Magnite, Inc. (MGNI) Earnings Call Transcript & Summary

Hi, everybody. This is Omar Dessouky. I cover video games and ad tech for the Internet team at BofA. My stocks under coverage are, on the video game side, Roblox, Take-Two, Electronic Arts and Playtika, on ad tech side, AppLovin, Magnite, DoubleVerify, Digital Turbine, Integral Ad Science and Cardlytics. And I have with me today the management team of Magnite. And I would just say that a number of stocks in my coverage universe have already been appreciated, and you can see the stock prices have run up quite a bit. They're amazing businesses, and we're still long on them. But I've tried to point out Magnite to a lot of my clients, and I think it's probably the most underappreciated stock in my stock coverage. So if you're looking at CTV players like the publishers or The Trade Desk or Roku, you should definitely look at Magnite as well because they're a key company in the ecosystem. So gentlemen, let me just jump in with the questions. I don't think you guys need an introduction unless you just want to say who you are, and then I'll move to...

I think it would be helpful for...

Yes. I'm Mike Laband. I lead revenue in the U.S. for Magnite.

And David Day, CFO.

Nick Kormeluk, IR.

Great. So a lot of investors think of Magnite as an SSP company. And they don't realize that in CTV, which we're going to focus on today, the ad server is the key differentiator. And let me also just preface this with the reason we're focusing on CTV is because that's what's expected to be the next big platform for media and advertising growth. So the next generation of SpringServe will be rolled out for the general availability in July. Can you explain to the audience how the next-gen ad server will be better and how it will enhance Magnite's competitive position?

Yes, sure. So over the last 2 to 3 years, we've taken this SpringServe ad technology that's been more focused on an ad server-like stand-alone product designated built for video and CTV. We've also then had this streaming SSP that has been plugged into it for a long time, along with other third-party providers and exchanges out there. What we've recently done for the first time ever is now actually pulled those 2 systems together and effectively built a universe that allows for publishers and media owners that leverage our ad serving and programmatic capabilities to kind of have this like one-stop shop that allows them to centralize their supply, how they operate across the board with the different types of inventory management, also their demand allocation, which is going to allow us in general to build more workflows to make it more seamless for publishers to operate their business. Really, though, what's more exciting is probably on the buy side for that. And what I mean by that is on the buy side, you have all these agencies, all these holding companies, all these premier brands that in the market for the last year have been pounding the table for we want a direct connection to the actual publisher and the supply. We want to eliminate hops on the supply side. Well, we've done just that. What we did is as a part of this implementation is now we can pull the programmatic technology out of what we call the SSP into the ad server itself, and that allows for key DSPs like a Trade Desk, a Google and Amazon to now be directly integrated to the actual ad server itself. You can have curation, things like agencies be built directly into the ad server itself. And now we've effectively made it so it's an easier way for publishers and buyers to come together and execute in a more seamless way.

Okay. So I'm going to get to the sort of like buy versus build question in a second. But one of the things we've been batting around is what is the sort of technological runway for ad servers. And what is like some of the performance metrics that further investment in an ad server by a company like Magnite could drive? How long is that runway? And what new markets and what new TAMs would it open up potentially for both publishers and advertisers?

Sure. I can answer the first part, and I'll let David and Nick take on the second part. But on the first part, I mean, from the ad serving side today, there's a number of different, we'll call it, network effects that you build out as a result of being the place where the inventory originates. And so when the video inventory is being set up in the actual ad server itself, what happens from that perspective is it allows us to be able to help optimizing and help inform the media owner on how best to optimize that inventory. And a part of that is not just monetization, but also things like user experience, things like net new formats that have come into play and really just different ways for buyers to now come in, in a closer way for the actual ad server itself. So where we see a lot of the innovation and where we focus is, one, from a workflow perspective, to building out net new units, so things like pause ads when you're watching television at home and you pause the show in the middle, you see an ad pop up sometimes that's net new placements that are coming into the place. And then finally, like live events is a huge piece where when you're the ad server and then also when you're essentially deploying the activation directly in the ad server for that, it's a lot easier from a live event aspect to bring in more of the linear inventory you see today to be more dynamically enabled for digital that allows us to have more opportunity on that side itself. So overall, there's a lot of different opportunities when the -- when you're essentially the origination of the point of sale, and it just gives us more flexibility to work with publishers itself. Anything you guys would want to add?

