it in house, you're thinking about taking your team, your marketing team that should be strategically thinking about growing your business. And you're now putting them into managing people tactically operating DSPs. And that's just not the best use of time. I think that we can all agree there. There's also the shared knowledge piece within agencies. We're there together. We're activating. We're learning as a trading team. We're all learning things constantly and sharing. So I think there's that value proposition where if you're in a brand, that's what you do, you do that every day. You're not seeing and experiencing and learning with other people that are doing the same operation as you every day. Yes. Barry, you touched on how that programmatic start really in performance, direct response, CPA type campaigns, that's still what we see. But we also see a lot more branding campaigns coming in. I love to hear what the types of metrics that you guys are seeing clients ask for, the types of campaigns, is that changing? I think for us, for a lot of our clients, sort of the watershed moment where we were able to justify investing a lot more in upper funnel branding through programmatic was when we were able to connect it to an attribution model so that we clearly understood how upper funnel was driving lower funnel. And when you have that kind of insight, then you can really invest a lot more, a lot more upper funnel. I didn't know that you guys were going to have an attribution component connected to the TV side. Guys have to talk to me about that. It's new. It's new. It's new. It's new. I know someone here about buying those algorithms. That sounds really interesting too. I didn't know about that. But that's really exciting. I mean, I think for you guys, and I think for the industry overall, unlocking that $70,000,000,000 plus in the U. S. That's in the TV marketplace, that's going to go connected, which will, that's going to be a gigantic opportunity for you guys, but also for us. I mean, you guys have TV buyers. We don't have that kind of group. So for us, our programmatic people are going to be buying television rather than TV buyers learning how to buy TV. So I'm really looking forward to seeing a lot of that money moving to this space. Yes, I would definitely agree with the attribution piece, seeing upper funnel branding and seeing it connected to the whole funnel has really driven a lot of investment from a branding perspective. I think just brand advertisers realizing they can get some more efficient reach with different types of media and diversifying their media. So if they were always just a big heavy TV advertiser, it makes sense to hit different experience and be able to creatively express what you want to say in a different environment, a different time, in a different place, reaching somebody in a more targeted way even than your traditional demo panel. I think actually from an attribution standpoint too, it's really demo panel. I think actually from an attribution standpoint too, it's really interesting. There was a few presentations ago the whole online to offline attribution. I'm a big category again where our offline sales are huge for us. And the fact that we can then tie back those offline sales back to digital impressions is just huge and allows us then to justify more spend into the digital landscape of everything. Yes. One thing we often think about is, sometimes it feels like branding for years has gotten kind of a free ride, where it's like you need the reach, you want to establish the brand, but you're not always holding to the same accountability that you might hold a lower funnel type of attachment. It feels like branding is starting to look a little bit more like direct response. Is that a fair assessment? Still getting a free ride? I don't think it's getting a free ride. Yes, I think it's not being held to the same standards direct response. But if you can measure it, why not measure it? And I think that's the answer that the solution that programmatic provides is if you can do it, why not do it. So I think that's kind of where branding sits now is give us everything you got and let's figure out a better way to use it. Well, I think the KPIs from like and we call it equity at our place. The equity side of the house really brings to everything is now starting to be measured very, very specifically. And to Brian's point, we've actually studied that bump we see from an acquisition standpoint when it's married with an equity message and it's a substantial bump. And one of the nice things is about all these new digital tools, let it be the ad servers, the programmatic buying, etcetera, so now we can start to force fit and force marry those different equity acquisition impressions together and this technology is what makes that possible. I don't think about certain channels getting free rides or free passes or anything. So we talk to our clients about harvesting intent before generating new demand. And so you size up all of the intent in the marketplace, you put a budget on it, you buy that before you bring new people into the market. I mean, after all, why wouldn't you do it that way, right? I mean, it makes perfect sense. But for a really long time, you couldn't do that because you couldn't understand who the hand raises are. You couldn't identify them and you couldn't find out ways to buy them. And programmatic has gone a long way in helping us do that. So when I talk about the signal based approach on how to think about it, you first have you bought all your current customers? Have you bought people that are shopping your brand? For a lot of brands, they're spending money in demand gen before they even need to because they haven't harvested all of that intent. So for us, it's about have we not is one channel getting a free ride or not or do we feel good about being on TV versus somewhere else. It's about how many hand raises have we found and kind of work our way out. A lot of what we talked about today is understanding programmatic in the industry. And this is not me fishing for compliments. I have not plugged this in advance. But how do you how would you guys describe for an audience like this how The Trade Desk differentiates itself from other DSPs? How are we different? I mean, I think a couple of ways. So we don't build our own tech. A lot of agencies build their own tech. We don't have engineers on staff. What we bring to our clients is customizable tech stack that they can plug and play that represents best in class across every sort of like tech need. And The Trade Desk is our preferred DSP. We have a couple of DSPs within our tech stack, but The Trade Desk is our preferred DSP. And for a couple of reasons, one, they're incredibly modular, which means that they plug and play really easily, which allows us to bring customizable best in class solutions around viewability, attribution, fraud, data intelligence, so on and so on and so on. We don't ever have any friction when we plug in the Trade Desk. The other thing that's been really great is their training. And I can't tell you how helpful it is to when you're bringing on new people and you're teaching them to be biddable buyers to help you figure out what's the right bid, what's the floor, what's the ceiling, how to do that. And they're very helpful around those two areas. So apart from just accessing inventory and all those kinds of things, I think those are 2 of like some of the important things for us. Yes, I think first and foremost, it's customer service. It's a fantastic tech, and it's one of the best in my opinion. But I think they go out of their way to teach and educate clients and help us be more powerful and more effective at every level, whether that's tactically a buyer to strategic planners, really helping maximize the tools and resources. And then of course, the tech is very customizable with the integrations, the constant rollouts of new things, just really staying at the forefront of where things are going and where they're heading and giving best in class tools. I think for us, it was really a threefold decision when we look at this landscape. From a just from an inventory standpoint, Trade Desk has access to a lot of the premium inventory that we were already naturally buying going publisher direct. Now we can access that same inventory via the programmatic dashboard. And then B, from a tech stack standpoint, we kind of selected Trade Desk in the middle of our development and all the partners we were already working with were already integrated into the Trade Desk platform. So it was a very nice organic fit to start working on Trade Desk as well. And then lastly, it was funny, one of the earlier presentations talked a lot about bid factors. When I sat through it, I've seen the Trade Desk platform literally one time because our agency manages it for us. And they were showing us the bid factor platform. And I was just really impressed with how much easier that makes it for the actual trader sitting at Starcom and doing all the different factors along and then also how they can pull reporting out of it. And as a client, I was looking at that thinking about it from an FTE standpoint. It's like this makes it easier for my Starcom team to work, therefore I have to pay less FTEs to actually do the work. That's great for him. No, yes, just to piggyback on that though, on the efficiency piece, what it allows the team to do then is spend more time analyzing and optimizing the campaign. So it's there's a time saving