We're going to get started. Wow. Great. Two seats here, one seat there, one seat here, one seat there. Anyone wants to sit? Feel free. Hi, Tom. Ok, let's start with a personal question, Michael, a hard personal question. What hidden skill that does not go on a resume do you have that has been mission-critical to your success professionally?
That would assume I have, like.
Any skills?
Skills.
Yeah. I don't think you have any, but they must be hidden, so you must have a long list of hidden skills.
They're all hidden. You know, thank you for sharing it in advance and not popping it on me in real time.
I did it to Tim Armstrong on stage. He's like, what are you asking me?
Yeah, I would like to think that one of my hidden skills is my EQ.
OK.
I think that if you look at the roles that I've played as a Public Company CEO, almost all of them have been taking over the role of a Founder, and so it's a little different when you come in from outside and you have a founder-led culture and you have to invariably do some tweaks to the culture because I don't always get the call when things are up and to the right, usually they change founders when things aren't going so well, so when you come into that sort of an environment, usually you need a culture change, if you buy companies, there's a whole culture mix that has to happen, and I think it's really tough to do that if you don't have a high EQ to kind of work through the subtleties of making it happen.
Great answer. However, what I want you to do then, which you have not been prepared for, is on this stage, either a year or two years ago, I asked you about your leadership style. And one of the things you said that no one else said is, as one of many brothers, you understood that the buck has to stop somewhere. And you decided in your upbringing that the buck was going to stop with you, that whatever mischief you and your siblings got into, you were the guy that ultimately was going to make the decision. And that's how you run your companies, that people can give you advice, but at the end of the day, you're the leader. It's up to you. And you take the heat if it all goes horribly wrong. So to me, that's like the almost opposite of EQ. An EQ guy, you would feel is a communal guy and get everybody on the same page. So can you consistently?
Yeah. I mean, I think, look at when you do try to build collaboration, the flip side to that is indecisiveness, right?
True.
So the superpower might be collaboration. The area of growth that you would need is, and you can get accused of that. There's no question that in times where people are like, damn, can he just pull the trigger? But I find that if you do it too soon, you ultimately have to pull the trigger. I just find that you need to build a consensus to do it, particularly in the situations that I outlined in terms of.
You need to make sure everybody's in the same timing pace. Even if you have to take an extra week to get them there, you just have to wait for them to get there. So, everybody's. Yesterday, I was on stage with a couple of Amazon people, and they're like, "debate and commit," and the idea is we debate ruthlessly, but when we come to a decision, everybody commits, even if it wasn't their point of view.
And even if it's wrong.
Yeah, yeah, yeah. Everybody commits because you made a decision. OK. Yeah, that seems great. OK, let's talk about CES. You and I saw each other at CES. I loved CES.
Great activation, didn't we have, right?
Great. Oh, my God. OK, I just want to tell you guys what he's talking about here. Total inside joke. So I go to meet him, and usually everybody's like in suite at the Wynn. No, I have to leave and go to a new hotel across the parking lot, basically, in the Cosmo. And I'm looking for this. Basically, what he means by activation is a room, right? And you go to the janitor's closet, and I'm like, where are you taking me? There's mops and there's brooms, and it says janitor on it. Michael's in the janitor's closet for his meetings? Like, couldn't they afford better? And you open it, and it's a speakeasy. You open it, and it's this big, beautiful room, twice the size of this room, maybe three times, huge bar, paneled room, low lighting, and it's all Magnite people. There are Magnite salesmen at every table, and Michael's got the back corner.
Literally, a barbershop that you walk through to get.
It is a barbershop.
The barbershop is functional, and your clients can get a haircut.
Yeah.
And because these guys are.
That's true. They're cutting hair.
We were getting calls in advance saying, Can I book a call?
Oh, a haircut at 2:30 . And you're walking by these men getting their haircut because it is a barbershop. There's no chicks. You're walking by, and then you go to the janitor's closet. You go in the back, and you're going through mops and brooms, and you're walking. It's like a hallway, but the door says janitor's closet, and you're like, where am I going?