Yes. From a TAM perspective, it's a pretty unique offering that exists there. And having that technical capability of bundling the ad server in together really does open the market up. It's not a traditional sell-side platform that we're addressing from a market perspective. We really are being hired as enterprise software to run ad operations for companies. And now what you're seeing is companies are choosing one person like ourselves to run the entire ad stack for them, look at a Netflix, look at a Disney, look at a Warner Bros. Neo announcement recently or some of these marketplaces. They choose one partner to build that out for them, to run it for them, to hold them accountable for it. And really, it's then other components of that stack that others might plug into or play a small part of. So we really do serve the entire market. And again, in that role, you're not choosing 2 partners for enterprise software. You're not choosing 2 ERP-like partners or 2 CRMs. You really do want one source of the truth. The thing that's also opening up to us is new partners. If we sat back a couple of years ago, we never would have said, hey, we've got a good shot at cracking some of the largest accounts out there that have their own internal ad stack. So think of it as an Amazon, think of it as an NBC. There's new inventory sources out there that we're able to service. And I think our integrated platform is truly what makes those bigger opportunities and opens those up to us.

And what is the actual process of improving that technology? Obviously, you guys have coders and software developers of some sort. But like what new technologies, AI, are you potentially integrating into the ad server? Just what is it that's actually going to improve that performance? And what are some of those performance metrics that your clients would look to see improve over time?

I think what's really -- I mean, this is going to get kind of detailed, but just to give 2 examples. If you're a publisher today and you're all watching content, it's a 45-minute show. You have 3 different ad breaks. And in those ad breaks, you're going to see what's called an ad pod that has 3 different slots or 4 different slots. What we've done is build technology to go to the publisher and say, let's optimize what we call these pods throughout the actual show that you're watching. And you can optimize the pod based on time. So if it's a 60-second ad placement, you can put four 15s or you can put two 30s, you can put one 30 and two 15s or you can optimize on monetization. So paying me the most amount of money, Magna, you possibly can by optimizing the pod for the total amount of ads you can find at the highest eCPM possible. While we built a lot of algorithms in different ways and auction mechanics to allow a publisher to optimize that type of pod, and that uses different types of learnings from over time, people to serve into that. That's one example. Another example would be you're starting to see an injection and a flow of more performance-based buyers into the actual ad space itself in CTV. And those performance-based buyers are generally coming in at lower eCPMs comparatively to a large national brand that has a higher budget from that perspective. But in aggregate, this performance-based market is a massive amount of thousands of small mom-and-pop shops that have a ton of different demand behind it, but they don't want to purchase at these higher rates. Well, what we've done is at different times of the day when their campaigns are performing better than others, we can actually call them more. We can prioritize their bids more, we can increase the win rate on that. And then you start to look at optimization on without -- throughout a day on a time basis to then serve into that. So that's just 2 examples of how we look at optimizations at the ad server and programmatic level. There's hundreds of different ways we operate that [ Choice ]. And a lot of the ways we do it is just listening to clients, understanding their needs, hearing how different buyers are operating and investing more on the product side to work closely with the clients to build out these solutions.

Yes. A couple of ways we utilize technology, AI, either large language models or also what we call machine learning. One would be in our curation. And so we use AI to help generate audiences. And so there's a lot of detailed work and a lot of manual effort in that process. And so that's one of the first places that we've implemented AI. And so that's significant for us. And then just from a sheer volume perspective, in our DV+ business, so the non-CTV part of our business, we ingest 1.5 trillion ad requests a day. So it's a massive, massive amount of volume. And a lot of that -- a lot of those ad requests that we ingest just aren't worth very much. And so we use very sophisticated algorithms that are -- they are consistently learning and we throw away 1.1 trillion of those ad requests just off the bat because we don't want to spend the processing time to process them and put them into auction and so forth. And then we run roughly 400 billion auctions a day, and then we're continually learning from that process. And so that's another important part of our ability to scale. So we don't have to add machines in data centers, and we don't have to do more coding and development in order to scale.

So -- and maybe this is the last question before I move on to the buy versus build decision at the publisher. When we think about performance metrics and advertising in general, it's thought that programmatic advertising is generally like low efficiency. Viewers still don't see the optimal ad, right? Or they don't see enough ads and perhaps the fill rate is low. I don't know what the actual state is. But how much room for improvement is there in terms of showing the right ad or the right combination of ads to the viewer? And how should investors think about that and how you're -- how much more room to go you guys have?

