All right. Hello, everyone. My name is Matt Cost from Morgan Stanley US Internet Team. Thrilled to be joined by Jeff Green, the CEO of The Trade Desk.
Thank you.
Thanks for being here.
Excited to be here.
Quickly, I'll run through the disclosures. For important disclosures, please see the Morgan Stanley Research Disclosure website at morganstanley.com/researchdisclosures. If you have any questions, please reach out to your MS sales representative. All right, with that out of the way, maybe, maybe, Jeff, let's start high level.
Okay.
Let's talk about what you're building at The Trade Desk to people who are maybe newer to the story. Give us an overview of what you're doing, where you fit into the ecosystem, and maybe what's changed about what you're doing over the past couple of years.
Sure. So, buying ads has become way more complicated. Almost every industry, over time, gets more complicated, not less complicated, partly because sophistication tends to go up and not down, and that's how you win and not lose. But there's also another problem, which is the way ads used to be bought is you called up a guy at ABC or The New York Times or Clear Channel, and you asked for a rate card, you bought that thing, and that was it. But now, in a digital world, creating essentially an infinite amount of inventory is not hard.
You've all been on pages where there's more ads than content, and the ad at the top of the page where something a little bit different than the ad at the bottom of the page, and it's a lot harder to go to the equivalent of Times Square and see how they all play out on any individual web page, especially because those can be tailored to you, and they can be different for every single user. So when you put that dilemma together, creating price discovery in a trillion-dollar market of advertising, or soon to be trillion-dollar market, is harder than ever. And for the biggest buyers in the world, who have especially made their differentiation by buying media effectively, sometimes just using their size in order to...
For Procter & Gamble to remain Procter & Gamble or Unilever to remain Unilever, they have to buy media effectively. They have to make you love their brand. And so they come to us, and we help them look through, today, 15 million ad opportunities every single second across the internet, across the world, to figure out which, call it 4,000, they should buy for their brands at that moment, figure out what those are worth to them, and say no to all the others. The way that our business has changed is when we first got started, it was less than 1 million ad opportunities every single second. Now, it's, again, 15 million, and it used to be just focused on the browsing web. That's where all of the real-time ad opportunities started, were inside of a browser.
and now they happen across all digital media, and the most important and impressive segment of that is in connected television. So the core or the epicenter of biddable or programmatic ad buying is now in CTV instead of inside of a browser. So that shift has just made the world dramatically different. Of course, COVID accelerated a bunch of changes, and we've, as a result, been at the epicenter of the entire changing media landscape, in part because it turns out that buying ads in a data-driven way is better than any other way, including the steak dinners and martini lunches. So as delicious as those were.
Well, speaking about martini lunches, let's, let's talk about conversations-
Oh, I thought you were gonna bring out the martinis.
That's actually what's in there.
Oh, okay. Very good.
Okay, fantastic.
So,
Who's counting?
If we think about conversations with advertisers, one thing that stood out to me from not just your results, but 4Q results for advertising in general, is it was not monolithic. It felt like coming out of last year, things were really looking up in the ad markets, and I think we saw that certainly from what you talked about on your earnings call. But it has been a mixed backdrop when we look across what we've learned from companies, you know, when they gave their fourth quarter results. So what are you hearing from your conversations with advertisers? Are there areas of strength or weakness you would call out, and what are their priorities for 2024?
Yeah, so, the relationship between the CMO and the CFO has never been closer than it is right now, among the biggest brands in the world. And I think that's partly because of COVID and just the discipline that required in large companies. But in the same way that the markets have been a little jittery, maybe ever since COVID-
... where you're trying to see what, what does the future hold, CFOs are trying to determine that so that they can figure out where they invest, and when there's growth opportunities, they want to invest. And there are companies, interestingly, in the same categories, some that are going in, like, all in, in the same moment that others are pulling back a little bit.