It saves by doing that extra touch. It saves us from having to run up and down elevators because people want to come to us.
Totally true.
As opposed to us having to chase them.
You're going to be my first meeting next year. If you're doing barbershop, I'm coming. I'm coming to the janitor's closet.
Dan's haircut.
Dan's getting his haircut. Anyway, let's talk about CES. I sent you my list of provocative statements, some of which were yours on that, my takeaways from CES. Pick one and tell me what you agree or disagree with.
Yeah, I mean, the one I disagree with and was the one that you said that individually said it was an Open Internet player was that the Open Internet is dead in the walled garden.
He just said it again on stage.
Yeah. I don't know where that data point is coming from because if anything would point to it is that there is more willingness than ever from a marketer to have less of their budget going to the traditional walled gardens and more going towards streaming. And streaming is certainly not a walled garden play. And so I think that that was, I found, provocative and disagreed with.
OK. Making it their provocative, right? I think what he meant, let me explain what he means and then see if you agree with this. I think his main point is that search, specifically Google search, is moving away from a bunch of links where you go next to the website, which has a bunch of ad units, and Google search is going more towards we're going to answer your question, so it lowers the volume of website traffic, so what he's saying is website traffic will now be on a structural decline as consumers increasingly use either Google search or OpenAI to answer questions that do not require links, and therefore, ad inventory availability will become less on websites, which is the Open Internet, over the next three years. Agree or disagree?
Definitely, search referral traffic will decline. There's no question about it. However, search referral traffic has been in decline for years now, just simply the way that Google's been doing this. You used to search for a flight to Chicago, and you'd get the Delta website. Now you get the results in a Google search. So the web referral traffic, search referral traffic, has been in decline. So I don't think this spells the end of traffic to websites. But as I pointed out before, too, streaming is considered the open web as well.
I see.
Streaming is not dependent upon search referral. I think that we're at the heyday of Open Internet, if you will, if you look at streaming as a component.
But an important point is a lot of people think that people that are not in the ad tech business, they think it's a walled garden. They think Disney's a walled garden. They don't define that as Open Internet. So yes, he is thinking you have 60% of your revenue comes from DV+, which is websites and the Open Internet, and 40% from connected television. You and I, because of the value leakage from CTV into the Open Internet, we call that part of the Open Internet. A lot of people who don't understand ad tech do not. They think those are walled gardens. So it's not smart. It's just definitions.
Humility is one of my hidden skills, too.
Self-confidence is one of your hidden skills, critical to your success. I think it goes on a resume, though, so I don't think it's responsive. OK, so let's talk about CTV, so connected television, smart TVs versus DV+. Let's say round numbers, you're 50/50. Because there's five CTV companies that have 75% of premium CTV ad units, has this moved Magnite to becoming more reactive and a price taker than when you were aggregating sort of powerless websites? And now you really just have to kowtow to people like Disney and Warner that are always used to getting their way. Please comment.
Yeah, they are used to getting their way, aren't they?
Yes, and they are difficult to work with.
I would say two things about that. Number one, we always knew take rates were going to be different in the CTV world, just given the value of the inventory. Again, you know, 3% off of a $60 CPM versus 15% off of a dollar CPM, you start to do the math on that one, and you're going to live with a lower take rate, but your net revenue situation is going to be better than the DV+ business. That was always going to be the case. That had zero to do with whether I can be a price taker or a price giver. It was just a reality. That brings a lower take rate to CTV. We have a product mix. It's one size fits all in DV+.
You take our demand facilitation, you take our exchange, and you pay us 14%. In CTV, you might just take our ad server. Ad serving is never going to be 14%, right? You might take our ad server and our streaming product, but you want to sell the programmatic ads. You don't want us to sell the programmatic ads. That takes another slice. Or you want to take all of it, and we bring in the demand, and that comes with a double-digit take rate. So I think that it's more of a product mix and the value of the inventory that has a difference in take rate with DV+, not necessarily our inability to prove our value to justify a take rate that's high.
