Let's get started here. I'm James michael Sherman -Lewis, and I'm on the internet team here at Citi. Joining us today are Sean Buckley, President of Revenue and Market Strategy, and Nick Kormeluk, SVP of Investor Relations.
Thank you both for being here. Let's start with CTV. Growth accelerated sequentially to 36% in Q2, fairly broad-based strength across the segments. Against that backdrop, has anything changed in budget conversations in the last few years that could suggest programmatic is increasingly becoming the default method for transacting TV?
Great to be here. I would say programmatic is just up and to the right in terms of becoming the default mechanism for transactions in CTV. With that being said, I think there are some pockets where the market has leaned on more of the traditional direct sold model. For example, historically speaking, business tied to the upfronts with the largest players would fall into that bucket.
Increasingly, though, the negotiations between the big buyers and the big media owners through the upfronts are moving money toward programmatic transactions as well, and that's a relatively new trend. That's a historically untapped pocket that's moving toward programmatic, and then live events would be another example that's probably even earlier on.
But historically speaking, either linear pass through, as in the same ad experience you'd get if you were watching on linear, even though you're watching via streaming, or traditional IO-based transactions dominated that space. Now you're seeing more and more business there move toward programmatic as well. So we have some of these untapped pockets that are moving toward programmatic, in addition to the historical video-on-demand space.
Yeah, that's helpful. We've talked about this dynamic of demand fragmenting and broadening out across new buyers. Also on the publisher side, roughly 30 global publishers control 80% of CTV inventory. How do you see the CTV market growing and evolving from here? How is Magnite specifically sustaining this K-growth above-market growth?
Yeah. Just start on the buy side, for sure, the national television business was dominated by a relatively small number of large advertisers. We are now seeing more of these small businesses looking for niche audiences, direct-to-consumer businesses opening up the funnel and moving into streaming television.
The addressability and targetability, and also the measurability provides that in terms of what those advertisers are looking for. We have seen a bunch of dedicated companies pop up on the buy side, and some of which you are probably familiar with, so the likes of an MNTN or a Tatari or a tvScientific, and we have worked very closely with all those companies, and I would say we are one of, if not the largest, supply sources for those businesses.
And they have worked hard to help onboard those small and mid-size businesses, which is a very broad category into the space. I would say a more recent development that we find very interesting is some of the M&A that has happened there. So Walmart with Vibe.co, Pinterest with tvScientific.
I think one of the big challenges has been how do you go out and attract that mid and long-tail advertiser at scale and also retain those advertisers? And doing that as a startup sustainably over the long term, I think we have had some questions about.
But certainly you see Walmart getting involved, Pinterest getting involved. They have those connection points with those types of advertisers, and we think that can really turbocharge the broadening on the buy side of advertisers in the streaming space.
Yeah, that makes a lot of sense. Magnite has this streamr.ai that lets partners help smaller advertisers create, buy, and measure campaigns without a direct sales force. So how do you view Magnite fitting into this SMB landscape?
Yeah. You bring up an interesting point. So we acquired a company called streamr.ai, and the core business model there was bringing down the creative barrier. So how do we help small and mid-size businesses get through that process, which was historically a big blocker? The time and the cost around building a TV creative was an issue.
Now, using AI, we have tools that bring down the time and cost dramatically. We made the strategic decision not to go directly to those small and mid-size businesses, but to license that technology to companies who in turn do.
And so it is a little bit of a different approach from our side, but that technology has been hugely popular as part of our portfolio, and we are working with tons of companies out there to help do that.
In addition, I think at the end of the day, all of those advertisers need to purchase supply through a scaled and effective player, and I think we are very well-suited to be that home, regardless of the type of advertiser.
You think creative is still kind of the gating bottleneck to broader SMB TV adoption? Or, given the advancements in generative AI, are there other pieces in the landscape that are preventing smaller advertisers today?
I certainly think it was, but I think the capabilities are coming along very well there, and I think that is starting to become more appreciated by the market. Also, those advertisers are often coming at it with a little bit of a different angle than a big television advertiser, and so they have more of a performance focus. So making the process of buying streaming look more like what they are accustomed to in other channels like social is also a big part of it.