Some that are becoming much more data-driven and pioneering new ways of doing things, while others are trying to retrench to safety. It's not that dissimilar from what I think happens in the equities market. I don't think they're that disconnected.
So the same uncertainty that went into the middle of Q4, that then was answered with, "No, things are better than we think," that we're seeing in equities markets today, the same thing is happening among the advertising departments of those same companies. So overall, things are better than they've been in a long time, and there's lots of positive sentiment, but there are individual companies, some of them are not doing well, and it's partly because of all of the change that has happened to all of them, where change and volatility is going to redefine winners and losers. That's sort of always the case.
... So maybe let's talk about, you talked about, you know, in your, in, in the first, first comments about how things have changed over the past couple of years. So where, where are the what are the one or two things over the past couple of years you're most proud of in terms of the progress you've made with the product and inroads with clients? And then when you think about that now, looking out over the next couple of years, what are the things that you think are the next product improvements, you know, inroads with advertisers you expect to make that you're really excited about?
Yeah, so, we've been public almost eight years. During the IPO process, I was pretty vocal that eventually Netflix has to show ads. Almost every single person I said that to said: "You're crazy, that'll never happen.
Including Reed, who told that to me directly. And seeing them evolve and change and being ahead of the CTV move is one of the things that I'm most proud of. We were also ahead of the identity shift, where initially we said: There's a way for us to pool the efforts of the open internet in cookies to make it so that the advantage of the walled garden goes away. We did that. We were successful. They changed tactics to make it so that they came after that sort of collaboration or union, and we responded with basically taking UID and turning it into UID 2.0.
which was built around email address and phone number, which we knew they couldn't disrupt, and that would be leveraging the strengths of especially CTV and audio to create a much better open internet. The fact that we pioneered that in 2017, I think, is when we launched that, long before all the cookies and tracking efforts of Apple and Google in particular, I think, has put us way ahead. We also invested in AI starting in 2018, so we've been in that market for quite a long time, long before it was the greatest fad I've seen since the dot-com exuberance.
So, if that weren't enough, also just last year, we made a massive pivot in our platform to move towards more data-driven audiences, leveraging UID 2, so that we could make personalization inside of CTV better and overhaul our entire platform again. Last year, on June sixth or 6/6, we launched Kokai, which we said would take about a year to roll out in its entirety, but we were just gonna keep giving gifts constantly in terms of upgrades to our platform. Those have been amazing in terms of just creating incremental performance for all of our advertisers. Some of the biggest of those changes will be coming in the first half of this year.
... as we continue to roll out over that year. And so getting ahead of the opportunity, so that as CTV pressures continue, that we're able to leverage all those. All of those investments sort of ahead of their time have been very good. And there's a bunch of them that we're making right now that haven't shown fruits yet-
... that I believe I'll be able to reference if you ask me that question three years from now. But, just those are cases where we were clearly ahead, and they've clearly paid off.
There's a bunch of stuff I want to revisit in there, but maybe sticking on the theme of CTV, you know, there's been a lot of focus year to date and through the end of last year on things that might be changing in the CTV market. And then there's, you know, it's a two-parter. Let me start with the first. Is there any impact that you would call out good, bad, or indifferent from Amazon launching its AVOD offering? Is there any volatility in CPMs or, you know, the volume of ad impressions being served that you would call out that are meaningful for your business one way or the other?
Yeah, in terms of market dynamics, it hasn't changed anything. It's created lots of questions, mostly from the investor community, to be honest. But I think it's mostly been positive in terms of the sentiment change that it's created.
What it really does require is that the biggest media companies in the world embrace programmatic and biddable, and get their collective shit together faster. In the same way that Netflix has always been a stalking horse to all of them, now Amazon serves as one to them, too. Whether that's real or imagined is unimportant.
It's making them go faster. So the fact that Amazon is such a small percentage, the fact that it doesn't really impact most of their businesses in any direct way, I don't think is that important.