and there's a redirecting of time to be more analytical, more strategic. And that's helping those younger potential newer buyers get smarter, faster and becoming more strategic faster. And then that's improving the overall workforce. It's like the platform self trains. I mean, they're just getting smarter and better by spending more and more time with data rather than recreating the wheel by not having a Bitfactory system. I mean, I think for this audience in particular, there are 2 things that you should pay attention to. 1 is the connected opportunity. I can't tell you how gigantic that is. Like when that money when that TV money becomes fully digital and is activated through platforms like The Trade Desk, that is going to be a gigantic opportunity. And then the other thing is the fact that they are a platform and that they're global. So The Media Kitchen likes to think of ourselves as a cloud based media agency. So media today lives in the cloud. You buy that media the same way in every single market, right? You buy Google the same way in every market, Facebook the same way in every market, you use The Trade Desk the same way in every single market, which means that I need I could buy media in Singapore from here. I don't need to have people on the ground, which is a big opportunity for media agencies. I'm not I don't have to just buy U. S. Customers anymore. I could buy my customers everywhere in the world. That's a big opportunity for agencies, which is going to allow us to accelerate more budget onto platforms like The Trade Desk. Last question for you guys and we'll let you go. As you look at 2018, 2019, programmatic continues to grow. What do you think are the biggest places that growth comes from? You've touched on CTV. Where else do you think you'll see higher adoption, whether it's new technology like cross device, new channels, new markets? What's your forecast for where that comes from? I already answered. Yes. I'm going to piggyback on the Connected TV piece. I would ride that one. That's just going to be huge. You have huge budgets that are being considered to shift around. I think people are reimagining display advertising too. I think better experiences, content, native, things like that, getting better and more strategic about that is a huge opportunity rather than just thinking about your general banner ad. But how can you partner with the reach that a banner ad gives you with an experience in rich media or something like that. So I think there has been opportunity, but I think in programmatic landscape that opportunity is growing in the private marketplaces with those vendors that have traditionally been direct IO. I think the rich media landscape and more experiential stuff online is growing heavily. Actually, you know, 100% agree. It's definitely going to be also in the TV space. Like TV is going to have a huge problem as millennials and high school kids now get older and older, they just simply don't watch linear TV. And so these guys' solutions start to help us still reach that audience in the TV type landscape. And also just from a cross device standpoint, the digital landscape is still so siloed between app and mobile and desktop and social and search, etcetera, that those walls need to come down. And from platforms like this, it really helps all those walls to come down. So we can see across the entire platform and decide what's the best way to target platform agnostic, that's just digital. Fantastic. Our next panel is our client team panel. They're going to introduce themselves and tell a little bit about what they do and give you a little bit more insight about what we're doing at The Trade Desk on a daily basis. Great. Thank you, Chris. I'm really excited about this panel. We heard from some of our biggest and best customers in the world talk about how our teams help them. And so you're going to hear for the next 30 minutes just how we do that. So Mark, you want to start introductions? Yes, sure. Hello? Hello? Alright. Well, I'll shout. So my name is Mark Davenport. I'm Senior Director of Analytics. I've been with the company for about 6 years. And when I started, I was originally a media trader. And so I've recently shifted roles a little bit, but up until very recently I was managing a team of media traders who were responsible for going into the agencies and helping them. Check. And responsible for helping clients optimize with the platform. Hi, everybody. I'm Kathleen Comer. I'm Vice President of Client Services, and I've been with the company for 6.5 years. I started with the company as a senior account manager managing some of our key strategic accounts, and now I oversee a global org of account managers who are teachers and consultants and relationship stewards and above all else product experts. Hey, guys. I'm Jed Diederich. I am the RVP of Business Development. So I lead our sales and biz dev efforts out of the New York office here. I started right after these 2 here about 6 years ago and was the first on the ground salesperson at The Trade Desk. So the biz dev team is the tip of the spear. We go out, we identify clients that we want to work with, we onboard them from a contractual standpoint, and then we help grow the business. Awesome. And just the thing I'm most proud of, so many things to be proud of, you heard Jeff talk about our culture and a place people want to love to come to work. So all three of these guys have been here for more than 6 years. They really were the start of our New York office, which is, of course, our biggest office. And their careers have blossomed and they've become such amazing leaders in our company. And I'm just really excited for everyone to get to hear from them because it's not something you guys get to do every day. So let's jump in. So we regularly talk about enabling our clients. We've heard some of our clients talking about that. So Kathleen, what do we really mean when we're talking about enabling our clients? Yes. So I really think that it boils down to we aim to make our clients heroes. With many of our top partners, we're in their offices every single week. We're answering their questions. We're reviewing campaign performance. We're making recommendations. We're giving suggestions. But above all else, we're making sure that our platform is delivering results for them so that they can deliver results to their clients. And then at a more macro level, many of our clients have quite literally built their business on our platform. They've been able to go out and pitch and win new business and support new channels like audio and connected TV and video and new markets, as Barry mentioned earlier, through our platform. They've been able to make the pie bigger for their agencies, so to speak, because they've built their business on our platform. Mark or Jean? Yes. So I think to me, what enabling clients means is so one of the things that Kathleen touched on was education and training. And so I think for us, a big part of that is teaching our clients, the industry, and teaching them how to trade, use the platform and just optimize media because this is a changing industry and kind of a new way of thinking about things. And then I think in addition to that, it also is more from the technology side. So because we're talking to clients every day about their key performance indicators and how they're using the platform, we can understand what their pain points are. And so we can take that information back to the product team and help us build better product to figure out whatever the hurdles are that they're having. You guys touched on a few really great things. And Kathleen talked about how our clients are actually building their business on top of us. And so I think enablement is that big, and it's also really individual and personal. So it's the people who we meet and train at the very start of their careers. We see through tough times like Q4, which we're in right now, and we see them through to their promotion where they go from a junior trader to being the head of a bigger trading team at a midsize agency to then you know, we see so many times they come back around at a consulting firm who's helping a brand decide what their model is going to be or they're on a programmatic steering committee at a brand. So this skill set is so valuable and this community is small enough that these people reemerge time and time again. Awesome. Let's stay with Jed. So talk to me a little bit more about how we at The Trade Desk impact our clients' programmatic strategy as the markets evolve and it's become more important. So Yes. I think about a few years ago, when we started our relationships with some of the really big agencies that are now more at scale spending 1,000,000 of dollars a month with us. And the way their media plans looked when we started working together, and programmatic was one small line on a much, much bigger plan. So they had their programmatic strategy, maybe they were running like a display media campaign with us for $50,000 And then they had a banner campaign with Hearst. And then they had a YouTube campaign. And they had a Facebook campaign. And they had like 12 other lines like a video campaign. And that was that small $15,000 test that Rob talked about earlier. And what we impact and what we help them impact and do is capture the rest of the media plan. So go from a line on the media plan to the means of executing the whole media plan. So we're helping our clients capture that, make marketing better by connecting the customer