Okay, like the answer a lot. One of the things that we're fighting about, I did a panel, standing room only, of future streaming. And one of the big debates we got into was sort of violent disagreement, actually, was true or false?
False.
Maybe. True or false, this house believes that CTV ad inventory is unlimited?
Unlimited. I would say we have an imbalance between demand and supply right now, but unlimited is absolutely not accurate. Unlimited is DV+. That's unlimited.
But aren't FAST channels only 40% selling out, which feels like CTV FAST channels are free ad-supported television?
Unlimited, to me, means the ability to create ad units on the fly. So if there's a, unfortunately, I've sat and you lived it or living it personally, L.A., right? When an event like that occurs, the traffic shoots up across news sites, across weather sites, across everything. Every page that loads, you just fire off five more ads. You're creating ads on the fly in real time. A fast channel is, if you're watching a program, it's 30 minutes. It's only four minutes of ad time. If more people watch it, you're surfacing more ads because you're not kind of creating them dynamically on the fly.
OK, I see what you're saying.
You have set slots, so it's finite in that respect, which will always garner more value.
OK, that's fair. Nobody mentioned that. I think with a 40% sellout and 300 FAST channels, it feels like ad inventory right now is unlimited, but you're making the point you just can't make them up over time. They will sell out.
An imbalance, say. There wasn't two years ago, and in two years from now, there won't be.
OK. OK, that's fair. Like, it's a short-term imbalance. It feels like unlimited, but they can't make them as more people come. OK, totally fair. Live sports. Let's talk about the Disney deal expansion to add live sports. How much revenue do you think live sports could add in the next year or two?
$20 million.
OK, next year? That's awesome. I'm publishing that tomorrow.
No, I think live sports is a journey, right? We talk about the reason why we emphasize live sports is because it's becoming so popular in streaming as a content category, right? You're seeing all the rights deal to streaming. I mean, watch two NFL games on Netflix this year.
Poorly done, by the way, but they'll learn.
They got better.
They got better. They're gaining market share.
Remember how bad Fox was?
Yes, I do.
They were horrible.
They learned.
Yeah, you have to.
They're fast failers.
Yeah, there you go. So I think that sports is going to be a journey, and too often we think of sports as being the NFL. We will never programmatically serve an ad in my lifetime, which is.
To the NFL.
To the NFL. The Super Bowl-going programmatic is not going to occur anytime soon. But can an NBA game? There's 280-some-odd NFL games.
Yeah, women's basketball.
There's 280 sports events on any given weekend, right? And all of those are within our addressable market. And so we are working closely with all the media rights holders of the Big 4 sporting leagues, but also much more motocross, like you name it. There's so much inventory that's streaming. You wouldn't ever start a league or air your media rights and try to go cable company, cable company, cable company to try to get it on cable. You just create a mobile app. You create a streaming app. And so all of this is getting monetized. And the reason why.
Wait, unmonetized. Is that what you meant to say?
Yeah, essentially, you'll never see an ad on some of these services because they're not big enough to deploy a direct sales team. And the programmatic ecosystem isn't wired well enough because it's been built on display open web, the DSPs in particular. It isn't structurally sound enough to distribute ads in a live environment. And that's what we're changing at Magnite. And it'll be, first half of this year, a build story, building a sports product. And the second half of the year will be activating.
You're doing live sports for broadcast, aren't you?
Correct.
Today.
Correct.
So I don't understand this notion of we're building in 2025. What are we building if we're already doing live sports for local TV?
Because it is so limited right now from a programmatic standpoint. I'll give you a case in point. If you're a demand-side platform and you have a budget from Allstate and they give you $1 million, but they want it to run over a 30-day period of time, so you use throttle.
Your DSP, your Trade Desk , throttle.