I think what's mostly important is that everybody in media is moving towards biddable as fast as possible because they need to increase $20-$40 CPMs without increasing the ad load at all, because if you do that, you can lose subscribers, and none of them can afford to lose subscribers. They all have to race to profitability, which means that they have to make more money without adding any more ads. The only way to do that is to increase the CPMs on the ads that you're showing.
and so they're all moving in that direction, which means more biddable, and it means we get a chance to participate in those auctions, and we're pretty consistently showing them that we'll pay more than others will, so we get access to more and more of them. And that virtuous cycle has been amazing for us the last few years.
The second part is on fragmentation in CTV. So, you know, I think in the past, you've described the CTV, the streaming entertainment ecosystem, as being very well fragmented or maybe perfectly fragmented, I think is the phrase you might have used... and we are seeing a trend, you know, at least to some degree, of some rebundling that's going on, and you have some big players like Amazon that are making a push into AVOD. I guess, is fragmentation still The Trade Desk's friend going forward? Is the fragmentation gonna change? How do you think about that?
Yeah, so, the moves in fragmentation are not significant.
And really what you want to—what we wanna make certain never happens is that there's enough consolidation that any one of the sellers can be draconian.
What you have in search, and at moments you've had in social, is you have enough concentration that a player can be draconian. Like, Google, have 75% market share, so I can be as draconian as I want. And no one has the luxury of doing that in television. Irrespective of Netflix's market share, they still need to run an auction and have integrity in their advertising process. I think one thing that we can learn from the sort of Twitter to X transition is that the advertising ecosystems are really fragile, and there's lots of choice that advertisers have, and that there are very few companies that can boast of being a must-have on a media plan. You can buy somewhere else.
That dynamic is our friend because we then get to look at the 15 million ads per second, ad opportunities per second, and then select the 4,000 that are best for Nike or P&G or Coca-Cola or anybody else. And if that decreases from 15 million to 14 million, okay.
Yeah, fair enough. Maybe shifting to retail media. So, you know, another big opportunity for The Trade Desk, a big focus from investors. And maybe help everyone in the room understand what your product offering is for retail media, how you fit into that landscape, and how do you see that changing going forward, and are advertisers' level of focus, is it likely to go up from here?
Yeah. So our message to retailers has been pretty consistent the last couple of years, but it was a massive change in the way that they were doing business. So I wanna maybe explain the way that they were doing business a few years ago, 'cause it's very different than the story that we're telling them. But in order to understand what's unique about our pitch, you kinda have to understand where they were. If you're sitting in a major retailer, and you're sitting on top of all this data that comes out of the point-of-sale transactions and all the data just from doing business, you could monetize that in a way that is consumer-friendly, doesn't create any risk with your consumer, and you can make tens or hundreds of millions of dollars, maybe even a $1 billion-plus. But it's really complicated.
The data ecosystem is super complicated. So what would happen is the C-suite would usually assign one person and a team of 10 to go monetize that, and it makes a ton of sense to go monetize that if you're gonna make incremental tens of millions or hundreds of millions, especially when essentially the cost of goods sold is zero because it's incidental to the business you're already doing, so you can add incremental EBITDA in addition to just bringing in that revenue. But, what we've been saying to retailers for years now is: "You're doing it wrong.
There's a better way to think about this, and I fear that you're gonna lose to those that are doing it best, and especially Amazon." Like, Amazon thinks about it the right way, which is, "I want to leverage our data to spin our flywheel." Everything at Amazon has been to spin the flywheel, meaning, "We want to create incremental revenue to our business." So instead of going to, to use an example, to Procter & Gamble, the largest advertiser in the world, and saying, "I wanna make $10 million at this brick-and-mortar shop by selling you data," it's a hard-fought negotiation to make that $10 million. Wouldn't it be better... And let's just assume that's Walmart, just to keep the example simple, and to recognize that the largest players in the world are all interested in this space.