experience across all of those channels. Yes. And I think another area where we gain ground is in delivering and helping agencies deliver insights about campaigns and about audiences. So one of my favorite examples was we were working with an agency and their advertiser who is a luxury automaker for a campaign. And the goal of the campaign was to have people go to the website and build their dream car and then click through, put their zip code in and find a dealer. And they were getting a lot of people who were coming to the website to build their dream car, but they weren't getting a lot of find a dealer like actual qualified leads. And so they were looking for help. And we were able to help the agency find that the vast majority of people coming to the website and building their dream car were college staged men, who maybe didn't quite have the income yet to buy a luxury car. Not that they weren't important to the brand in the whole life cycle of customers to the brand, but they weren't the customers that they wanted to be talking to for that specific campaign because it was about qualified lead generation. So what we were able to do was eliminate that waste, divert budget away from non qualified customers and spend the budget on people who actually were intending to buy. So even though we arguably cut the audience or made it more targeted because it was more successful, more targeted at who their customers really were, we're able to spend more marketing dollars. So you might say, well, you cut the audience by 80%, how did that happen? But targeting and performance wins every time. Exactly. And actually, just one thing to add to that. I think also, our independence gives a lot of the brands that we work with the comfort to give a lot of their first party data to our platform to inform a lot of that modeling. So I think because of the trust that we've been able to gain with agencies and advertisers, allows us to go out and find some of those cool insights. Perfect. And so tacking on that, I talked a lot about growth during my prior time on the stage. I think we'll grow I believe we'll grow faster than the rest of the industry for as long as I can see in the future. So to you guys, to make it real, a day in the life or a month in the life, what is the blueprint that you guys think about? How do we win more spend on our platform? Why do clients spend more with us? Yes. So Mark just mentioned it and it's trust. I think it was Brian who called out earlier. Many of our clients start their day in our platform. The very first thing that they log into is our platform. The first phone call that they make is to their account manager if they have a question or a one off. They trust our team as advisors and consultants. We actually we had a client who recently won an award from the DRUM for best digital trading team. And immediately when they won the award, they e mailed our account team and said, thank you so much. We could not have ever won this without the training and education that you've given us. We would not have been able to do this without our ad tech partners at The Trade Desk. Mark or Jade? Yes. I mean, I think so it's about trust, but it's also about credibility. And so I think a big part of the way that we grow spend on the platform is oftentimes building the trust and the credibility through that first initial campaign. So typically, an agency will come to us, they'll have that first test budget that might not be the biggest budget in the world, but it's their opportunity to get to know us. Through that process, we make sure that whatever their goal is that they're trying to optimize to, that they are able to hit that goal very effectively, which gives us more credibility to then go talk about new channels. And so maybe that first test campaign starts out as desktop display. And if we are able to hopefully crush that goal in the initial test, then we can start to go talk to them about cross device targeting through and layering on mobile in app inventory, and that can increase their reach and decrease the cost per action. I think what my team is doing, as we're helping our clients grow that trust and credibility to themselves and grow the pie within that advertiser in that first test, we're walking down the hall and finding the next team that we don't work with yet. So the next brand that isn't on our platform within that same agency and saying, your neighbors won an award by working with us. And we were able to really grow and show a comprehensive media strategy down the hall. So let's kick off another $15,000 test. Great. And so I think Dave talked a lot, probably did as good a job as we've ever done in rooms like this, talking about our technology and why it actually is different and matters and drives better performance. But you guys are the business team. You're talking to the clients every day. So how do you guys think about what we do is different from a technological perspective? So Mark, your teams live in the platform more than any other. So why don't you start? How do you think about the technology differentiation of The Trade Desk? Yes. I mean, I think from a trading perspective, I think there's 3 things. So bit factors is definitely one of the core differentiators. So we are the only DSP in the space that has that fundamental architecture, which gives traders a lot of control that they don't get through other systems. In addition to that, I think our reporting and transparency is also a huge differentiator, which when I first started out in this industry seemed kind of crazy that we'd be talking to these agency traders who are running media on behalf of Fortune 1,000 companies inside of other platforms, and they're just not getting any reporting. So they're not actually able to see where their media dollars are being spent. A big part of our technology differentiators is our reporting. And so we've always wanted to tie reporting and insights to action. And so I guess that's kind of the second thing. The third thing I would say is workflow. So because we are buy side focused, a huge part of our product offering is around workflow. And so we take we make it really easy to go from initial campaign setup to pulling reports and finding insights and then actually making those insights actionable through the platform. I would add our agility. We released dozens of products throughout the year, and we integrate with dozens of inventory and data providers. We actively seek out feedback from our clients to inform our product roadmap, and our clients know that they can count on us to deliver what they need. I would also call out, and this was mentioned earlier, our open APIs. That's a massive differentiator, especially in the context of or in contrast to the walled garden approach. We encourage our clients to build to the gap at the top of our stack, and we encourage them to build in their own proprietary advantage, and that is unique. Jed, you're often the 1st person to go talk to our clients about The Trade Desk, and of course, technology is part of that conversation. So how do you introduce it and how do you frame that for clients? Yes. I often start with our business model, which I really do think is fundamentally different than all of our competitive set. And that is the foundation for trying to build trust after that. So what does that mean really on a day to day basis? Mark talked about transparency. That's the difference between someone asking you, where did I place my ad and me saying here or I won't tell you. So you can imagine how we earn trust by showing them absolutely everything that we know. It's also not being a media owner or a content creator or an arbitrager. So we don't go out and pre buy media and then not represent our fees. So when we're coming and saying, we recommend a different channel allocation or we'd suggest shifting some money out of your digital audio campaign into your CTV campaign, they know that we don't make more money by selling them one ad over another. So that's just the day to day trust manifestation, I would say. Awesome. Let's stay with Jed and let's stay on the BD front because folks are in this room is particularly interested in our growth and again the stuff we talked about. So let's talk about getting new agencies on to working with us and then brands within agencies that we already work with, how do we win them over? So I have a few questions. I'll start with Jed and then go to the rest of the panel. So first, how does the process work of going and winning a new brand, Jed? So in terms of finding a new agency partner, we're I've been a part of a few sales organizations. And what I'll say about The Trade Desk is that we're picky about who we want to work with. We're looking for agencies who are going to really own the capability and take it in house and then go win more. And so we're starting those conversations. We're participating in really in-depth technology reviews. We're demoing the platform. We're getting through a contract stage. And then once we get in there, we're mapping out different parts of the organization. So who are the decision makers sitting on top of different brand spend, on different channel spend? And then we're setting up meetings so that our bigger business team can go in and start that long growth curve. Cool. And then touching on what Brian and I touched on a little bit earlier. Are you pitching agencies or brands? Or is it a combination of both? What does that look like today? Yes, sure. So because we're so focused on the self