When you see inventory spikes that occur on days that normally don't have inventory spikes, like a Saturday afternoon, you immediately think it's fraud or bot traffic, so you really tamp down your spend. Meanwhile, that's a college football game. They can't recognize it, and they throttle the spend even if they did recognize it. What we have to do is educate the DSP ecosystem, educate the buy side that this is a live event, this is a sporting event, and if you connect through the Magnite plumbing, it will all get distributed evenly. The frequency caps will be there. The competitive separation in the pipes will be there. We'll cache it ahead of time so that if it goes into double overtime, we can serve it instantly. All this piping, this infrastructure, as simple as it sounds, hasn't really been built. That's what we're doing, and that's how we're going to change it.
So here's a question I don't understand. There is data that comes across the bid stream. Why don't you just say sports or NFL or NBA? Why isn't it just as easy as putting three initials at the front of the bid stream so that The Trade Desk says, if it says NFL or NBA, bid for that?
Because even if you did that, their algorithms tamp down how much spend they would do. So if Allstate.
Even if it says NFL at the front.
They'd have to rewire their whole algorithm to do that.
It seems like the algorithm could just read the metadata and have code that says, you know, buy this because it's NFL or NBA.
But if you've been buying 15 years on the Open Internet banners and that's how you built and that's how it's.
But why is it on you? Why doesn't it on him? Why isn't it on the trade desk to do this? Why is it on you to do this?
I think the media owners have tried for that to occur. It's not occurring fast enough, and so we're going to help them make it occur.
Well, I hope you're charging for this because this seems like it's a Trade Desk problem and you're paying for it to fix it.
Yeah, at a 40% take rate.
OK, never mind. I take it back. I forget that. Let's talk about Netflix as a growth driver. One of the reasons that you're showing up on a lot of charts as a stock to buy is that I think all of us analysts are raising our estimates for 2025 because you have a bunch of Netflix ad inventory and services coming, and you've said publicly, I think on the third quarter call, maybe even on the second quarter call, that Netflix will go from being like zero a year ago to your largest client by the end of the year.
With an asterisk, our largest CTV client, of which the existing largest client doesn't represent more than 10% o f CTV, right? [crosstalk]
But do you have a bigger DV client than you do CTV client?
DV is a bigger business as you put it.
That's true, 60%.
60% of the business is DV. So when you start to extrapolate that, don't use the whole [crosstalk]
I see. OK, you're right, and I was saying the biggest client, not the biggest CTV client. Fair enough, so it's going to be the biggest CTV client. Let's talk about the timing of the Netflix fee ramp in 2025 and 2026. At maturity, not near term, but at maturity, how big could you know, when do you think that is, and how big could Netflix be?
I think we've said pretty consistently exiting 2025. We believe it could be our largest CTV client. As it relates to how you pace that out over a quarter, so much is dependent upon Netflix's launch plans. We just go along for the ride, right? Everything's built. If they want to start running in Japan tomorrow, we're ready to go. It's up to them to decide when they want to start running in Japan. So much of this is out of our control as it relates to the timing. They've been very vocal about their interest in expanding into 13 international markets, including the U.S. I think that number's kind of the tip of the iceberg. Again, we're all set. Everything's wired. Systems are in place. DSP pipes are connected. It's just a question of when they want to flip the switch and launch.
OK. So it seems sort of risky business to make a projection that you control none of. I mean, you could say over time.
I think David Day made that projection. I want to be very clear.
David Day?
Nick is the IR guy keeps talking about it, so.
The funniest thing, we're sitting in this barbershop, and I ask him a question, and he doesn't answer. So I ask it again. He goes, "Stop it. I'm going to tell Nick you're bullying me." I just thought that was the funniest line. Generative AI. So a lot of my ad tech companies, because you guys do think of yourselves as technology first, are using generative AI. So tell us how you're using generative AI today to either lower costs or increase revenue growth rates and how you look at the roadmap for generative AI in 2025.
To be clear, we've used machine learning, which is the predecessor of generative AI, for years.