You could get an incremental $10-$20 million from Procter & Gamble at Walmart, but that's not exciting to either of you, especially given that P&G spends $6-$8 billion in advertising. Wouldn't it be better if all of that money were spent optimizing to selling product in Walmart? Wouldn't that be better than making $10 million? But what that requires is a buy-in across the whole company. It requires thinking holistically. It requires thinking about that flywheel.
and so what we've done is we've gone into all the retailers and say: "Think about it holistically, and think about how you can improve your own marketing efforts, but you also can make more money this way." so if you can make it so that P&G is optimizing to selling their products in your store, you will make more money in your store, but they're also willing to pay you for the data that will do that. But you've got to make measurement free, and then you make money on the targeting, and of course, you make money when you sell more of their product, whether it's soap or Doritos or Coca-Cola or whatever. You sell more of that product in the Walmart, that is the best outcome for everybody, and that's a better ecosystem, and it's more competitive to Amazon.
So because of the fact that we're representing the majority of the S&P 500 or Fortune 500 brands, we have a tremendous influence in helping them put that data to work, and of course, where the data is of most use is in figuring out what ads to actually buy. And that's the business we're in, deciding from the 15 million, which 4,000 Nike should buy.
So leveraging that data from a variety of sources, so that we can help Nike make the best decisions, figure out what ads are working, what aren't, be willing to pay for it when it makes sense, but then also to create more business inside of those brick-and-mortar companies, which creates a more robust global economy and a more competitive digital economy, is great for us, especially when we're essentially trying to make the open internet compete with a few walled gardens, like Facebook and Google in particular, but also Apple and Amazon to some extent. So if we're really trying to make the collective of everyone else do well, this is one of the very best things we can do. It's one of the most exciting things happening in all of advertising, but not just all of advertising, all of media, all of digital.
That trend is one of the hottest trends. You hear retail media networks, they all take different flavors, but to me, that's the holy grail of retail media buying and selling.
So you're talking about Walmart in there. I'm sure most people are aware that, you know, you signed a major partnership back in, I think, 2021 with them on the retail media side. And now, you know, they're looking into merging with Vizio. I guess, what impact do you see that having on their business, on the retail media landscape more broadly, and what does that represent to you?
Well, so, the very first thing that I thought when I heard about the deal, and I heard about it before it was announced publicly, was, "Oh, this is great news for us.
And the reason why is, Vizio has been, at times, fairly short-sighted in the way they think about monetizing their ads. They have a lot of ads. Because, OEMs have a dilemma, right? Which is that, they make almost nothing on the boxes, and in fact, margins have, in some cases, gone to zero or, and some predict below zero. So, they, as a result, have tried to create ad businesses, and you can create these silos that make it so that you can't buy holistically and manage reach and frequency and compare what you're buying in this silo to what's happening in this silo, and that makes buying a very small silo less appealing.
So, stepping outside of, like, selling in a very short-term way to selling more holistically and thinking more long term, that's what excited me about the deal. Plus, the fact that, because it's a very different business than the DSP business or any of the things that we do with Walmart, it reinforced that our partnership is going to do really well and perhaps have new opportunities inside of Walmart to buy additional inventory and whatnot. So, I'm overall excited about it, and what it represents for the ecosystem, but also what it represents for Walmart.
Another big topic for investors is cookies. It came up a lot, even in the Q&A, you know, with the sell side on the earnings call. So I guess, you know, I think you did a good job on that call, talking about your position with relation to cookie deprecation, but maybe give us a quick refresher for the audience here today about your strategy there and how you're thinking about the usage of 1P versus 3P data.
Yeah. So, a lot to unpack here, there. Because there is a lot of, of sort of false conclusions and honestly, ill-informed, discussion about cookies, so I'm, I'm grateful to have the forum to try to unpack it. First of all, cookies were never the best mechanism to create personalization on the internet. We all retrofitted them, because we needed a way to create personalization. It was a way to do it, but it's a really complicated, system that isn't really built well, especially for opt-out. So it's not the best way to manage the, the opt-out process, which is a critical part of the privacy process.