serve clients, we're always starting with the agencies. Agency trading teams are the cornerstones of all of our relationships. But what's happening more and more is those guys are leveraging our partnerships as we build trust to walk us into their clients. And they're using that to win more spend for themselves and make different channel recommendations for their marketing spend. So I think of a recent example of a major bank, where an agency had set the stage for wanting channel consolidation, platform consolidation onto one platform and wanted us to come in and actually pitch the brand with them. And by the time we got there, the brand had already actually gotten half of our pitch. So it was a much easier job for my team as you might imagine. So in that conversation, obviously, sort of 2 heads are better than 1, and all three groups sort of have their own value. We were sort of preaching to the converted because the agency believes so much in the way that we our business model and our technology, they had done a lot of the sales job for us with the brand, and they were just really interested to meet our team and learn more about us and how we could help. So that's a great example. So last question for Jed in terms of BD. What's the onboarding or excuse me, what's the average sales cycle like? We get that question a lot in this room. So Sure. I wish there was one answer. It would make my life a lot easier if there's one. Consistent answer, but the reality is that it really varies a lot. We have sometimes it's a client who's moved over who already really knows us. So it's a matter of days. Other times it's a major brand initiative to identify a new technology partner, and that can literally be more than a year in the process. But what I will say is, it's not the contract signature that we think about in terms of our sales cycle. It's the long growth curve that happens as we go from that $15,000 test to the prospecting budgets that Rob was talking about earlier to new channels and new advertisers. And that's what happens in the months quarters years after my team has really identified that opportunity. Awesome. So we won all this new business and then Kathleen's Client Services and Mark's Trading and Analytics. What does onboarding look like? I've put that same slide up for 15 months and it never I don't feel like it comes to life. Just too hard to describe. So can you guys bring that to life? You guys live it or your teams live it every day? I think we would say onboarding is where we shine. We onboarded a very large advertiser alongside that advertiser's global agencies this summer and every single person involved at the end of the transition said that was the smoothest transition we've ever seen. I think it was because we were so globally coordinated, but also because we had such deep seated relationships with each of the agencies involved. But to take it to the ground level and talk about what the onboarding process is actually like, we're in the office, We're on the ground with over a dozen traders setting up over 100 campaigns. In some cases, we're talking about campaign goals. We're going over best practices. We're educating. We're training. We're walking through our platform. For the first couple of what we call office hours, we may come to the table with some suggestions. For the next couple, we'll say, hey, you guys come to the table and bring us your recommendations, and then we'll talk about, yes, I would like to see you make that change. I think that'll make a great impact. But always, always, we're driving to a place of self sufficiency so that the agency can self sufficiently work in the platform. And that's why any given account manager at The Trade Desk can work on a portfolio of different clients because we teach and we train in the effort of making the agency completely self sufficient. And I would just add to that. So I think the key thing from a trading perspective is the initial investment. And so a lot of times during the onboarding process, they'll come to us with that initial test campaign. And it's in onboarding that campaign into the platform and the initial optimizations where we not only teach the client how to use the platform, but we also win their trust and the credibility that if we're recommending something to them, they're going to see success with it. And I think that really is the key point. And then I think also is just the self sufficiency side of it. And so we will be very hands on at the beginning just to ensure that they are successful. But then the goal the ultimate goal is for them to do it themselves. And so the first couple of weeks might look like us providing the recommendations to them. But then after the next couple of weeks and once they've sort of gotten a sense of how reporting and optimization works within the platform, we'll ask them to come to the table with recommendations and then we'll just review them and see whether we think they're good or not. Yes. We literally have a mantra inside of our company that is we have to go teach the world to trade, like it's literally that. And we've done that thousands and thousands and thousands of times. I think one of the panelists on Brian's panel, one of our clients said, hey, we invest greatly in trying to increase the pool of labor, increase the pool of people who have that skill set. And I think we spend so much of our time trying to help them get there faster. That's a big core part of what we do. So another sort of hot topic, I guess, continue to get a litany of articles and various sources about brands bringing programmatic in house. It's in house guys. What are the agencies going to do? What's the trade desk going to do? But I think Brian and I both spoke to sort of the real situation. But you guys are talking to clients every day, you're living it. What does it look like from your perspective? Are brands really bringing this in house? I'll start. I mean, I will say that the grain of truth to all of this talk is how much more seriously that brands are taking programmatic. So they really do want to know us. They want to call us. They want to talk to us. Sometimes they want to own the contract with us. And that's because of how when they recognize the value of programmatic and they recognize how much data they can get out of it, they do things like hire a programmatic expert or sometimes establish a programmatic steering committee of a few people within their marketing department. And that part is really true. The part that's always overstated, as Brian talked about, the black or whiteness of actually staffing out people to do pixel implementation, creative QA, all of the day to day management and reporting and infrastructure that is required to actually tie this into the rest of your marketing channels. So I think that part is often overstated. We usually either see a few hires or somebody take a big swing at this and then 8 to 12 months later completely come back around. That's happened a few times too. Yes, I'll say this. When we go into an agency and we see them handling everything from creative upload and QA to campaign activation to billing reconciliation with vendors to workflow management, planning, providing their point of view on marketplace initiatives. There is so much that goes into this. I can personally vouch for the amount of work that agencies do because I used to work at an agency. And there is such a level and depth of specific expertise in activating and managing media investment on the part of the agency. And that is simply not easy to replicate in house. I think an agency and training traders on behalf of, an agency and training traders on behalf of a CPG client in the same week that there were AdExchange or Ad Age articles about how this client was actually taking things in house. And so I think, yes, not quite black and white. I think that was like something like 12 months ago, and you were training 15 or 20 traders at the agency to work on that CPG. And I think if you went in there this week, there's like 40 traders. So the investment has only gone up on the agency side, not down. So cool. So let's talk about what are some of the unique characteristics that you guys think of that made The Trade Desk a successful company as it is today. So how do you guys think about The Trade Desk and what we've done and why we've been successful given how close you are to our clients? I have to point to culture because I mentioned I've worked on a couple of sales teams. This is the only sales team that I've ever been a part of where it's like a pretty common occurrence to hear people say it's the best job they've ever had. And it's also the only one that has not been part of a company that is sales first. We're not sales first at The Trade Desk. We're product first. And as a salesperson, I really appreciate that, because our team goes out and feels really good about the product that we represent. And I think that, that comes through for everybody across our business teams. Mark? Yes. I mean, I think there's so to me, I kind of think of 2 things. There's the internal cultural side of it, and then there's also the technology side of it. And so I think from a cultural perspective, there's any number of things that I think have made us successful. I think maybe the biggest is humility. And so I think a big part of what our job on more the business side is to make our clients heroes. And we like to hire smart, creative problem solvers, but we also like to hire people that are okay giving the glory to