Makes sense. I want to hit more on TV and the growth drivers, but before we do that, let us talk Google very quickly. Last week, the court ruled in favor of behavioral remedies instead of structural remedies in Google's ad tech trial, which I think is consistent with your view that behavioral remedies would be the more meaningful lever going forward. Now that we have a ruling in hand, which behavioral changes are you maybe focused on, and what do you think about the path from here?
Yeah. Great question. Yes, we have all been waiting, Wall Street ourselves, very impatiently for a long period of time for this to come through. Now, what we do have at the moment is a 1.5-page order, which is, again, still cryptic and still not the full set of behavioral remedies that has been shared with both parties.
So what has been shared with both parties is sealed. It is sealed for 14 days. At that point, what we know is Google and the DOJ will have a chance to redact anything that they would not like seen. The judge will rule on that, and sometime between day 14 and 30, we will get a full viewing of what the remedies are, and what the behavioral remedies will be. So I think at the moment, we know no structural, as you pointed out.
We know there are behavioral remedies, and everybody is overanalyzing, and rightfully so, overanalyzing that most of the behavioral remedies proposed by the parties will be ordered. We are guessing at what that might mean. The Street is trying to guess at what that might mean. The industry is guessing what that might mean.
So today, we know nothing more than that, but it is encouraging, right? To hear most, right, is an encouraging word. What we also know is some of the remedies proposed by parties are in conflict with one another, and I will not go through those, but we do not know how those tiebreakers work as well. But the word most gives us encouragement that, is it just the ad server that will be changed in how it works with the exchange? Is it something on the demand side that will be changed?
We honestly do not know. We would like to think the word most suggests that, but we will find out more as there is more information shared. As a reminder, there is not a whole lot that changes from how we run our business to what this means.
The reason I am answering this question as opposed to Sean is there is nothing here today that Sean can say, "Hey, I am going to take action against this and do a lot of things with my partners, customers, and buyers to be able to influence what this outcome is until we know what that is."
Even then, a lot of what changes on the auction side is a more fair auction that can transact. That means our win rate in those auctions can increase. We already see and have at-bats there, so we already play in that ecosystem.
It simply means our win rate goes somewhere higher than very low single digits, where in the trials Google's win rate was much higher than double digits. So again, understanding what exactly that means, what countermeasures might be taken, we do not understand if this affects the demand side.
That would be a very good guy for our industry and our business. But we do not know if that will be inclusive or not, so we sit here kind of waiting to be able to react. What we have shared in the past is that for every market share point that we would win as part of this, we believe it is $50 million in contribution ex-TAC. Today, as a reminder, we sit at 6%-7% of overall market share. Outside of Google share, we sit at 16%-17% of rest of market excluding Google.
We think that there's an opportunity here, obviously, for that to move and that $50 million in contribution ex-TAC, which is really net revenue. In that case, that flows through at very high incremental margins, 90%-95% incremental margin.
This is meaningful. We don't know the answer yet, so hopefully, as the next couple of weeks pass, 3, 4 weeks pass, we'll have more information on it, but that's the best that we can share. Again, overanalyze that one set of words that I think everybody's laser-focused on.
Makes sense. Consistent execution from here regardless of what happens.
None of this is baked into what Street numbers have, what our expectations around the DV+ business would be. Again, none of that has been kind of priced in, if you will, for expectations yet.
Understood. Let's move on to SpringServe. It's evolved into this kind of OS for CTV monetization, extending into mediation, demand facilitation, and yield management. What SpringServe capabilities do you think are delivering the greatest value today, and how are they translating into deeper relationships with Magnite?
Yeah, I mean, SpringServe has been a transformational acquisition for our CTV business in particular. I think, first and foremost, at a foundational level, when we integrate with media partners in the streaming space, so publishers, there is really two main models. We either integrate as the ad server, which is relatively straightforward, and SpringServe in particular has done very well there with regards to new media.
So if you look at, for example, some of the OEMs like a Vizio or an LG who use SpringServe as the ad server and run their ads business kind of wire to wire on our technology. The other model is what we call a mediation platform. This is different than what you have historically looked at an exchange or an SSP, right. This is more of a control center where they operate most or all of their programmatic business.