So in general, when you look at the personalization of ads itself into these different placements, a lot of it comes to content signals and audience signals that are passed in the bid request out to a buyer, meaning things like what is the site, the series, the show of the actual content itself, advertisers really want that information. Am I going to be in the first ad slot or the last ad slot and what types of audience-based data can be passed in that? The biggest thing I'm sure a lot of people have heard is like universal IDs, which is this way to effectively make it so that there's a common language across the industry of, hey, if I'm using this universal ID for this type of entity or this other type of one, there's at least an understanding of like who that person might be. So what we found over the last year to 2 years as there's been more adoption of the high-level term, which we call an EID -- EUID being Trade Desk one is like one of the largest ones, then Yahoo! has their own and others in LiveRamp. We found that more and more ads have become more personalized as a result of that. There's definitely still room to grow. There's a lot of ways where you can deploy different ways to pull in information about the aspect of a person. That's obviously in a privacy-compliant way to leverage different audiences to then layer on to that. So some of the largest, I would say, media owners out there, like the OEMs like Roku and LG and Samsung, all the different companies as well as some of the major programmers have really valuable ACR data or first-party data that they know based on people's watching viewer habits. And they can layer that data in, I would say, probably better than a lot of these different platforms out there to be able to personalize ads better from that. And so we do a ton of these audience-based types of deals. The audience is profiled and effectively only managed on the Magnite side. It's not passed out in the actual request because of fear of some privacy aspect, but allows the publishers to build out these different types of deals itself.

And I'll add one more important initiative that layers into what Mike just said is, especially in CTV, streamers have first-party data that they're very protective of. And so we've spent a lot of effort to be able to take their first-party data and ingest it into our curation tool, anonymize it and allow it to be leveraged without leaking out into the ecosystem because those streamers don't want their valuable data to be utilized by, let's say, a DSP in their general ID that Mike was talking about and be able to leverage it off property, say, for example. And so we've put some development efforts into that, again, to continually trying to refine and improve and expand our -- the way that we can assist in targeting.

I guess I was kind of trying to get to maybe what's the upside in CPMs from this incremental efficiency, right? Because people like to compare CPMs across platforms. And if one platform is helping CPMs just grow dramatically, that's great for the publisher and obviously great for your business. We hope that's Magnite. But how do we think about that?

Yes. I would say that the easy lift button, which is causing programmatic to be fully embraced is if you layer on data, if you curate an audience, if you're actually delivering against a user, that's the easy button to get a higher CPM. So whether it's Roku or whether it's Disney or whether it's Netflix, if somebody layers data on top of inventory that they're selling, that's the 10%, 15%, 20% lift that you get automatically. Magnite is doing all of that execution to be able to do that. So you're hearing in every conversation that publishers are having and content owners and ad inventory owners have, they're pushing programmatic, and we're enabling that to happen. You think about some inventory, it's not all created equal. You could be watching a show on an app, and you might have 3 people who own certain slots in that ad as part of a carriage, right? The TV OEM might have it, the streamer might have it and then the content owner that has licensed their content out might own it. And then they're optimizing for different things. Sometimes they're optimizing for user experience, meaning they're trying to do as little advertising and have as little of an ad load as possible to minimize potential churn. In other cases, they're just grabbing for money as much as they can and they release those rules, release their price floors, release those competitive processes that allow more buyers to be able to come in and buy. So it really depends on what they're trying to achieve as well.

Okay. That makes perfect sense. Thanks for clarifying. Well, let me move on to kind of the next question. Can, should and will publishers build their own ad server technology in-house? Right? There were some -- I think there's a podcast by Disney at one point that created some confusion among some of my clients and Magnite. Can they build their own and should they build their own ad server technology in-house versus leveraging a third party like Magnite? Can you help us understand the trade-offs and how expensive it actually is to do so?

Yes. I mean like you've seen -- when you look at the total ecosystem of CTV, there's only a handful of companies that could even attempt to do that because of the cost of the original capital to do it. Some of these companies have spent billions of dollars in their own legacy technologies to build out a full ad stack when it comes to the ad server itself. I will say, if you were to ask different types of partners in the CTV space, each one of them would talk about how they might own some aspect of their ad tech ecosystem. That could be just as simple as we own the digital ad insertion technology that sits in front of our third-party ad server or they look at the order management system to look at different ways that they do that. The point I'm making is to build the ad server itself takes a ton of time, a ton of capital and a lot of resource to continually keep it updated. Then you look at the aspects of what plugs into the ad server, and that's the programmatic capability, that's the user experience capability, it's all the creative aspects and workflow. And what we're finding is that as the advertiser side and the buy side continues to inject more and more, we'll call it, third and first-party data into their different planning tools, it would require publishers to then build and support each one of those custom different ones. It would require them to not only support it at the actual data provider side, but also at the DSP level. So there's just a major continually large, I would say, aspect to jump on top of that, that it's hard for these different media to consider doing that, which is why you've only seen a handful of them build out their own proprietary ad server. The rest of them are effectively leaving it to us and others to essentially keep that business and continue to customize on top of something that's, I think, pretty well understood while they can focus on other things like the customization of their ad placements and how they want to manage different other aspects of their monetization platform.