someone else. And I think that's a huge part of our success. And then I think on the technology side, I mean, a couple of people have referenced it today, but we really do innovate. And so we have 45 or 46 different releases every year, which gives us a ton of credibility when we're talking to clients about whatever problem it is they have and whatever they need in the platform. We can actually go and say like, Okay, well, in one of these 45 releases that we're going to have over the next year, we'll definitely get that in for you. We have a singular mission. We want to be the very best buy side technology platform out there. And I think when everybody is driving toward that same goal, toward that same mission, everybody is rowing in the same boat, that's really unique and special. And I'll also say from a personal perspective, The Trade Desk has had the single most positive impact on my own career out of any other place that I've worked. And I couldn't be more committed to making sure now that every single person on my team has that same experience. And I also think that, that is extremely unique and differentiated and very, very special. Great. So with that sentiment, last question, and this is probably even more personal that. So at the end of the day, what makes you guys successful in your roles and what do you value about it? So why don't we start with Kathleen? When my incentives and the incentives of my team are tightly aligned with the business objectives of our clients And my clients are seeing growth and success because of the expertise of my team. And my team is having fun doing it, and they're learning, and they feel like they're developing new skills, that is success for me. So I think for me so due to the nature of what my role is and what my team's role is, we're very close to the data and we're very close to the execution layer. This is an industry that is completely changing the way advertising has been bought, which means there's a lot of room for innovation. And one of the things that I've really enjoyed the most about this company besides the people and everything else is just the opportunity to innovate and do cool things. And so to me, success is seeing the team that I've now started leading be also excited about that and excited to come into work every day, excited to be actually playing with the data and changing the way media is bought and innovating. I think that's pretty unique and pretty cool. When I think about what success means for the biz dev team and for me, it's building this scaled sales and biz dev organization that looks really different from different media sales organizations that I've seen in the past. And it's full of people who are really great listeners and who are truly consultative sellers. And who are the start of that conversation, this long conversation that we talked about, about the value of independence and full alignment with the technology provider, but also people who are able to see that through. So they see it from the intra stages with the programmatic trading teams into the investment teams, and then as we talked about, into the brand teams as well. So helping everybody along the way see that value and improve the marketing they're doing. Awesome. Thank you guys very much. You've done a personal plug, an extraordinary job of building the trade desk. The 3 of these folks have as much credit as any of the C levels you see on stage today. And I've never been more humbled or proud to work with folks than these 3 in my life. So thank you guys very much for bringing The Trade Desk business to life, I hope, and hopefully that was helpful for everyone out there listening. So thank you guys. Thank you so much, Kathleen, Jed, Mark and Rob. Our next present in our last presentation before closing remarks is our CFO, Paul Ross. Paul? Good afternoon, everyone, and thank you for coming, and thank you, Chris, for everything. I joined The Trade Desk 3 years ago back in 2014, primarily focused on building up the company's infrastructure to get ready for life as a public company. So I'll take a couple of minutes just to discuss where we are today, what levers we have for growth in the future, and also how we think about cash and working capital. Everything you've heard earlier from Rob and Brian and Jeff and Dave are really consistent with the financial results that see now if you're not already familiar with them. So the programmatic industry has been tracking to roughly a 33% CAGR through 2017. Magna Global says 33%, IDC says 37%. And we've consistently outgrown the industry by at least 2x. We've grown much faster than the industry, but also much faster than the other SaaS companies that we're often compared against as we continue to win market share, both due to our differentiated technology, which of course yields higher ROIs for our clients, and of course, our business model, which is conflict of interest free. So, 2 stage. So clearly, we hit an inflection point with operating leverage, and we've had really solid operating margins since 2014. This is a result of the business model, which of course is self serve, which means we don't need to increase operating expenses commensurately with revenue. And I think as Rob said earlier, we don't have to double our staff in order to double our business. And importantly, I should mention that these are results from periods where we've been investing really heavily for future growth, given the massive cam that we have. So in terms of profitability, the first spend on our platform was in 2011, and of course, our Q1 of profitability was in 2012, again, a testament to the tech and the business model. And we've been producing real GAAP net income every year since 2013. So when we look at the latest 2017 estimates from NASDAQ First Call, and this is how we stack up against other SaaS and tech companies in our world. And while our revenue growth rate, of course, reflects the secular shift going on in the advertising industry, as well as our continuing capture of market share, it's interesting to note that our EBITDA margins put us in league with the largest and most profitable players in the business. So we've been growing by faster than the industry in a number of categories. And there really is no reason why we can't continue to outperform the industry for the foreseeable future. The channel growth and the international growth, in particular, are two areas to keep an eye on. And while mobile and video continue to grow rapidly, we expect the newer channels like audio and, of course, connected TV to provide the fuel for many years of growth in the future. And of course, the platform is super sticky. Once clients come to our platform, they never leave and they continue to spend more. So a couple things to highlight from our long term model. As Jeff and Rob discussed earlier in the presentation, we expect the take rate to be relatively consistent where it has been between 15% 20%. And while we've been trending on the upper portion of that range, we would expect that the rate will move down as more spend comes onto the platform and as volume discounts kick in. But we also sell data and other value add products that make up for that. Also, we don't see a whole lot of take rate compression because we already have one of the lower take rates in the industry, and we continue to add more value to our clients than we charge them. Over the long term, I don't know if you can see it behind the chairs there, but we're targeting an adjusted EBITDA percentage of 40%. And for 2017, we forecast an EBITDA margin of 29%, which, of course, reflects all the growth investing that we're doing, especially in international markets and for additional features on the platform. So as far as cash goes, this is a question I get a lot. So in addition to the seasonality and the volatility in cash flows that we see on a quarter over quarter basis, It's true that in general our payment terms are such that we pay for inventory before we collect, which has recently consumed about 20 days worth of working capital, but we've also continued to generate positive cash flow. And the reason for that is because our EBITDA margins are higher than the working capital consumption. And so as our EBITDA margins continue to increase toward that 40% long term, you'll see more and more cash flow generated. And in addition to that, Jeff referenced the gold standard earlier, which also includes other features that more align the payables and the receivable cycle. So we should be squeezing that 20 days down over time, which, of course, the result of that will be an increase in cash flows. I think this is my last slide. I just wanted to highlight our venture capital ownership since the IPO. When we went public, I think almost 14,000,000 shares were owned by early investors. The blue line there is the VC ownership going down as we've taken them out through the follow on offerings that we've done, all the while the share price was increasing, too bad for them. So where we stand today, right now, I would estimate that less than 10% of our shares are held by those early investors, and that's inclusive of the 5% stake that Highwind currently holds. They're the gray line across the bottom of the chart on the right, and they've indicated that they are super, super long term holders. Thank you very much. Thank you, Paul. Now for closing remarks, we'll have Jeff Green. Then we will just take about a minute, have management team come up and go into the Q and A. Jeff? That was started by our