Welcome to the March podcast. This is Arie Paparro. We're coming off an amazing week. We had an amazing time here at March Live on Tuesday and Wednesday. We had a thousand people. Incredible vibes, great weather. Probably a lot of the people listening to this were there. It was great if I had a chance to see you. If I didn't, maybe next time.
We are going to be broadcasting a bunch of the sessions from that on this channel over the next couple of weeks, but the most anticipated session by far was the closing session where I had a heart conversation with Jeff Green, the CEO of the Trade Desk. Jeff has been in the news quite a bit. He's been posting on LinkedIn. He's been buying his own stock and otherwise kind of shaking things up. And it was a packed house. Everyone wanted to hear what he had to say. And we had a good conversation. We even broke some news about his AI plans, about what the Trade Desk is offering its clients in AI. We also got some details on Ventura, their CTV offering and Open Path and what's going on there. So, I hope you enjoy it. It was a great conversation. Eric and I will be back next Friday with our usual conversations and we'll, like I said, we'll be probably publishing a bunch of the March Live conversations on this podcast channel, mostly on Mondays moving forward for the next couple of weeks. So, I hope you enjoy that. Thank you very much for attending if you were there. If not, I hope you enjoy this conversation.
You've been in the mix in the last week or so. We had our prep call like a week ago and you said, 'Oh, a lot more is coming.' And I was like, 'Oh my god, I can't take it.' So, what's going on?
Well, yeah, a lot is happening. I mean obviously probably the biggest news for me was making the biggest purchase of my life in an insider stock buy that I learned only afterwards was the third largest ever. Just a commentary on my convictions in our future, right?
Which then kicked off a whole bunch of other discussion.
Do you just like put it into Schwab like market order press go? So, I won't tell you which brokerage, but this one I did. Yes.
I mean I'm working with their trading desk.
Yeah. I imagine it is at a firm like that where I kept it simple.
So I want to know your note on LinkedIn about the purchase had a couple of nuggets in it I wanted to kind of ask you about.
We're off script already, too, but...
No, no, no. We're not. I mean you wrote this so I have to ask you about things you wrote. So first thing you said was soon you will have access to AI chat inventory. So I assume that's reference to ChatGPT. Is that deal done?
So I of course can't commentate on any deal that we have not publicly announced yet but I'll just say in general all of the AI companies have a serious problem in some ways which is that their capex is massive, their valuations are massive and they have really high expectations on the way that they're going to make money and almost all of them are either exploring or should be exploring what's happening inside of the ad ecosystem. And I think wisely many of them are looking at it and saying we're not just going to copy and paste what Google did in AdWords before. There's a lot more to think about. There's obviously even a lot more value that you can add. I think very very few have really contemplated the fact that this isn't really conducive to keyword-based marketing. If you look back to when the AdWords model was built, right, the average query was less than two words, refinance mortgage was what you would type in. And around the time that business model was born, what we're putting in prompts is less keyword anchored, which creates an opportunity for something with more value as well as the answer is more valued. I took the time to write six sentences and now you're spending a whole bunch of compute. If I'm going to watch maybe I should watch a video, right?
From my perspective, sort of putting on my product hat and a former DSP hat, it seems pretty hard to translate, you know, even if it's 20 words of great intent into the kinds of campaign targeting available in a system like yours. So how far along has that thinking gone?
I would say it's still, I mean we have tons of ideas that I think are way beyond anything that we do right now in contextual or things like that. And of course AI itself can help with that, right? Like the AI movement of the last few years has really been led by LLM. So of course there's an opportunity for that to play a role in the actual targeting. But it's still very nascent because there are business models that we don't want to harden until we can believe that they can be huge.
Yeah. The second thing you said in that post was that you were going to have access to retail media inventory and I forget the way you phrased it, but it really sounded like you were talking about product listing ads of some kind, which I don't also believe you currently do at scale.