Okay. I'm going to ask you a question that's unscripted here. Do you know how much you've invested in your ad server since you've bought it, like in terms of improving it? If you could tell us, that would be interesting to know.

Well, that's tough to quantify, but we have a product and engineering team of, call it, 350-plus people. And we've had a good chunk of those folks for the last 3 or 4 years. I mean, call it, half of that team working on our technology since we merged with Telaria in 2020 and purchased SpotX in -- and that ad server SpringServe in the middle of 2021. So it's a pretty significant effort that we've continually been working on.

I think what's really unique too about our ad server is that we have these different buy-side tools that build out curation and SPO aspects for buyers today. So like ClearLine is a big name that you see us talk a lot about in market. You have different buyers that are using that ClearLine tool to effectively aggregate inventory and then either load up an asset to target inside of it or resell to a third-party DSP. Well, it's interesting because we essentially took our SpringServe tool set, and we just made a buy-side version of it. So what we've done is now any time we build development and code into the SpringServe tool set, it automatically makes it available to our buy-side tools, too. So it's this effect that we allow us to say we're improving a targeting or a pacing algorithm, that targeting and pacing algorithm can now impact buyers that leverage our tool on this side. We've purposely done that. It allowed our engineering teams and product teams to work more closer together on these different opportunities to enhance overall the product itself. So there's another aspect to that and how it has come all together from that perspective.

Okay. Very interesting. Then just maybe sort of an educational question here. Is an integrated ad solution, ad server plus SSP always more performant than the combination of different point solutions. And I suppose that one could start to ask whether the SSP market itself will consolidate around Magnite possibly.

I mean we believe that in general, I mean there's always going to be workflow enhancements as a user and having one tool they can operate in. And there's other different ways we can optimize things in an easier way. And just in general, we do have to be interoperable with how a publisher might operate. So if someone does have a proprietary ad server, we operate and integrate with them in some sort of advanced way to make sure it's easier to essentially operate their business. But from the programmatic landscape in CTV, I would say the origination of inventory really stems from 2 major players out there and other players that are trying to get in and trying to build into the CTV space are really plugging into our systems today. So if you look at the hops and you look at from what a buyer sees, they want to find the most direct way. Well, the direct path is directly into us and one other player. And I think we've done a really good job articulating that and building it out to the buy side, too.

So you guys have, at least in my opinion, kind of been on a tear in terms of announcing deals in the recent past here. The most recent one was with Amazon. And I think it's worth maybe explaining the significance of that deal to the audience here. Justin Post covers Amazon, my boss.

I might touch on it and turn it over to Mike since he's closest to it. I think the big significance is this is really Amazon for the first time for the portion of inventory they own some of, meaning partner inventory, allowing somebody else to transact in inventory they own. So this is on Fire. Previously, we had partner inventory, both in Fire and in Prime. But the significance is really of strategic significance, meaning that they're trusting Magnite and probably finding in our partnership some value to us monetizing on their behalf and not just allowing the partners to transact. So Mike, maybe walk people through exactly kind of what this looks like and what a good example to kind of illustrate what we're talking about.

Yes, sure. So for a long time, we've worked with Amazon Publisher Services across both of our DV+ and our CTV business. We've enhanced the implementation to allow for a third-party supplier, we'll call it like a Warner Bros. Discovery to be able to more effectively sell their share of inventory they might have on a Fire TV or even Amazon Prime to Magnite and then operate their business through that way. So it is an enhancement of that integration. But now also, as Nick mentioned, Amazon on the Fire TV side has been able to effectively take a seat on Magnite and be a "Publisher" in our ecosystem for us to help start to monetize more of the share of the inventory that they get to, to operate and bring more monetization from that perspective.

So our relationship with Amazon just continues to grow, which you wouldn't have thought would be possible a couple of years ago.