team. Thank you, guys. I'm going to get us back on track in terms of schedule. So we're just maybe 10, 15 minutes behind and I'm just going to take 2 minutes to wrap up. So first, what we've tried to do today is just be really transparent about our business and share with you the way we think about our business. So I'm sure at times you've wondered what the acronyms mean. And I know we've gotten into some of the specifics that might have tested the limits of what you know and understand or even care to know and understand about digital advertising. But we wanted you to see the world through our eyes. Hopefully, we've given you a little bit of glimpse of our culture, as you've gotten to hear from almost 10 of the members of our team today, and how much they love being here and how much optimism we have for the future. But I also just wanted to highlight, in the last year, we've seen amazing performance in the stock, but that's never been what we were after. We've always been about creating a long term business, and we just believe that if we focus on the long term, we do the right thing for advertisers, we do the right thing for our team, that, that stuff will take care of itself. And there's no possible chance that we're in this amazing and surreal position if it weren't for the shareholders who believed in us at the very beginning, the employees that make up our 650 person company, our amazing Board of Directors. And now I just wanted to say thank you to our public shareholders as there's been a changing of the guard over the last year for most of our existing shareholders to many of you in this room as well as those of you listening. I just wanted to say thank you, because it's not possible without you. And I believe that the best is yet to come. And we have our very best and brightest days ahead of us. So with that, I'll welcome the management team back up and we'll start the Q and A. I'm going to stand because I think we only have 4. And Mark and then actually, Chris, you're in charge of question allocation, but I'll start with Mark. All right, I'll deliver it. We need to well, we need to get one out there, right? Are you going to do this? Is that right number? Who's going to? Oh, you 2 are. There. Yeah, we'll get this started. So Jeff, two questions. First is, can you talk about any expectations you want to set in terms of being able to get through to some of those walled gardens? I don't know, give it Facebook or Snapchat. Like what do you think your odds are? When we could actually see that happen? And secondly, just talk about the risk of Trade Desk being out Trade Desk. So as I listened to how you're going with agencies to advertisers, it sounded great. But at the back of my mind, I remember a few years ago, one of the things that really differentiated the story was that you weren't working with brands and some of your competitors that did pissed off agencies that created your opportunity. So how do you what kind of guardrails are you trying to set up out there for your employees so that you don't make that mistake that your those companies did in the past? Yes. So I'll go in reverse order and make sure that I well, actually, I'll start at the top. I mean, if I miss anything, just me. So in terms of social networks and walled gardens, which most of them are the social networks and nearly all of the social networks at the top, Facebook and the smaller side, Snapchat, they're all walled gardens. I think they increase in likelihood to breaking down the walls based on size and desperation. So the less profitable or the less growing you are, the more likely you are to take down walls. My prediction is that at least one of the major social networks will take down their walls in 2017 or 2018. So in the next 15 months, somebody is going to take down walls. And it will set a trend for the others that are reluctant to do it. And frankly, the only reason they've endured the pain this long is not because it's not in their best business interest. They're worried that strategically once they take them down, they can't go back is I think the concern. On the agency direct side, and I'll let Brian add anything that I miss on this. It's not a change in strategy for us to get closer to the brands. We're doing it arm in arm with the agency. So I don't believe there's any incremental risk of any consequence in getting closer relationships with the brand. And in fact, I think it's important to get to the next level, which is getting the brands to trust both of us with their data to a greater degree. So part of what I think we can learn from what P and G did with Audience Science, where Audience Science had taken an approach a few years ago, which was very anti agency. They went out of business despite having the biggest spender or biggest advertiser in the world. That model just doesn't work even when you have the biggest. So if you can give many of the tasks that give you scale and require lots of labor to the agencies and still let the brands have some amount of market intelligence in them and continue to build up their expertise in data science and then leverage that through great technology and make that available to both of them and giving them reassurance about the way the data gets used, we think that's the best way to operate in the future. But it doesn't really change anything about what we do. I would add to that. Like, you could ask any of our agency partners. For years, when an advertiser reaches out to us directly, our entire sales team knows, hey, before you have a conversation, do anything, go check it with the agency. And so we'll routinely go to the agency and say, hey, this brand reached out, how do you want us to handle it? And 95% of the time, they're like, great, good, that's good for all of us. Go talk to them because they're comfortable that they trust us at this point, that they know we're not no one's trying to cut them out of the picture, we've developed that trust. But we also have our sales team follow pretty strict rules around, we're really loyal to our clients, we don't want to change that in any way. Jeff, I had two questions. First one on Connected TV, you talked about that being the largest channel in 5 years. How do you think about the progression from here just the ramp in 2018, 2019, 2020 to kind of the 5 year goal? And then in terms of data, you talked about also the 5 year goal of growing usage for data at twice the media spend. Can you just talk about what opportunities you see in data for you guys to add? How do you think about the monetization opportunity for that as well? Thanks. Yes. So the only way that we do that we make connected TV the biggest channel in 5 years just because of the growth rate of all the others and because of the sort of head start that they have in just mass is if there's an acceleration in total dollars available. So it's because we're looking at that 10x increase in inventory that we're betting that we're right about consumers are tapped out on the number of subscriptions that they can pay that don't include ads. And that consumers over time, especially as the ad experiences get better, are going to prefer on average to have an ad funded model that costs them less than not. So if you believe that, then you're going to continue to see inventory increase, perhaps not at 10x just because that was a small base and 1,000 percent is a lot to keep up with. But if you keep up in total dollars growing at an accelerated rate, then that can happen. And we're betting that it is, otherwise I wouldn't have made that assertion. The same thing on data, which is we've just scratched the surface. And you heard from some of our clients about how 3rd party data has not been priced right for most of the history of programmatic, where Barry covered it really well, which is it wasn't uncommon for people to try to layer $1 worth of data on top of $0.30 worth of social media. And you just can't enrich the ad that much with basic third party data. So the data has had to get better, which also means finer grain. So the era of using age, gender, income as a primary data source is over. And instead, if you can use more intent data and even evolving from this is an auto intender to this person is in market for a Toyota, that level of grain is making it so much more effective, making it so that people are willing to pay more. But most importantly, the amount of data exhaust that all of us create is just going up massively and it didn't used to be collected at all. So it's only in the last few years that people have really started collecting it and thinking about how they can use that data. So the amount of sort of actionable data has gone up so exponentially that it's ridiculous that we wouldn't use it. So of all the numbers I shared, like maybe the one that could be more aggressive is that one. Hi, Youssef Sculley at SunTrust. Thank you very much for doing this. It was very insightful. Two questions. 