That's right. We don't do it at scale. We've done some basic testing, but the premise is essentially this and this is the way that our decisions have been guided from inception. We tend to start with a macroeconomic view. And you've heard us talk a lot in the last few years about orienting around premium. And sometimes I make the analogy to clothing where there are enough clothes in the world for every single person to have a wardrobe. It is not hard to get clothing to every person in the world. It's a tragedy that that isn't better done. But that doesn't change the fact that on Fifth Avenue you can still buy a t-shirt for $500. So there is a world of premium fashion. There is a massive stratification as it relates to gradients of fashion and the exact same thing as it relates to content or even words on the internet. There is premium content that is going for massive premiums, unprecedented premiums in live sports. I mean, the price of Paramount was so different than the market cap of the company 5 years ago. There are just so many examples of how the premium multiple on amazing content is going higher. And the exact same thing is true of ads. Most ad placements are not worth much, but there are many that are worth a tremendous amount. And that's where we've always been gravitating towards. We've been going to the places where there is premium. And one of the most amazing and one of the things that's great about premium, by the way, is that the best way to monetize it is with an auction because there's high demand for those things. That's the reason why they're created that way. So sponsored listings are some of the more effective places for ads to be placed. However, many of the retailers just want more demand in order to move prices up. So we want more efficacy. They want more demand. And there's some great conversations.
Yeah. I mean retailers are very clean environments for the most part to show ads. One of the challenges has been where with non-endemic that they don't retailers don't love the idea of ads that click out where you're not converting. Is there anything in your product strategy you can kind of give us a hint about as it relates there?
Nothing that I can say at this point. Those retailers that obviously want to sell their own products before they sell somebody else's products have some policies that each of them need to make on their own. And those obviously their choice, but there's lots of data to show there can be reasons to do it.
So the third thing you said in that post and I don't want to like rat hole on this one forever is that the Amazon DSP might not exist in 5 years or I don't know if you said it won't exist, might not exist, whatever you said. Can you give us like your top level thinking on that?
Yeah. So, I think this year you'll get another case study from Google on just all the difficulties of antitrust have really come from all their participation in the open internet. Like I think you're going to hear from them a lot of like we wish we were never in this, like they don't, all of their troubles come from the open internet and none of the money does for them. And I believe there's more at risk for Amazon because of the fact that it touches retail, because it touches AWS and cloud to put those things at risk I think would be a strategic mistake for Amazon. So I believe that their role in the open internet or their stay in the open internet will be shorter than Google's because of the macro impacts on their business.
That's interesting. I mean, the open internet is dangerous for these big companies to play in because there's a lot of people involved, a lot of politics. It wasn't in our notes, but I wonder, do you have any thoughts on what's going to happen with the Google trial? We're waiting any day now for a final judgment.
Honestly, while I do hope that there's some teeth to the ruling, I'm not sure that it matters that much. I think the pain has been felt enough that they're never going to try it again. And the case study for not doing it is very present. And that's part of the reason why I make that assertion about Amazon. And I do want to be clear, I'm not saying that Amazon isn't doing well in ads. Like if there's anything that I've been misquoted on on Amazon, it is this. Amazon has a great ads business. They have an amazing future in ads. Those will always be built around sponsored listings and Prime Video and that's where almost all the money comes from today. That future I believe is very bright for them. But everybody who has owned and operated inventory where you make almost 100% margin on and in sponsored listings that's the way they look at it in Prime Video. It's a much more complicated equation. But whenever you're comparing that to instead of me making the majority of a dollar, I'll just give 80 cents to my biggest competitor fighting over sports rights that is Disney. I would rather not. I think it's when you play that out that they rethink their strategy because there isn't enough upside in their DSP as a standalone versus all the other avenues they can pursue, especially Prime Video and sponsored listing.
Yeah. Well, the danger is that if you make the analogy to Google and especially to the DV360 DSP that it still exists and it's a major player but they're sort of siphoning the money over to YouTube. So, you pointed this out. I wish that the press would have covered this as well as you did. I really do.
I consider myself the press kind of.
I guess the more mainstream trade press didn't cover it nearly as much as you did. And actually the tweet that you put out was so good. It was the simplest way to show like what had happened from 2018 onward YouTube versus their open web.