You guys want to sit, Walter? You want to sit? There's a seat here. Yeah, right here. Come on up. Yeah, same thing. He's always standing around. He's always loitering.
Just disrupt my whole spiel and sit down. I'm glad you're comfortable.
Generative AI. He doesn't like to talk about this anyway, so gave him time to think. Generative AI.
To give you some, the way our system works is ad requests.
Yeah.
They're our lifeblood. It's our oxygen. On any given day, we'll do a trillion and a half ad requests. If I took a trillion and a half ad requests to auction, we'd be out of business. It's just too expensive to do it, and too few are bought. And so we somehow, through machine learning, figured out through patterns of who buys what, how to take a trillion and a half and bring it down to $400 billion. And that's what we bring to auction. So that's an incredible cost savings for us. And then we help educate our DSP partners on how to only take the good stuff too and help their CPU costs go down as well. So it's called traffic shaping. Some take it like Google and The Trade Desk say, no, thank you. Don't ever throttle us.
Some of the younger DSPs will take it all day long and say, great, you're just sending me the stuff I like. So in and out, we do machine learning. How generative AI is working right now is more of an internal tool. It's obviously helping our coders. It's helping our salespeople. And it's helping our marketing folks. Although I'm getting kind of sick and tired of those stock-generated images of like a Magnite logo floating. And everything looks like Roku City.
In purple?
Yeah, in purple.
Purple.
Exactly, and where we're going to go with that is you're starting to see some outward facing for our clients' use of AI in this respect to our Demand Manager product. So when you implement header bidding, it's not as simple as just saying, I'm going to shoot one impression to 10 SSPs.
Yeah.
You're shooting 213 different configurations of that one impression because one's on a mobile device, one's on an iPad, one's on a mobile website. Some are in foreign countries where you have to have a different rendering. So you're doing a lot of work.
This is you.
No, the poor media owner.
The media owner.
Our Demand Manager product makes that easy. But now with AI, it makes it even easier and quicker to say every 15 minutes, that's not to create a mix of that package. We should do 230 items and mix it this way and send it to those people. So it's really helping the efficiency, and we cited a case study where one publisher made 17%-30% more in revenue just by having more intelligent ways to reconfigure the ad requests that they send to SSPs.
So this is a tool you're giving your publisher clients like Disney, Hulu.
Exactly. It makes your Demand Manager, and that makes Demand Manager more attractive, and that's just the tip of the iceberg, and then more and more.
This is in connected television?
No, this is in the.
DV+.
DV+.
DV+ business.
But you can see that working.
Yeah, yeah. Well, that's the bigger business.
I mean, ultimately, we sit on our goal is data, right?
Yes.
We have tens of thousands of publishers, and every day we store all that data about what happened that day, and it's of very little use to publishers because it's so voluminous. You imagine taking that data, structuring it, and dropping it on an LLM that we'll just rent. We're not going to create one, and then becoming much more sophisticated in terms of price flooring. You know, price flooring is something that people set and forget, essentially.
I think people need to know what's price flooring. You need to tell them what a price flooring is.
You know, somehow or another, a certain publisher may say, I'm never going to sell my inventory for less than $2. Meanwhile, the market's all at $1.50. And so you're just essentially trying to keep an artificial price floor in place when you should be selling to that person for $1.50, that person for $2.10, that person for $1.70. And when you add it all up, you get more than $2.
Right.
But if you just set it at $2, you're just missing all these opportunities. So we have that data to make publishers that intelligent. And so that's what we're working on.
That'll be a new product that you offer or you build it in.
Or it just becomes the product, right? In other words, all the tools that they use today become just much easier to use.
Okay. One of the things that became clear or that people said at CES was that there's real downward pressure on CPM, so pricing for connected televisions, that FAST channels are now selling for $6, down from $8. And that's hurting the Roku numbers, which are down to $18 from $22. And premium inventory has gone from like $30 to $20. And they're blaming it on Amazon Prime, dumped a bunch of, like you said, imbalance, dumped a bunch of connected television ad inventory. And Netflix is putting ad inventory into the marketplace too. Do you have a point of view on cost per thousand and pricing? You said there's an imbalance. Is this a perpetual imbalance? Is it a short-term imbalance? How do we get back in balance where demand exceeds supply in the connected television industry so we can have pricing power? Because that hurts you. I mean, you get a % of ad revenue.