1, first starting with China. Can you speak a little and I know it's very early, but can you speak to the competitive landscape that you're seeing there right now? And in order for you to achieve the goal that you've outlined, what is the next progression of your relationship with local players? So you have a deal with Baidu. How quickly do you think you can get the other players, the Alibavas of the world, the Tencent, etcetera? And how important is it for that goal? And second, can you speak to the potential impact of Apple's recent move to eliminate or the basically the cookie tracking from the new iOS? And how does that impact your targeting abilities? I guess it's really impacts the whole industry, but for you in particular. You bet. So I'll try to remember them all. If I miss any, please let me know and I'll chime in on the China questions and then I'll answer the data if you can then supplement with anything I miss. So on the China front, how do we expand from where we are in order to hit those big goals? There's no question that we have to get closer with the local players. So we have a great relationship with Baidu. We have a growing relationship with Alibaba and Tencent. There are no formal deals with either of those to date. I'm definitely spending more time in China. I mean, I'll be in Shanghai next week. We will get closer with the major players. As it relates to the competitive landscape in China, the second or maybe it was the first question you asked, one of the questions you asked. The way that I would say China has developed in the state of the market right now is something more like 2,008 or 2009 in the U. S, which is there are a lot of ad networks. In fact, DSPs there are not called DSPs, they're called D SPANs, which is DSP Plus Ad Network, AN, DSPAN. And they refer to themselves that way. And it's not uncommon in their pitches and the way that the traditional DSPAN pitch goes is, we're a DSP, we're an SSP, we're an ad exchange, we're a data company, we're a mobile company, we built the following 27 apps. Like they do everything. And there hasn't been a lot of trust on sharing data. And part of the reason why RTB or programmatic sort of price discoverable programmatic on a per impression basis is not very big there is because there has just been concern about sharing data. So one of the things that makes the opportunity so great right now is that the big companies are just swimming in inventory that they need help monetizing elsewhere, which by the way is a commentary on what happens to walled gardens eventually here, but they're swimming in inventory because you can't monetize it all yourself. So then they're asking the question, who can I trust? All these companies are trying to do everything. Then there's these multinational players who have a proven track record of working with the sell side, who don't own a single cent of media and represent some of the largest multinationals in the world. That's where we step in and partner with them. As it relates to your Apple question, I don't think it has a significant impact on our business. And it's largely because it all hinges on how intelligent you think the tracking actually is and how aggressive Apple is actually going to be about preventing cookies. And I personally think this feels a little bit like what Do Not Track was, which was the press was way louder than the outcome or the impact. And I think this one looks a lot more like that. But I think we're in a better position than anybody to sort of respond to it if they decide to be aggressive. So I don't think it has a major impact on our business. I don't know whether you would elaborate. Yes. Two main things to me are overall the size of the impacted segment is not that big. So like it's Safari like we've also it's Safari Mobile Web, so there's also in app, there's also Chrome Phones, there's like there's lots of ways to go out and do marketing, so there's outside of it. But then also taking away the cookie doesn't necessarily mean that we can't market. And we're actually better than I think anyone else at making a great campaign thing even when we don't have a cookie. And there's a lot we can still do. And so I wish they wouldn't do it, but I don't I also don't think there's a big impact from it. And I think it's actually in some ways an opportunity for us to innovate and continue to be the best at dealing with that environment. So just to put some numbers to what he said. So mobile market share for Safari is less than 4%. So it's just not big. So we would miss out on a percentage of mobile web, which is really small. And just to be clear, the effective mobile advertising is in app, because all of us know when you're on mobile web and you see those tiny little banners that you accidentally click on, that's not the most effective advertising you've ever run into. And then Safari, all that market share I'm sorry, Safari Desktop, that market share is also quite small as most of that has gone to Chrome and Google, who owns almost 70% of the market share alone, because of the fact that they're in the business of making money off of advertising. So I'm actually not certain that if Apple is too aggressive, I personally think that they may lose market share to more market share to Google because many would argue that Chrome has a superior experience already. But if you're punishing publishers for making money, number 1, they're coming after you, which could create a negative impact on your in app marketplace, which is totally dependent on publishers. And if they start to say download Chrome because we make more money in Chrome, then perhaps Apple has been too aggressive. And then if they do the same thing, then Google is going to be aggressive about doing that too. So there's a scenario where publishers as well as Google are working to combat an overreaching Apple from negatively impacting the Internet experience. Or I'm right, and it's not that big a deal. Those are the only major outcomes, I think. Two questions for you guys. 1, Dave, can you talk a little bit about the architecture and why Google or others can't replicate that over time? And then the second question was really about the walled gardens, but really Amazon in particular, Jeff. As you think about 2, 3, 4 years from now and Amazon's media business is not $2,000,000,000 but maybe $5,000,000,000 or $10,000,000,000 How and then the DSP business, not just on Amazon, but off Amazon, how does that compete with you and others? Awesome. You want to take the first part and I'll take the Amazon question? Yes. So sure, they could replicate. A lot of people would replicate it. I talked a lot about it. But I also mentioned in one slide that it's very difficult to do. And so I feel a lot of confidence in just the number of road map road bumps that are in front of anybody trying to come down the path. And then also like my word count today is overemphasizing like one area of our platform. We do lots of stuff. And so that's kind of like one thing that we're good at. But there can And like with the speed that we release product, like we continue to innovate. We're not sitting around counting on that to be our sole defense. So Brian, on the Amazon question, if I miss anything, I know we talked about this earlier, feel free to clarify or add punctuation. So I think there's 3 or 4 companies in the world that every company in the world has to be careful of and watch. And perhaps none of them are more worthy of watching than Amazon, right? I mean, they're in the cloud business. They build phones if they want to. They win they're in the top 3 Emmy winners. I mean, they build TV programs. They're the largest retailer in the world. So I mean, it's so surreal to sit in Board meetings with the CFO of Netflix and us both talk about watching Amazon, right, despite the fact we're not even in the same business, right? So everybody has to watch Amazon. But the reason why I don't think that this part of Amazon's business is going to be big is for that exact reason, because I've spent enough time with CMOs to know that they're afraid of Amazon. They're terrified, especially you take companies like CPG Companies who are now selling their products directly on Amazon because they look at Amazon with more fear than they ever did a company like Walmart. So those same companies who we talked about earlier that in order to compete with Dropbox have to activate their data, that in order to be good at it, they're going to have to give their data to the company they trust to buy all their media. And so I just don't think Amazon is likely I think they're even less likely than Google or Facebook to get the trust of the biggest brands in the world with their data. So when I make my list of companies that I worry about, Amazon is probably number 3 on the list, not even number 1. And it's because on most of my brands that come through our agencies, Amazon is their number 1. So it makes it so I worry less. I would just add that agencies and marketers are not sitting around asking for more walled gardens. They're asking for fewer. The fragmentation is a big enough problem, trying to frequency cap, get insights, attribution across everything you buy. Adding another layer of that where you can't get visibility is not something you hear a lot of. It's the opposite. Exactly. Okay. Hey, Jeff. Two questions. I guess the first one is, if Amazon is the 3rd company you're worried about, who's 1 and 2? Google, Facebook. Okay. And then the second question is, can you talk a little bit about market consolidation? We've seen some of the early ad tech companies get taken out. How do you think future consolidation looks like? And if I lead you a little bit, do you think about AT and T, Comcast, maybe even Disney coming into the advertising market? And if so, is there any risk there to The Trade Desk? So I think there's no question that companies like that are going to get into the advertising space. So I think you're going to see a meaningful amount of those, the AT and Ts, the Comcast of the world get into the space. I mean Verizon already did. They'll do it. Singtel