← AdTechGod Pod
Loading summary
Transcript22 lines
- [00:01]
A
Hey, this is Ari with Market, and unless you've been living under a rock, you've probably heard that Market Live is coming up October 27th in New York City. The last Marketecture Live was sold out, and this one will surely be as well, with speakers like Mark Grether of PayPal, Eric Seufert of Mobile Dev Memo, and Jenny Wall from Videoamp. Plus, I'll be recording my podcast live with the one and only Antonio Garcia Martinez, author of Chaos Monkeys and now part of the team building at Coinbase. It's a stacked agenda and we hope to see you there. Go to marketlive.com and grab your ticket while they're still available. That's marketlive.com this podcast is brought to you by Vyant. Still investing in paid search? Hoping to grow your business? Tides are shifting. AI tools have arrived and in the end, search is simply capturing existing customers that already know your brand. The real growth, it's in creating new demand on ctv where attention is high and co viewing is the norm. Vyant's AI powered DSP puts your brand on the largest screen in the home across premium streaming content, including live sports, driving real outcomes. See why Vayant leads in CTV@vayantctv.com that's vayantctv.com welcome to the Ad Tech Godpod.
- [01:21]
B
Your window into the world of advertising technology and the people behind it. I'm your host, AdTech God. Welcome to the AdTech God Pod where we're joined by the builders of ad tech companies. Today's guest is Joe Hirsch, CEO of Swivel and former CEO of Springserve, which was acquired by Magnite in 2021. Joe has a track record of building and leading some of the most innovative companies in the space, especially in the CTV and OTT space. I'm familiar with Joe and his work, so I'm excited to have him here. Joe, welcome to the pod.
- [01:57]
C
Thanks Ad Tech God.
- [01:58]
B
Thank you. And thanks for taking time out of your busy day. I know right before we started you said you had a lot of things to do, so thanks for making this one of those things you had to do.
- [02:06]
C
Yeah, well, I just prioritize you over those things.
- [02:09]
B
That's amazing. Joe, question for you. I'm familiar with your work at SpringServ. I'm familiar with the acquisition that happened at Magnite, but I'd love for you to take it back a little bit. Just give me an idea of how you got into the industry and what brought you to the point to founding and starting Swivel and what, what purpose are you trying to solve with it if you want to take it all the way back to start.
- [02:29]
C
It's a long history. It starts all the way back in junior high. Originally I was involved in affiliate marketing as a youth. I remember searching online. This would have been something like Dogpile or Lycos, a pre Google search engine for ways to make money online. And it led me to affiliate marketing. And the first affiliate program that I ever used was the Amazon affiliate program which at the time gave you 15% rev share on all sales driven to Amazon products. That was primarily books since that's what they sold at the time. And because my primary interest was video games, the books that I was promoting via affiliate marketing were for the cheat codes on Super Nintendo games. Those were pretty popular back when I was in junior high. And you know that that kind of led me down a path which is well, now that I'm an affiliate marketer, how am I going to drive sales to my affiliate offers? Inevitably I figured out that you could drive users to an affiliate offer via display banners. Which kind of led to kind of an early ad network that my partner at the time, Hagen Major and I started called Haguenet Media Network. We built Haignet up to represent over a hundred websites. At that time those websites were primarily video game websites but also some professional wrestling was back when the WWF was still popular. Again, another childhood interest. Eventually that network grew. We renamed it to Click Hype. I ran it through college. During college, my freshman year I got evicted from the dorms. And so I, I asked my brother who also went to the same college as I did. We were at the University of Arizona at the time. I asked him, you know, is there any place that I could stay? And he recommended that I stay with this, this guy who had a, a spare top bunk in the fraternity house. And that guy's name was David Zaplatol. I stayed with David Zaplatol. I taught him how to use the ad server which was kind of the primary mechanism for serving banners in the affiliate network at the time. David went on to co found kind of CPX Interactive which was the Post Buds Inc. Ad network slash affiliate program. And then around junior year I sold that company to a group out of Los Angeles called Spark Studios. At Spark Studios I ran affiliate offers and kind of continued to sharpen my ad operation skills. Ultimately I left Spark Studios and started Yellowhammer Media Group. Yellowhammer Media Group was also a huge kind of performance marketing house. Performance marketing was, you know, still Very popular. I mean this, this would have been 14 years ago and performance was still kind of driven on the open web. And primarily when we started, we were big into popunders because pop unders were driving double digit CPMs. You know, that was kind of a wild and exciting time. As Yellowhammer evolved, we, what we found was we needed a solution to kind of improve our ability to serve video ads. We looked at a variety of solutions on the market. We looked at things like Live Rail, we looked at things like Adapt TV and Bright Roll and Freewheel. And none of those kind of met our needs at the time. So while we were at Yellowhammer, we co founded another company, SpringServe. Inevitably we sold Yellowhammer, reinvested our resources into SpringServe, SpringServe being a video ad serving platform. We built and ran springserve for eight years and then ultimately sold it to Magnite. So that's kind of a brief history of the universe.