I believe that Google looked at this and said, "This is an opportunity for us to hide behind privacy," to do something that is really self-serving for them, which is to leverage their Single Sign-On, where they have, I think, nearly two billion users, and say, "We can use our Single Sign-On," which is essentially your Gmail or your Google login, "to provide personalization. We know that when we do that and when Apple does that, the two of us will have a personalization mechanism that others won't have at scale.
and if we get rid of cookies, there are all these companies that are using cookies to create personalization that will make it so that we don't have to compete with them in quite the same way." because, as I've mentioned, one of the things I'm most proud of having done in the last few years is really addressing the issues of identity. We were ahead of this. We knew that this would likely happen. We knew that Apple was being aggressive on this front early on, and so we created UID 2.
UID 2 is a currency of identity for the open internet, for everyone that's not Google and Apple, that makes it so that you can turn an email address into an ID without having to directly interact with the consumer in a new way. And what I mean by that is, instead of us going and knocking on 2 billion consumers' doors and saying, "Would you like to sign up for a new service?" and it's not to say that there aren't some reasons to do that, and we do have some products that do that, but that's not the core of the strategy.
The core of the strategy is go partner with all the companies that already have logins, whether that's every other website, whether that's gaming websites, and also all the companies that key their data off of that, which is really every consumer data company in the world. So there's not really a data entry in a database at an Experian or anywhere else for that matter, that doesn't also have, on the same row, an email address. And so as long as they get consent from the user somewhere along the way, and whether that's on a form at a McDonald's website so that they can put their first-party data to work, or whether that's in whatever devices they're using to collect their CRM data, we make it so that that can be created into a UID2.
So we started by partnering with companies like AWS and Snowflake, and the infrastructure of the internet, so that at the core of where all data transactions are done on the internet, we could create this new identity fabric. And by partnering with AWS and partnering with Snowflake... And by the way, they didn't do this because they were being benevolent or even necessarily forward-thinking in some cases. They did it because in order to compete with each other, they have to make it so that they can put the data to work. What good is it if I put all my data in Snowflake, if I can't put it to work?
And so because this ID is a way to take otherwise locked-up data and put it, again, after consent, to work in the advertising ecosystem, so I can figure out what to buy and sell, we made it possible for every advertiser, every data company, to put their data to work in a more efficient way that was also more privacy safe, so that the user can opt out, and it makes them so that they can finally have a bit more courage. So UID2 has become the primary currency of the open internet. We also said, one, we're not gonna make any money on it. It's really easy to get adoption when you give it away for free. And two, we also were gonna make it interoperable with any other currencies that we thought were privacy safe or sufficiently privacy safe.
And so it's interoperable with almost everything else, so if there's a paid service that some like, like LiveRamp, we're just interoperable with it. So, we don't... We're not trying to compete with LiveRamp, we're interoperable with them. So by taking that approach, we have created an alternative currency, that is powering almost all of CTV. We had an event, last week in New York City. Was it last week or the week before? This year has gone by so fast. But in the last two weeks, we had an event in New York City, and a VP from Disney was on stage explaining that, yields went up 3-4x using UID2, and it's just economically obvious to continue to use UID2.
We had basically people from every major media company on stage, and all of them said similar things. So it's become the primary currency of CTV. We expect the same thing to happen in audio, and it's already happening in other parts of the web. The long pole there is in creating authentication, where a lot of what's done on the web isn't on the other side of a login, unlike in CTV, where everything's on the other side of a login. You're logged in to experience that. But the rest of the internet is getting there and will get there faster simply because of the need to do so in order to keep CPMs high, which they all desperately need.
Right.