Yeah. So I don't buy the Amazon Prime being the cause of it. I think if anything, people have been a little surprised about how few people watch Amazon Prime versus subscribe to Prime for two-day free shipping.
OK, that's true.
The buyers really played their hand beautifully during the upfronts. They used that in Netflix as leverage to correct pricing across to Disney's, the Paramount's, the Warner Bros. Discovery's. The cavalry never arrived in terms of that being the reason for plummeting of CPMs. It was just a recalibration. They should have never been charging $60 for untargeted ads anyway. They did it because it was scarce and they were the new shiny object.
This is Netflix he's talking about.
Yeah, or anybody in.
I don't think anybody paid that except Microsoft for a trip.
You know, when Peacock launched, they were trying to get $70 bucks. When Disney+ launched, they were trying to get $60, $70. It should have always been around linear pricing because they're not doing a lot of targeting, ergo they're not using programmatic enough. And so therefore, if they're just doing a replace.
Run of schedule.
Million dollars to they should have charged what linear gets, even though there's fewer ads. And you can make an argument that it's a less cluttered environment. It deserves a slight premium. The big whoosh of inventory has really occurred with the LG's, Vizio's, Samsung's, Roku's. FAST channels, as you have pointed out on multiple occasions, are kind of like the surprise winner, the unsung hero where they're chewing up a lot of viewer hours, and they're creating a lot of inventory. And so I think.
Cheap inventory.
Correct. It was always inexpensive. They get to overlay data on it, like ACR data that can really jack a $6 CPM up to $18 . But generally speaking, I think that the Amazon thing's overplayed. Netflix is overplayed because Netflix hasn't really even activated the programmatic to speak of. It's really more of macro. The macro environment for ad spend has been challenging for the last year and a half. And I think a lot of folks were very hopeful that Q3, Q4, 2024 would start kind of a launch pad back to normalcy for 2025. And I guess we'll see how that plays out.
OK. Interesting. Because I mean, when you look at the eMarketer overall numbers for connected television, I think they're showing 12% or 13% growth, which is double digits. So that wouldn't imply that there's weakness in the macro for connected television.
Yeah, but there's just outpacing in terms of supply.
So supply is growing 20%. Overall is growing 13%. And we get pricing pressure.
As weird as it sounds, since I get paid on the CPM, essentially, my percentage take rate is on the value of the unit. You would think that if it halved, then I would have. The reality is there's so much supply, and they're leaning more into programmatic and channels like LG, Vizio, that I'm doing more than I ever had because the budgets are there. They're just buying more units. So it could hurt an individual CTV publisher. But for us, we're kind of not agnostic to the value from a CPM standpoint, but it certainly isn't corollary where if CPMs halve, my revenue halves. The budget stays the same. They just buy twice as much inventory.
OK. Makes sense to me. OK. Great. Questions. We have a lot of time. But with questions, I want to make sure with all these people that we don't have questions. Yes, sir.
[audio distortion] question? Why do the DSPs grow faster than the SSPs?
I think that it's a great question. We're often asked that. I think if you say the DSPs, I think you're referring to Trade Desk.
Period. Full stop.
What's that?
Buying.
I would. OK. I would challenge that one.
I look at the publicly traded DSPs' growth rates. I look at the publicly traded SSPs' growth rates. And the DSPs' growth rates are much more attractive than the SSPs' yourself.
Again, I would say that's a Trade Desk story. It's not a bias.
Let's answer the question. Disagree. By the way.
No, you're much more intelligent than I am.
Go. Let's answer.