It's striking. The graphs that they showed into evidence were striking. I want to defend Ad Exchanger a little bit because I wrote a book about Google and like 90% of the footnotes in that book were Ad Exchanger. So you know this first draft of history thing was very much the case there. So I want to give them some props up on stage here.
So, let's change subjects. Let's talk about transparency versus outcomes. And I know you're going to object to that framing. And I spent a lot of time in the last market on this. It feels to a lot of people that transparency is a little bit of for show that, you know, people like to talk about transparency, but really it's about, you know, ROAS or iROAS and it doesn't matter how you get there. What's your reaction to that?
Yeah, I disagree with the framing. Partly because I just think there's a more interesting conversation to be had around closed versus open systems and you know I've been one that's talked a lot about the open internet versus walled gardens but I think we're elevating the conversation to more about open systems versus closed systems. And in the open closed systems, and this actually has been a lot of the conversations that I think are happening inside of many of the biggest AI companies in the world, you have to power an ecosystem in order to be successful. They need to power hundreds thousands of companies that are innovating in order to really do well. I would argue that the greatest asset at Apple is their app marketplace where they powered innovation of other companies. So, I think the danger in some of the closed systems and what we've historically sometimes referred to as walled gardens is that the transparency around their measurement or the way they count outcomes is so self-serving. Like I'm often saying that I believe the greatest accomplishment in some of the walled gardens is not connecting billions of people to each other. It's not in the millions of hours that are uploaded every second. It's actually that they've taken credit for all the brand building that other companies have been doing and they've managed to take credit for all of that performance with their products. That is at the core of the problem with the closed advertising ecosystems. So you're right, outcomes are outcomes. But if you say this one platform sold all of my products, like I was talking to a CPG recently where you just look at the growth of those companies versus the growth of theirs. Like for 50 years, these companies were growing at 10% a year and then all of a sudden that stopped with the advent of digital where now the companies especially those with closed ecosystems are growing rapidly and they are not. There's no doubt it's extractive to some extent. It works though to some extent. So you're kind of balancing the two things.
The open conversation's interesting, open, transparent. All these words kind of mean different things, but like you basically if you're in danger, I think not you, but you thank God in danger of like creating like Linux for the desktop, right? Where it's like total control, total openness, and yet you know, it's hard to use and you know it's more effort. And on the other hand, you have a beautiful new Mac that you can just buy and it all works.
I think that's right. I think that's actually fair. I believe even at times inside of Trade Desk, we've been dogmatic about transparency. I sometimes that can translate into one time at Band Camp and then you just spew. And we at times have been way too open to the point that it's not practical transparency.
Right. What does that mean? What's not practical?
Well, so like I'll use another analogy sometimes I think as it relates to the way that we invoice. Like you get an invoice from...
I had this problem with Beeswax too.
From Facebook you get like a one-line invoice. We give you like 30 pages. And so it's the equivalent of going to a restaurant at the end of the meal getting a list of every ingredient you just consumed. And then we debate the price of croutons or chili powder. And that I just don't think is an effective way to be transparent. What you're trying to create is understanding, right? And we're constantly looking through the rubric now of how do we create understanding? How do you make this accessible? That's practical transparency.
I highly recommend not being transparent on your invoices for my experience. Also, there's just no benefit to that. You know, follow up. You could get a transparent invoice if you ask for it, but not on the first go.
So transparency kind of folds into the next topic, open path, supply, etc. We had Michael Sullivan up here earlier, and he covered it, so I don't want to cover it too much. But I do think it's kind of interesting. I'm kind of I'll take your side of this conversation, which is, you know, Yahoo has its own supply, Microsoft has its own supply. You know, Google has its own supply. You know, 70% of DV360 bids on supply controlled by Google. Why do you get such a hassle on this? Why is everyone up your ass on Open Path?
Yeah. So, I think the primary reason is we have said we're going to create a stalking horse and a canary in the coal mine and that we're trying to create the most efficient supply path for the open internet. I think there are a significant percentage of companies in our space that are instead of trying to create efficient supply paths are trying to exploit inefficient supply paths. And so of course you wouldn't want a product like that, right? It is a disruptive product because it's trying to make the supply chain better and the people that are monetizing an ineffective supply chain hate it.