So it's been one of the most exciting things that I've ever been a part of because it shapes the open internet. And when you take CTV plus retail, plus the identity currency that we've created for the open internet, we've created a chance for the open internet to thrive in a way that walled gardens couldn't, and to create a real challenge to the walled gardens that never was possible before. And so the way this has shaped out is so good for us because we've made it so all the premium content is on our side. And so we have all of the premium content of television, all the premium content of audio, and that is competing with cat videos and hate speech, and I'm really happy with that division.
Yeah, fair enough. I mean, maybe staying with that theme of competing against the walled garden, I mean, what feedback are you getting and do you get from advertisers who use both products from Google, like DV360 and The Trade Desk? You know, what value are you providing that's unique, and what feedback are you getting about, like, the relative value of what you're doing versus big competitors like Google?
Yeah, I'm really happy at the sequence of questions that we're going after because it makes it easier to explain. Because from all the work we just did, and I recognize that it's fairly esoteric, and half of you in the room are like: "I don't fully understand the identity theme, but it seems like a big deal." It's a very big deal. Because what it does is it unlocks data, and the proof is in the number of data elements we're using to make decisions. And you can imagine that if you used to make decisions on, should I buy this ad, using two pieces of data, this person looked at shoes 10 minutes ago.
All of you've lived this, where one piece of data haunts you on the internet forever. Like, I wish I'd never looked at that hotel, 'cause I've already been on the trip, and for six months, they haunt you about where to go. That's because all the ad decisions were made with two pieces of data. Now every ad decision is made with ten pieces of data. And if you think of that in an AI world, where you're trying to teach models to make better decisions, you need more data to put to work. So as we get more and more data, we make better decisions.
So in the current landscape, the very best way to compete is to go to the biggest brands in the world and say, "If I could just have the information about the last 5,000 people who bought your product, I could look-alike model, I could do all of this, mathematical work to figure out where they are elsewhere in the ecosystem." Well, if you're Google and you do that, you have the risk of them feeling like you're competing with them or using that to gouge them on search, or to gouge them in some other way that you're selling media, especially if you've proven that the primary objective you have is to sell ads on Google Search and YouTube.
So the thing that we do, that is really hard to do in advertising, and again, the case study that is Twitter moving to X, I think also reinforces this: maintaining trust with advertisers is hard, and aligning your interest with advertisers is hard. And so we have gone to them from the minute we started our company and said, "We are never going to own media," because I think I'd get rid of my single biggest strategic advantage when I do that, which is my objectivity. Because I look at 15 million ads every single second and then say, "I'm going to objectively decide which 4,000 I'm going to buy for you." I'm not going to say, "Great news! The greatest 4,000 for you are all on YouTube. They're phenomenal.
That's why we bought all of those." I'm never going to say that to them. I'm gonna go to them and say, "I, I don't care if we buy Paramount or Disney+ or Spotify or Yahoo. We're going to objectively help you figure out, in a data-driven way, what to buy." And so because we have objectivity, we have trust, and because we have trust, we get access to their data so that we can make more informed decisions. That improves all of our modeling, that improves our media buying on their behalf, and the flywheel spins, and we continue to get better and better at the decisions we make, and we, as a result, add more value, and that's what has made us continue to perform as a public company, despite the fact that most in ad tech have disappointed all of you.
Maybe from a product perspective and, you know, building something differentiated, let's talk about the new forward markets. That's something you started talking about a year or two ago. It seems, from my perspective, to solve a very big problem, which is replicating some of the benefits of the way that traditional linear TV ads are bought and sold in a new CTV ecosystem. So how are those conversations going? How is the rollout of that product going? And maybe just give us an update on the new forward market.
Yeah. So, forward market is going to continue to be a slow rollout. It's gonna continue to take years to reach its maturity, but I do believe it has the potential to be massive. In the same way that a forward or futures markets in equities are a huge part of the marketplace, like, it just makes sense that as markets get sophisticated, and especially as this is, again, approaching a $1 trillion-dollar industry, that you would have sophisticated, forward contracts or forward products.