Yes. So I mean, Amazon is a very specific animal, right? Obviously, they have very high standards. But maybe if it's possible to say, the fact that Magnite is at Amazon standard, what does that say about where Magnite is potentially for the rest of the big publishers out there. Is it as simple to say that...

There used to be -- was it CBS Must See TV or whatever. I think what you're seeing in all of this is a real flywheel that's developing. And so where we represent somewhat unique demand on the demand side. We have unique supply on the supply side. And so if you're a buyer, you just -- you really need to go through Magnite. And to your point on the supply side, we're just -- we're one of the channels you almost have to go through. And so that's creating this push and pull and this exchange and this kind of virtuous cycle that -- and you see that in the momentum that we have within Amazon, with other publishers and with other buyers. And so that's lifting the business in a really significant way.

That makes sense. Flywheel tipping point. We're familiar with those terms. Great. So maybe with the last 5 minutes here, let's touch on DV+. And on the Google antitrust ruling, maybe could you walk us through the different scenarios that you see could potentially play out and the implications for Magnite?

I'll start, and then I'll hand it over to David. So I think from -- I think we initially, just like the rest of the market, looked at this with a really, really elongated time line where who knew when something actually would happen after appeals and after rulings, et cetera, et cetera. There was a pretty significant change that happened about a month, 1.5 months ago. And this is right before the judge broke for recess before the remedy hearings begin on September 22, talked about separating behavioral remedies and structural remedies. And even though both can be appealed that the behavioral remedies could be implemented and in place even during an appeal process. So that accelerated the timing of which something could happen into the 2026, first part of '26 at the early part. So we anticipate the September 22 remedy hearings will take anywhere from a few days to a few weeks, similar to what's happened on the search side. The judge will then rule and then we'll look for what is implemented. Google, even as a worst-case scenario, has put forth some behavioral changes that would shake some market share loose. Now clearly, they could retract that. Clearly, they could appeal that. But I think even that starts to move the needle on some share redistribution. Could that be 1%, could that be 10%, could be 20%? Everybody's got their decision tree, their NPV, their time frame for doing so. But we're pretty excited the fact that some things will change. When we talked a little earlier about our ability to service this, this really does come through at a very, very high margin, and David can kind of walk through an example that we used on our last earnings call, but there's really not much that we have to do because our machine learning tools and how we shape and how we win inventory doesn't need to be adjusted for this relative to things like a political campaign and sending the right traffic to bidding activity that naturally happens in our ecosystem. So David, maybe share that example for folks that didn't...

Yes. So to put this in context, we estimate that Google has roughly 60% market share of this DV+ world and that we have mid-single digits. And so there's -- and we believe we're the second largest player. So there's a huge discrepancy in market share. And they've built that market share because they've -- Google has had their thumb on the scale. So this monopolistic behavior has created an uneven playing field. And so now with these behavioral remedies going into place, we estimate that 100 basis points or just 1% of market share shift to us would represent roughly $50 million of revenue to us. And to put that in perspective -- and the flow-through of that is roughly 90% flow-through to adjusted EBITDA and then 100% of that would flow through to free cash flow. And so it's really significant when you consider that -- 2024, we had $118 million of free cash flow. And so you layer on a couple of hundred basis points of market share and you're almost doubling your free cash flow. And so it could be very significant. And I'm not limiting what kind of market share we could take to that 100 couple basis point example. So it's significant, and we'll have a little more visibility, we think, into this from what will be put into place from a ruling perspective, a remedy perspective by the end of this year, and again, could see benefits early next year.

So I found it intriguing, and I think it's worth the audience knowing that I think one of the reasons you mentioned that the flow-through would be so high is because you already process a lot of those bid requests, right?

Exactly. So we're seeing them all, but we're losing the auction to Google because they've got their thumb on the scale. And so now all those sunk costs and our win rate goes up. And so it's just pure upside with no additional cost.

Right. And what were kind of the numbers there? You had laid out like 1.5 trillion bid requests and then you kind of broke that down like 500 million of those are Google. Would you mind going through those numbers just as we finish up here?

Yes. Yes, we see the 1.5 trillion bid requests. We run about 400 billion auctions. We don't share our exact win rate of those 400 billion auctions, but it's very, very small. And so it doesn't take a lot to really move the needle on our win rate.

Got it. All right. Well, we're out of time. Gentlemen, thank you very much.

All right. Thank you.

Thank you.