- [06:21]
B
So moving from the affiliate market into the ad serving market with springserve, I saw an article from Ad Exchanger and the title literally said four guys who built ad servers are out to fix ad servers. And so post acquisition of springserve, post your time at Magnite, what do you mean by fix ad serving? What gap did you see in the market that you felt that it needed to be addressed? And why did you create Swivel?
- [06:45]
C
I think it was a personal experience, you know, at springserve for a considerable period of time I ran the kind of operational side of the platform and interacted with all of the clients, the ad ops teams, et cetera. You know, I observed kind of the change logs of the ad server, I. E. Everything that somebody did that, you know, required that they save an action was registered to the change log such that I, you know, had kind of a encyclopedic understanding of every workflow that could possibly exist. Post insertion order for both direct sold and programmatic campaigns for all things video and also the home screen at the end. The thing that you see most is that there's so much repetitive labor in operations that prevents operations people from doing higher order work. And so when I think about fixing it, the ad servers aren't necessarily broken. It's the business model such that you require an infinitely scaling number of ad ops people to address a scaling business. Traditionally it's kind of a stair step. And that stair step is, you know, for X millions of dollars under management, I need Y heads and you continuously need more people as you manage more budget. But those people are, they're all doing the same thing. They're all kind of doing repetitive tasks. And so whether there's something broken with the ad server or not, it's broken insofar as it's hard to take a group of people and give them the tools that they need to scale a business, you know, basically to. To infinity. So I think about those workflows as broken more so than the ad servers themselves.
- [08:42]
B
Great. I mean, I do recall working on multiple ad servers. It's been quite some time now for me, but a lot of that is repetitive. Is Swivel just learning from your actions or are you commanding it? How does it work and integrate with the Adopts process to help make the right decisions on behalf of the Adopts person?