I think it's been a story of brilliant execution on The Trade Desk side where they've been able to kind of run the table. They got a lucky break in terms of a bunch of DSP acquisitions that didn't pan out. Oracle bought some. InMobi bought some. Singtel bought some. And so I think that they have been able to be in a pretty much dominant position on the DSP side. I also think that one of the big advantages that they've had is in this third-party cookie world, they got to enjoy the economics of data creation. When you went to buy moms with young kids and you could just put in your shopping cart in your DSP that data segment, The Trade Desk was able to resell that from any number of data sources. And that carried with it terrific margins.
So their take rate might have been 10%. The margins on all the data collection, another 10%. There you have 20%. So if you look at that take rate versus our take rate, it's close to double. What our thesis is, is that that world is kind of sunsetting largely because of deprecation of third-party cookies and the rise of streaming, which was never dependent on cookies. And now the Disneys of the world, the Paramounts of the world, the Rokus of the world, they want to participate in the data economics, not having the third parties do it. And so therefore, they're holding closely onto it. And we feel as their SSP partner, we can participate in those economics. And we already are with someone like an LG with their ACR data. We think it's kind of another chapter where we'll have an opportunity to increase our take rate and margin.
Do you think The Trade Desk's growth rate will come down, yours will come up [audio distortion]
Or we'll just go up, and they'll stay the same.
Disney inventory are you getting, and what kind of inventory is it? In other words, are you getting premium fixed price inventory to sell, or are you just doing the programmatic? I don't know how much programmatic Disney is going to throw out there. [audio distortion]
No, their aspiration, they've said it publicly, is to exit 2025 with over 50% of their inventory transacted programmatically. So a significant bunch of it. And it is the good stuff. It's the premium stuff. The difference maybe in a Disney example than some other examples would be, they definitely want to sell it all themselves. And so they use me as kind of plumbing to make it happen between the ad agency, the DSP, and the Disney inventory. So in that scenario, my take rate is on the lower end of the scale because I'm not piping in the demand.
OK. So let's stay on Disney. Is it your point of view, because we talked a little bit about Netflix, Disney started with you, big client, and then over time sort of moved to their own backbone tech structure. And they made you an ad server, which is the low end of your take rates. Is it your feeling that Netflix will follow that same path?
I think that Netflix is starting with the belief that they want to be very careful about the user experience, very careful about user data in that they're kind of a crawl, walk, kind of run. So I think they'll start off with a lot of direct sold programmatic, not relying upon us to pipe in demand out of the gates. But as they expand to international markets where the dominance of maybe two DSPs, it becomes more of a six DSP play, that I think you'll start to see a willingness on their part to be able to have us be able to be a demand facilitator. So I think it'll look a little bit like Disney out of the gates. But I think it will evolve much quicker than, say, a Disney evolved, largely because they're not burdened with linear inventory and a linear sales structure.
They're building from scratch. They just hired a brand new head of sales a month ago, and so they're very programmatic first, programmatic driven. I think they believe that there's value in having salespeople for big sponsorship events, selling NFL games, but if it's just a normal Netflix show and it's a 30-second ad, why do I need salespeople involved in that?
OK. Questions, you guys? Any other? Yes, sir. Hi, John.
C PMs on CTV in the early days for Disney, et cetera, should have looked more like linear because they weren't targeted. Does that suggest that when targeted ads become more prevalent, CPMs will go back up? If so, what is the tipping point for that to happen? Or does targeted ads simply become table stakes [audio distortion].
That's a great question. I think targeting will be important to some advertisers and not as important. It's amazing how many people are still like, I don't really care if an eight-year-old sees a Mercedes-Benz commercial. It's aspirational. One day they will buy a Mercedes-Benz. I'm building a brand. Where there are some advertisers who will be like, that's ridiculous. I'm advertising to an eight-year-old. So I will pay higher not to reach that. And so I think that you're going to see kind of the brand broadcast guys that will have a lot of spend, will be able to dictate pricing on that, and then a slew of advertisers that are new to the medium that will demand targeting. And if it all backs out into their ROI, they're agnostic as it relates to $40 CPM versus an $8 CPM. It's all about the performance of the unit.