But there is this dilemma in advertising, which is that there's less fungibility in our world than there is in yours, meaning that a share of Microsoft is worth, is worth the same as another share of Microsoft, but an ad on MSN can be worth $100 to Nike 1 second, and then the next second be worth $0.05 because it's a different user at a different part of the page or a different part of the site. So that volatility, in some ways, is a trader's dream, but it also creates a much more difficult dilemma for price discovery, but also for forecasting. So what we've enabled in forward contracts is the ability for advertisers to lock up inventory and buy it in advance, and agree on a price with the publisher.
The thing that makes it hard, though, is right now, they all take risk, like a penny at a time, because when you're buying... even at a $40 CPM, when you divide that by 1,000, it's really small. And so, you don't have to take much price discovery risk a penny at a time, but a forward contract, if you buy $100 million of inventory, you could lose a lot of money. So the forecasting engines of knowing what's going to happen, it's not hard—it's not as hard to figure out what is the right value of this ad at this moment, right now. We used to say that's a hard job in ad tech. It is a hard job in ad tech, but you know what's even harder?
Figuring out how many of those there are going to be over the next two weeks.
Or figuring out when they're gonna call time out in the Super Bowl, or is the Super Bowl gonna be even close? Are people how many people are gonna be watching in the fourth quarter? Those dilemmas make it so creating the forward market depends on creating better forecasting, and the industry has to improve its forecasting. But as we've done in every part of our business, it's better to be early than late. We've essentially laid the pipes or the foundation for us to have a phenomenal business built on top of the spot market. We know we'll be a part of it. We know that we'll be the pioneer in terms of defining it and making certain that it is good for us and good for the ecosystem.
We've done that. We've done $ millions across it, but in order for it to scale to the $ billions that it should be, it just requires the ecosystem to grow up, and the ecosystem has a lot of work to do as it transitions, especially from linear to digital. So it's gonna take some time, but again, we're ahead.
I wanna close by talking about Kokai. I mean, you alluded at the very beginning of the conversation to how, you know, it is something you announced in June last year. It's this, you know, ongoing process of product improvements and rollouts and some important stuff happening this year. So maybe talk to us a little bit about what is changing, why it matters, and how in particular it interacts with generative AI, which I know is a big topic for the audience, and certainly for the Kokai initiative as well.
Yeah. So let me start with the last part of that, 'cause it dovetails into my last answer as well. I think there is so much hype around AI, and there's so much that has been talked about in LLMs, that I think that some of the more obvious use cases have been passed over and aren't really talked about as much. Parsing through language to find truth is hard. It is much harder than parsing through data, which by its very nature is more binary. And so there is this opportunity for us to look through massive data sets. When you're looking at 15 million ad opportunities every single second, there is a ton of data about what's available. So we pointed our AI efforts about forecasting.
If you make a change, if you change your frequency from, I wanna show every user 2 times per hour the same ad, if I change that to 3 times per hour or 3 times per 2 hour, and then if they see that, show them a different ad, what happens to my ability to spend, and what happens to my performance? It is much better to point our AI at predicting what's going to happen there so that we can essentially optimize before spending a dollar.
So that instead of making some changes and then come back in three days and see what happened, which has been the cadence of trading in our world, and we do call it trading, and most of the traders in our company, half of which have come from equities and commodities, they are trading on that sort of cadence until we introduce AI-driven forecasting in Kokai.
So Kokai is really about taking those rich, we call them seeds, of audience that we can then lookalike model, and then predict against this massive data stream what's going to happen, and then injecting better prediction. And then, as a result of being able to change that cadence, you'd really have to re-architect the entire workflows. So we have reinvented our entire platform to respond to all these new capabilities, the way you navigate it, what you click on, when you click on it, it's all different. And so we re-engineered our entire platform. We started on this, I think it's fair to say, before AI became the craze that it is right now. But as a result, we're seeing just massive benefits, and we'll continue to see that roll out as we see benefits over the years to come.
Awesome. Jeff, thank you so much for being here.
Thank you.