- [09:01]
C
Sure. Well, I think before I answer that, you said something critical, which is multiple ad servers. And if you think about the name of Swivel, Swivel kind of originates from a problem space that internally and many people refer to as Swivel Chair. And Swivel Chair is just the operational burden of operating in multiple platforms, in multiple tabs on your browser and moving between those browser tabs to kind of execute your operational needs. So if you think about just rudimentary operational workflows in the ad server, you know, after the sale is complete, you've got like this insertion order. And the insertion order goes into something like a CRM or an order management system. And then depending on the sophistication of your order management system and your operational organization, the insertion order could be automatically trafficked into an ad server or an activation platform. But what we see is that it's partially trafficked. It's usually a shell, and the shell is inclusive of some order details, but not all order details. And then an operator is responsible for kind of finishing that trafficking operation. That could include things like adding the creative, adding audience, targeting, associating inventory, excluding sensitive attributes. All the minutiae that is kind of hard to capture in the order itself is usually executed by humans. If you, if you just think about like as the flow continues, you know, the order is trafficked, okay. And then after the order is trafficked, it's flighted, it's scheduled, it goes live. You know, whether it's programmatic or direct. Once it's live, there's usually some in flight campaign optimization that includes things like making sure your campaign is budgeting, making sure your campaign is pacing. And then of course, the end all, be all goal, making sure your campaign is producing the KPI that the advertiser intended. That KPI could be return on ad spend, that KPI could be, you know, some kind of in store transaction or test drives. I mean these are all the performance metrics that you hear all the time. And then after that, you know, you do that kind of performance check in on a regular cadence. Depending on the length of the flight date, you might do it, you know, if you divide the flight date into quarters, you might do it four times per the quarter of the flight date. And then, you know, if you're a seller of media, there's a process in between yield management where you're, you're trying to maximize the value of your inventory. After you've done that, the campaign finishes and then you have post campaign reporting. The post campaign reporting process, again like another manual workflow, usually involves pulling reports. It could involve pulling reports from multiple platforms. You make this kind of post campaign report, you send it out and God willing the advertiser rebooks all of those things, right? Each one of those workflows or operations requires a multi step process or an orchestration between multiple platforms. And those are all processes that Swivel seeks to automate, orchestrate or allow kind of a one off operation within the platform such that, you know, all of the things that I said you might do infrequently, you might do them across a certain number of objects. You can only do them, you know, when you're awake. You can only do them, you know, when you're at your desk. And Swivel kind of allows you to do it at scale across all objects, 24 hours a day at an unlimited capacity.
- [12:53]
B
How does Swivel, what is the, I guess the process or how does it work when, when trying to optimize for your, your advertiser's outcomes and what they're looking for is it just utilizing the data that you input or the campaign parameters that you have and then optimizing the same way a person would by hand. But this is now automated based off of the goals that you've set. Is it writing on behalf of you? Is it like doing the optimization on behalf of you and fully automated through the process?
- [13:24]
C
I think that's a incredibly important thing to discuss. So when Swivel kind of starts its process, we kind of have a three phase process. So phase one is to replicate human behavior, as you mentioned. And so the replication of human behavior as it pertains to adopts, I look at it in this kind of black and white world where you have this thing that we refer to as a playbook, and the playbook is Sometimes it's well documented, sometimes it's kind of tribal. And so I guess when it's tribal, it's not as it's, it's a little more opaque. When it's documented, the playbook includes basically the instructions that the organization has acquired over time for how an operation occurs. So, you know, let's just say that a campaign is under pacing. Well, there's an order of operations. It's like, okay, the campaign is underpacing. What do we do next? Okay, and the first thing that you might do, you might have the budget or the budgeting set to even daily pacing. So the first thing that you might do is change even daily pacing to daily asap. And that would be in kind of this playbook that you have. And if that doesn't resolve delivery, you might say to yourself, well, why don't we change the frequency cap? Why don't we increment the frequency cap up? And that'll allow, you know, the same number of users to get displayed more ads and that'll allow the campaign to spend more freely and hopefully that'll allow the budget to be spent. And if that doesn't work, maybe we increase the aperture of the audience by looking for a similar audience from a different audience provider. And if that doesn't work, what if we go from zip code targeting to DM it? And all of those things are the replicatable processes that we first seek to do to resolve different operational workflows. So after we replicate human behavior, our second goal is to scale it. So if these processes occurred, like I was saying, you know, four times per campaign flight, the scaling element is taking that time period and condensing it. So if something used to occur every seven days, we might do it every day. And if something occurred every day, we might do it every hour. If something occurred every hour, we might try to do it intra hour. And then lastly, our goal would be to attempt to improve human behavior. So if a human had a playbook, and we thought that the playbook could be improved, that would be like the final evolutionary step. So just to tie it back to performance, which you had kind of honed in on, let's, let's think about what a human operator might do to improve performance of a CTV campaign today. Okay, so the first thing they might do, and most operational workflows start with a report. So you might run a report by app name, app bundle, okay? And then you would look at that report and you would say, I mean, hopefully you have some kind of attribution and your attribution is like, the majority of my conversions are coming from these 20 app name app bundles. I'm going to exclude all the app name and app bundles that aren't driving conversions. All right, so that would be like a performance optimization playbook that we could replicate and then make it run on a schedule at a certain frequency. And then you can compound those playbooks. Like if the next thing that you did was maybe you ran a report by geography to see, you know, where the majority of my conversions coming from. Are they coming from coastal cities? Are they like urban or rural or state specific? I might make another list. And that list might be an include or an exclude to narrow the aperture of the inventory such that I'm targeting a subset of inventory that is a higher probability of producing a conversion outcome. And this just kind of is a never ending iterative process such that the way that humans do it, we want to do it as well.