Other questions? OK. Let's move to Amazon Prime. So when you think about this closed-loop attribution, this retail media, Wall Street's madly in love with RMNs. And so RMN is a retail media network, the best execution of which is Amazon Prime Video sells an ad. And then they tell the brand that advertised how many actual units got sold on the Amazon e-commerce site. And so they can show brand lift, incremental sales from the actual connected television ad. Similarly, Walmart Vizio is the same strategy. You'll put an ad. I think the vision is you put an ad on Vizio. And then Walmart's going to tell you in that town that you advertised how many Walmart online and offline purchases got made. So they're going to tie actual purchases to connected television. Wall Street loves this because this shows clear return on investment of advertising. Tell me what you think about that and what it does to the TAM of what's available to you if you don't have a clear path to purchase for your connected television ad units?
Yeah. Well, you know when you think about a Walmart, Amazon's a little trickier. We'll get to that. But Walmart, to run those ads, they can run those ads on walmart.com. Right? So those.
They're running them on Vizio TVs. It's a connected television ad play.
Right. And so if you start to move off of like when you run it on Vizio, I'm involved in the process. I'm the ad server. I'm the SSP, and they're using a DSP to come and buy the inventory, so it's just a very smart way to buy an ad if they're buying it through me. And because we have so many deep relationships with the CTV publishers, we're the beneficiary of Retail Media Networks. It's not a negative. Criteo buys a ton from me. Walmart buys a ton from me. Amazon buys a ton from us, so even Prime can't satisfy all of Amazon's demand. If they are trying to reach someone.
Do that sentence slower. Even Prime can't satisfy all Amazon demand. I don't understand that sentence.
If you think about video advertising, broadcast quality. So if you're Procter & Gamble and you obviously sell products on Amazon, the unit that you prefer to use is a broadcast quality TV creative. That doesn't really run on amazon.com. Right? So it has to run in broadcast kind of environments.
Doesn't it run on Amazon Prime Video?
100%. But this is where it gets to. But Amazon Prime Video isn't limitless in terms of its inventory. And it may not have that profile exactly of the person that they're looking to reach at that moment. That person that they're looking to reach as an Amazon member could be watching a Disney show. So they're going to advertise on Disney to reach that person. Half the Prime membership, probably more than half, probably never turns on Prime Video. So if you're going to target them in a broadcast environment, you're going to do it off property, not owned and operated. [crosstalk] And that's where you come to us to buy it. So it's just another form of demand that we're able to help facilitate supply for.
I see. OK. Yeah, that makes sense. I hadn't thought of that. OK. Any other questions before we call it? Because we're running out of time, I think. Let me ask Minnie Mouse. Yeah, we have five more minutes. Any other questions? Go ahead. Yeah. And then we'll call it. Go ahead.
Revenue-wise, is your revenue model volume times take rate?
It's a percentage of media spend. Correct.
Right. And so let's say you grow, so I think you, what, did 8% growth in the last quarter? So is volume higher than 8% and your take rate's going down? Or how should we, can you fill, how do you get to 8% growth if it's volume times price?
It was 12%. Yeah, that's fine. So for 60% of our business, there's a direct correlation between spend, take rate. Take rate's been stable. That's the DV+ business. So if we grow spend 10%, revenue will grow 10%. It's a little bit different in CTV because we have different products. And one of the products that has been used dramatically by the plus services is one of our lower-end take rate products. And so therefore, you saw for the past several quarters spend maybe growing at 30% clip and revenue only growing at 10%. That has started to compress for one, because folks are using some of our other products and so it's mixed. And the other part is comps. We're coming up against those comps. We're getting much closer to a one-for-one scenario in terms of growth of ad spend and growth of revenue in the CTV space.
[audio distortion]
It's just a product mix. It's not compression in the sense that, not to be semantic, but compression to me says someone came to me and said, cut your take rate or you don't get.