did when they bought a bunch of ad tech assets recently, roughly $1,000,000,000 worth. And I think you're going to see the same thing from the software guys too. So at some point, Salesforce, who the way I understand it, they meet regularly, they create a list of what are all of the opportunities of using CRM data to get into other spaces and advertising and marketing have been on the list for a very long time that they keep just moving it down. But at some point, they get to that. At some point, SAP may get to it. At some point, IBM says like we should finally do this. At some point, Oracle says, okay, we took a big bite in data. Maybe now we want to get into media. I don't think all of those companies necessarily do it, but I think many of them will try. And then it's a question of did you get in too late? Did you buy the right tech? Did you put it all together? Can you move fast enough? Can you not mess it up? If your IBM and your track record in M and A isn't awesome, like can you move in this really fast paced space? And then most importantly, with every one of those companies, can you be objective enough, right? So if you're compromising on objectivity, you better be giving them something better than Google or Facebook. And that's the open question on all of those companies. Can they be agile? Can they execute? Is advertising enough in their DNA as well as are they objective enough? And I put us in a great position against any of those companies for the same reason that Mark Zuckerberg says anybody could have built Facebook. We won because we cared more and we were more focused. I just hopefully, we've demonstrated today that we care a lot and we're pretty focused. Hi, Dylan Haber from RBC here. Just a quick question. Earlier you highlighted that you expect the tax rate to sell out from 20 Yes. So one of the reasons why I put that tenant there, something that we had 10 years ago, is because when I was at Microsoft, I remember one debate in particular where I was on the strategy team for the Online Services division. And we basically said the only way that Microsoft has any chance to compete with Google is for us to predict the future of what this industry looks like at end state and then build to that because if we just chase Google's tail, we're dead. And we basically created this coalition with AOL and Yahoo! And it's probably the wrong word. I'm sure there's an antitrust lawyer at Microsoft that doesn't mind me saying that. But there was some group that we were working together to try to figure out like what to do. And we would debate endlessly about what end state looked like. And there was one point that we all agreed on that at end state, when an advertiser pays $1 the tech fees are going to be roughly a quarter and then the rest is going to end up in the publisher's pocket and that we as an industry weren't doing a good enough job of compressing that yet. And so I just started with that premise because of that. We started our take rates at 15% because we believe that was sustainable until end state. And we've been fortunate to see those grow just a little bit, but they've always stayed in that range. So we still believe that that's the case. That's the premise that we started with. And as we get more and more evidence and 10 years has gone by, it still feels roughly true. Now in terms of the breakdown between exchanges and SSPs, I think you asked a really astute question because, number 1, in our industry, we tend to talk about ad exchanges and SSPs like they're the same thing and they are not and they shouldn't be and at end state, they won't be. And the reason why they won't be is because the publisher needs somebody to focus on yield management for them, where they're figuring out how to segment inventory, what data to use. They're a player in the game, not a referee. And exchanges tend to be referees that don't make a ton of money. It is really hard to make money and the NASDAQ is perhaps a really great evidence of that. You have to be big, really big and then you charge a very small amount. So that remaining 10 ish percent, I think a point, maybe 2, goes to the exchange and the SSPs take the rest. You have to be good at yield management to have any chance to be in the SSP game 10 years from now. Sure. So in terms of what impact data has on take rate and is it included in our take rate? Yes, we included in it. It's really hard to say. When we were on the IPO roadshow, I just kept saying, and I think we probably put too many word count against take rate today. I recognize it can be important for some of your models, but that's not the way we think about it, which is, if I can get more total dollars, more total revenue by having an 18.3% take rate versus an 18.6 percent take rate, I'm going to do it, right? So all we're after is growing the revenue as much and as fast as possible. So data is one of the opportunities to do that. And every channel has slight differences in their cost. And so it bounces around a lot as channels grow. There's lots of differences. So we don't sweat when we see any bouncing around, which is part of the reason we don't report on it. So we don't consider it a core metric. But we're confident that it's roughly in the right place so that we can make decisions that optimize for the long term and frankly optimize to those revenue numbers. In terms of the product that's coming out, really what we're doing is we're changing our whole user experience. And we're reducing clicks. We're changing the UI. We're narrowing the distance between reporting and insights and action. So it takes less work to action and it's easier to connect the dots between insight and action. And we believe that will create double digit percentages at least double digit percentages of efficacy as well as reduce the amount of time that our users have to spend in the platform in order to accomplish their goals. And that overall is going to just continue to distance us from everyone else. So it's been a massive undertaking. There's still lots of work to do, but it will launch it in the first half of next year. I think that was Brian. Sorry, sorry. 2 cooks in the kitchen, sorry. When we're looking at the inventory ramping like it is, could you argue that, hey, that's low hanging fruit, that's Hulu and Sling and there's none sure, you're fleecing and getting that inventory and look at that inventory now. But to grow that 10x inventory, do we need linear TV guys, Fios guys to make more of their inventory available? Are there barriers to entry right now or barriers to tool sets in order for them to show their inventory digitally to you guys? I'll take a stab at it, and then Brian, you should add to it. So I don't think there are meaningful barriers to entry for people making their content available. So we're super excited about the growth we've seen, for instance, on CBS.com, who has what of course, they call FEPs, I know you know, but maybe not everybody in the room knows, full episode players where you can go to cbs.com, watch the exact same content that you would having cable, but the ad experience is different, more data driven, more customized, but it's exact same content. We think more and more of that's going to happen and it already is from all of the major players in broadcast and linear television. So and whether that's at Disney or whether that's at Turner, we think that trend continues. And in fact, they're looking for more ways to create a better experience. And it goes back to that when advertisers are willing to pay almost double the CPM, sometimes more than double the CPM to have data driven connected TV. And it's partly because of scarcity, which won't last because more of it's going to come on, but that's exactly what's going to make more of it come on. So that sort of virtuous cycle that is making it grow, which is then accelerated by a consumer saying I'd rather do it this way, just makes it so that I really think we just have to wait and be patient, not piss anybody off along the way to continue to see it grow. I would just add to that, that it took the last 18 months to really get the pipes in place. Like 2 years ago, when we go to buy connected TV inventory, we'd go to large networks or content owners and they'd be like, oh, we don't have an SSP in place. So we'd give them a tag directly, which isn't the same as buying programmatically. And so over the past 2 years since, we've seen companies like Spot Exchange and Telaria and Google on the SSP side develop better technology if you want to sell connected to base. That was step 1. The pipes are in place. They're more robust now, so that enables more transactions. The other thing I think you'll see starting to happen this year, we'll see more next year, is the buy side. So large agencies and marketers saying, I'll still do an upfront with you at the network or the cable operator, whoever, but I want to transact a certain percentage of it programmatically. I want to bring data. And that has been, A, wasn't always possible in the past and B, they weren't demanding that. I think that's going to change. Great. That's going to wrap up our event. We're right about at 5:30. We do have cocktails and hors d'oeuvres that will be out here for another 30, 45 minutes or so before NASDAQ escorts us out. This concludes the webcast as well. We thank you so much for coming today. We're really humbled at the response and the interest. And have a good night. And please reach out if you have any additional questions. Thank you. Thank you very much.

• Slides