We integrate with an existing ad server. It could be home-built or could be another player in the space, but then they house their programmatic business more holistically in our platform. That would include other exchanges and SSPs, some of these DSP direct integrations that are talked about in the market.
They also use us for other capabilities like creative review and things like that. At a foundational level, I think we have a very differentiated integration point with virtually all of the partners we work with in streaming, and we work with virtually all of the ad-supported streaming services, certainly here in the U.S. That foundation gives us massive differentiation between what most others are doing in the space. You can then obviously layer on unique capabilities.
For example, more recently, we have come to market with a bunch of agentic capabilities, which we will talk about. But that foundation is what gives us the ability. We are so deeply embedded in their workflows, and we are such a big part of how their business operates. It gives us a unique launchpad for all the other things that we do.
There have also been some applications on the buy side. For example, our ClearLine technology, the direct buying UI that we offer for specific purposes, was built foundationally on the SpringServe technology, and we have developed marketplaces for some of the big buyers.
For example, some of the holding companies which we have announced, which is also foundationally built on the SpringServe technology when it comes to CTV. Yeah, that acquisition, albeit small on the onset, has unlocked a lot of potential for the company.
Makes sense. Let's move, if we can, to live sports. Banner year this year, World Cup, et cetera, represents about 40% of TV ad spend, but relatively low programmatic penetration. Even as I think you've noted 56% growth year-over-year in live sports ad spend from January to July, which was fascinating.
What are the primary barriers here keeping programmatic from becoming a more meaningful part of live sports monetization, and how is Magnite positioning with some of your newer tools like Live Scheduler?
Yeah. There's kind of an additional step that I like to clarify with folks as the sports monetization moves from traditional television into streaming. You'll hear stats that a huge part of the audience watched X event on streaming. But the reality is there's an additional step, which is you have to move from linear passthrough, which I touched on earlier, to dynamic advertising.
The media owners, even though the audience has moved and is viewing through some streaming service, they have to flip another switch, which is, okay, we're going to enable dynamic advertising in this event, not just pass through the linear ad load.
Once the switch is flipped for dynamic advertising, that is what then enables programmatic to move in and capture that opportunity as well. Obviously, we're working closely with the major media owners to help them understand.
It's a sizable transition. There's a business decision. There's a high level of technical sophistication, and so there's folks both looking at the user experience and also crunching the numbers to say, like, "Does it make sense to flip the switch on this event?" It can literally be done event by event.
So we're working closely with media owners to help give them confidence that, yeah, this will be a net positive for you financially, and we're ready to support very high concurrency, high-profile live events with our technology, and you're able to do it dynamically.
You mentioned one example of a recent product release, which was a scheduling capability, to have people go in and sort of manually, one by one, set up all these live events, which are often happening when nobody's in the chair from a trading perspective, right? It's happening at night.
It's happening on weekends. So creating a more robust scheduling capability for media owners to go in and do a lot of that work in advance and bring some automation to that was a huge deal. Before that, we had released a product called Live Stream Acceleration, which if we were to just take the traffic on a live event and send it out to DSPs as is, it creates a lot of issues around pacing.
Essentially, you can bet that if we were to do that, it would not function properly. So we have a suite of tools that spreads that traffic, makes it more digestible for the buy side technologies, and therefore helps live events monetize much more smoothly and effectively. Those are two examples of technology we've built to help bridge this gap and bring live into the programmatic ecosystem.
Yeah, I think the topic of sports and flipping the switches brings us to a very parallel subject, which is data. Ad tech data historically being concentrated on the buy side for targeting, but also first-party data CTV now increasingly sits with publishers. How has the balance of power really shifted between the buy side and the supply side in advertising, and what role does Magnite play in the data landscape?
Yeah, this is an important trend that I think the market is really starting to pay attention to. If you go back and see TV, we have been doing this for a long time. The cohort I mentioned earlier, the OEMs, smart TV manufacturers, they have historically done it this way for a long time, as in taking their valuable first-party data, their ACR, automatic content recognition data, and enabling that through the supply side technology, Magnite, and saying,
"Okay, a buyer can come and say, 'I want to target this segment,'" but when we activate that segment, we are going to do it on the supply side preemptively and then send that traffic out through a deal, and that traffic is already refined against what the buyer was looking for.
Versus take that data and hand it over to buy side technology and then start to lose control of how that data is used. Except for very few exceptions, the OEMs have remained very firmly on the supply side track in terms of how they activate that valuable first-party data. What we have seen happen over the last 18 months, 2 years, is other folks are seeing the value proposition of activating on the supply side.
Commerce players for sure, and even the major buyers. We have done a bunch of work integrating the data assets that have either been built or have been acquired at the major holding companies, and they are seeing the value in integrating and activating audiences on the supply side.
I think the Walmart announcement that we did with Walmart a few months ago was a big signpost for the industry that was like, "Wow, if Walmart is moving in that direction, this is probably a trend we need to pay close attention to." That has accelerated this conversation meaningfully.
But we see it basically happening from all constituents now evaluating, "Okay, if I can activate decisioning and audiences on the supply side, I am starting to really see the benefits of doing that.
Yeah. Makes a lot of sense. As we think about that kind of future of buying, you have buyer and seller agents that are being tested with a handful of partners, a handful of million or so of transactions to date. Today you just launched your first agentic campaign in EMEA, I think with M6 France.
Right. Yeah.
How do you view the broader adoption curve for agentic advertising capabilities, and moving advertisers from really kind of experimentation to a true scalable channel?
Yeah, the first set of announcements that we had, we brought our own buyer agent, our own seller agent to the market. So enabling partners to use our tools, tap into the efficiency and natural language capabilities of the LLMs, and start transacting that way. Very different workflow.
You are prompting and the buyer is, for example, prompting and pushing or uploading a campaign brief and saying, "Hey, here's what I want to execute on." More recently, we announced our orchestration layer. This is, I think, a unique setup in the market because it builds on the foundation that I discussed earlier.
I really think you have to have the right depth in terms of your partnerships with the major media owners, and you really have to work with everyone that the buyers want to reach, and we are uniquely positioned in that way.
What we are doing there is we are enabling the media owners to upload their full inventory footprint, all the packages, the different offerings, and they can even customize it by buyer. We have this amazing repository of all of the inventory intelligence from the major media owners now around the world.
That gives buyer agents, whether it be our own or we are interoperable on this front, so we are allowing buyers to bring their own agentic tools and integrate. We put it in a machine-readable format, so essentially those agents can now evaluate the broader inventory landscape and execute on campaigns. Most of the transactions have been one-to-one in nature to this point, so it has been one major buyer interacting with a major media owner.
We certainly see a world where it moves, for example, to one-to-many, where it's a buyer who wants to interact across many media owners, and I think that reinforces our position and makes it even more important.
Do you think one-to-many is the future of agentic buying? I think we heard from the last speaker that MCP is now table stakes. I think you certainly share that view. How do you see the ecosystem evolving one-to-one versus one-to-many over time?
I think the reality is, particularly in CTV, you'll see both. There's certainly an opportunity for agentic to transition some of the remaining traditional direct sold, right? There has been some barriers with programmatic, and I think agentic could create a streamlined option to tap into that pool of budgets, which would also be upside for us. I also think you're inevitably going to see the multi-publisher and optimization across media owners for sure. So I think both.
Makes sense. The flip side of LLMs and advertising is called the traditional web. I know you've spoken to the open web and display portion of DV+ has faced, I think it was high single digit declines from referral traffic pressures, potentially in part from AI overviews.
Even as your mobile in-app and commerce media have returned DV+ to overall growth. How do you think about the rate of change within DV+'s portfolio mix, and how are you helping publishers adapt?
Yeah. So I would probably classify the web category. It is facing structural headwinds, not necessarily specific to us, but industry-wide, as you alluded. It is a meaningful portion of the DV+ business, but it is not the majority. We certainly have other areas of the business like mobile app, audio, digital out of home, and those, for the most part, are material growers.
I think what will be interesting from a perspective standpoint is as you look out 2 or 3 years into the future, I have little doubt that, as a percentage, web will be a notably smaller part of that business. Mobile app in particular, but certainly audio and digital out of home will be a notably larger part of that business.
I think that will perhaps change the perspective on the DV+ business, and we think those other formats have material long-term growth opportunities.
Makes sense. Can we talk briefly about commerce media?
Sure.
More than 20 deployed commerce media partnerships across DV+ and CTV. Very long kind of enterprise list. How do you think about the growth opportunity here? Maybe the go to market and the path to scale together.
Back to your earlier point around supply-side decisioning, I think this is a category where that is showing up in spades, and increasingly, the major commerce players are choosing to activate their data on the supply side. There are a lot of reasons for that, as we alluded to earlier. In this case, just to highlight one, workflow flexibility is becoming very important for this cohort.
If you were to integrate your data into a DSP, you kind of have that as the workflow model you have to bring to your end customers, which are brands and agencies as a commerce player for the most part. If you integrate your data with a supply-side technology like Magnite, your buyers can then show up with their preferred workflow. For example, whichever DSP they are currently using.
If they are not using the one you had selected, you no longer need to sort of go through a workflow process change, which can be very challenging in our industry. That workflow flexibility aspect is one of a number of reasons as to why the commerce players are drifting in this direction.
I mentioned the move or the announcement we had with Walmart, which obviously is one of the largest constituents in that space, and we have made a number of announcements there, and so we have a lot of momentum in that category.
Great. Should we hit on financials now? Your EBITDA margin guidance was raised to at least 37%, with free cash flow guidance, I think in the high 40% range. How do you think about capital allocation in particular, following your deleverage with the debt paydown, ranking buybacks, organic reinvestment, and potential tuck-in M&A.
Sure. I will start with the operating leverage and touch on capital allocation. Because we have paid down debt to zero, I will punt that over to Sean, which will play a bigger role in capital allocation now that we have a lot of free cash flow. On the operating leverage side, we have talked about for a long time that at roughly 7%-8% annual revenue top-line growth, we cover our costs.
We are about margin neutral there. What we have said is once you start getting above that range, you will start getting a little bit of margin contribution. For the last few years before this one, we grew in the 10%-11% range and started to see about 100-150 basis points of margin expansion. This year started with EBITDA margins in the high 34s, 34.89% from a street expectation perspective.
But as revenues inflected to now get to the, call it, 13%, 14%-ish range above 11%, you've seen the operating leverage really play out as expected. I think why it's caught a little more attention, and that was a bit of the surprise question that came in following this earnings report.
We know that it performed as expected, and our $10 million revenue beat on the top line in Q2 translated to an $8 million EBITDA beat on the bottom line. The operating leverage played out exactly as we indicated it would.
But now that we actually performed it versus just said it, I think it got a little additional attention. So we think that that will hold, we think above 13%, 14% growth rates. There is 80-plus percent flow-through for top-line revenue beats down to the EBITDA margin line. So you've seen it in operating and EBITDA margins.
You'll continue to see that as we're growing at a sustained elevated rate. What that allowed us to do over the last several years is to be able to pay down our debt. So we started, and frightfully so for some investors with a 6x leverage scenario to buy the acquisitions that we did.
We haven't been in market doing any large deals for the last four years, so it's really been organic growth since then. But we've paid down. We're now net debt zero. So we've gotten to a position where the vast majority, especially when our shares were at far lower levels than they are today to start off the year, with fairly aggressive buybacks in the market both from a withhold to cover perspective as well as open market purchases.
We've also said that greater than 50% of our free cash flow will be targeted towards buybacks. The rest of that is really towards any M&A, and there's not significant M&A out there.
I'm not trying to signal that we're doing anything different in market today. streamr.ai was our latest deal that we did, but I'll turn it over to Sean to maybe talk about the types of things really from a tuck-in and a roadmap perspective that we might be interested in, things that leverage our market position. But let me turn it over to Sean to speak on that in that regard.
Yeah, just to jump on that. Look, we feel really good about the technology that we have in-house, and we're fortunate that we aren't absolutely in need of something here or there. We feel great about our stack and what we're able to do with our customers and what we're able to continue to build organically from here.
With that being said, we always have our eyes open for the bolt-on type of acquisitions, particularly adjacent technologies. We talked a little bit about SpringServe and how fundamentally positive that has been for the business.
Nick mentioned streamr.ai, and I think one interesting part about that acquisition was we talked about the capabilities in terms of the product that they provided to the market in terms of what we initially purchased.
The reality is the knock-on effects of the way we've been able to leverage those capabilities and the team, it has dramatically impacted and accelerated our agentic roadmap more broadly across the company.
We've been able to expand their product, for example, from purely video to the home screen environment on the OEMs and solve some major problems for those constituents, leading to that environment likely becoming much more programmatically enabled than it has been in the past.
We've been able to take a relatively small team and a specific set of capabilities and then apply that across many other areas of the business, which has been hugely beneficial to us. Those are the types of things we are always on the lookout for.
Yeah, perfect. So many different areas to follow up from that, but I want to make sure we take some audience questions if there are any. Raise your hand if you have a question. Otherwise, I can keep going. All right, perfect. We've hit on improving incremental margins. AI is an efficiency but also a product velocity lever.
I think you've spoken in the past about some benefits from the internal load balancer saving $20,000 a day, agents replacing ops contractors. How do you think about this equation of leveraging AI internally for efficiencies, but also trying to drive a little bit of faster product velocity? With that, if I can squeeze one more, how do you think about open versus maybe closed frontier models?
Yeah. I guess to answer your first question, we do have it broken out. There's internal efficiencies and then there's external customer-facing products. We talked a lot about the external side. From an internal standpoint, we certainly are seeing material value.
One example would be that we created our own in-house platform, and what that's enabled us to do is things that historically we've had to send through the full product and engineering pipeline, which has a resource demand and also has a timeline against it.
We've used those capabilities to put in the hands of our broader team outside of products and engineering, folks like our operations group, giving them the ability to now develop those tools without going through the full engineering and product roadmap.
If there's a need to develop an internal capability, efficiency, or tool, they now are able to leverage the AI programming features to do that themselves without having to go through that full internal pipeline.
We've started to see material benefits from that in terms of our team's speed to market, in terms of releasing those internal capabilities and making our internal teams more efficient and effective based on that. We're starting to see material internal benefits as well.
On the AI build-out, you have a new North Carolina data center coming online. At the same time, your $60 million CapEx guidance is, I think, a step down from some prior years. How sustainable is that lower CapEx level against the backdrop of a market that is seeing real price inflation from AI build-outs?
I think Northern California, but-
Thank you. Sorry.
All good. We've released some documentation recently to the market around our hybrid architecture. You'll talk to some folks in the market, and they'll say, "Fully on-prem is the only way to go." There'll be other opinions where it's like everything in the cloud, and in our view, you need a hybrid setup.
Obviously, the cost-effectiveness of the on-prem side is critical, but there are capabilities. We talked a lot about live events. I think that's certainly necessary in terms of supporting those huge spikes in concurrency.
To give you a more extreme example, we did the Cricket World Cup in India, and so having the cloud capabilities to be able to handle that effectively for our partners there was absolutely critical. We are very centered around this sort of hybrid approach.
One of the things we continue to work on is taking the CTV business, which was historically completely on the cloud, and balancing that across, what's our more typical sort of base load need?
Operating that increasingly on-prem, but then, as I mentioned, continuing to leverage the benefits of the cloud where and when it makes sense for things like live events. I think that's kind of our philosophy on a go-forward basis, and we feel very confident in that architecture for our business.
Related to this investment-
Yeah, to that $60 million number that we've given for this year. We think that's a relatively safe number to use in a kind of going future, so we'll get additional CapEx leverage. What we've quoted is that using and running things on-prem versus cloud, especially to the base load that Sean mentioned, we get a 4x efficiency for being on-prem versus cloud.
So everything that we can do on-prem at a base load where you know exactly how predictable your loads are, there is a massive financial benefit to doing so. To the degree that our business continues to elevate and grow at a faster rate, that base load could grow.
So CapEx could follow in support of growth of the business, which again, is success-based capital. So that's really the only linkage to that $60 million going number, but we'll get our number going higher. But we believe we're in the right range as we look towards next year.
Great. Last 20 seconds. CFO transition, David Day set to retire at the end of September. Any update on succession timing or the potential profile?
Not yet, but imminent.
Perfect. Thank you, Sean and Nick, for being here today. This was a great discussion.
Thanks for having us.
Really appreciate it.