- [17:45]
B
Yeah, I mean, you mentioned something on the frequency side. You're number one on the frequency piece. You're right. I mean, optimizing once a day quartiles throughout the campaign is one thing. Being able to optimize within hours or hour of that required optimization is potentially huge. I mean, one thing for me is on the frequency side of these optimizations, some of it is not scalable. I mean, you cannot have someone running reports every hour making decisions, are utilizing the data that you receive through, you know, an API to make those decisions quickly. Because there's a lot of data that pumps through that even a very experienced ad operations person cannot possibly process that much data within an hour or even sometimes a day to make educated decisions on their campaigns. Is this also a piece of like the back end, how you're ingesting the data, how you're analyzing that data and you're pushing out decisions. Is that a big piece of what happens on the, on the back end of the tech?
- [18:44]
C
Yeah, absolutely. And I think the, the closer and closer that you get to real time decisioning, that granularity continues to improve performance. You know, one, one thing that I would note is that we're currently limited on the downstream platform's fidelity. So there's a couple of elements. One, how quickly can you get the data? And I think you mentioned that. And we can only execute an incremental operation when we get incrementally new data. So if we can get the data every 15 minutes, then yes, we could do an operation or an automation every, you know, four times an hour. Again, like four times an hour across a thousand Objects or thousands of objects in an ad platform. That's still a tremendous amount of actioning that today. And we see this all the time. So I can, I can go into somebody's ad platforms and I can look at how many actions they're doing a day, and a team might be able to produce a thousand or 1500 or 2000 actions in a day. And Swivel can take that number and 10x it, we could do 15,000 or 20,000 or even 100,000 actions per day. Just as a byproduct of having a machine augment what a human already does.
- [20:07]
B
Incredible. Where do you see things going, Joe? You know, AI is obviously a massive piece of your business. You're leaning in very heavy into it being able to utilize the data, the optimizations, the integrations to just drive better outcomes for your clients. But where do you see the overall industry moving? And it doesn't have to be about AI, it can be about CTV or advertising. What trends are you seeing? What trends are you hearing about? And what kind of excites you for the next 12 to 18 months?
- [20:34]
C
So today the primary user of ad platforms, except for a handful of platforms like SpringServe, is a human, right? Humans today are the primary person that logs into and pushes buttons and pulls levers in ad platforms. But the future of adopts is that an agent will be the primary user of the ad platform. And you see this on the open web, right? So today the primary user of a website is a human. But if you look at trends, you can see that crawlers and agents and scrapers are quickly becoming the primary user of websites as people start using natural language or AI or something like ChatGPT or Claude or Perplexity to execute search. And so as you see that humans no longer are the primary person that interfaces with a platform or a website, there are certain inevitable conclusions, which is, well, now they're interacting with something else. And that something else, at least in my opinion, is something like Swivel, which is this ability to augment what you were doing by overlaying artificial intelligence onto what you were already doing. And you said we're leaning into AI, and I agree with that. But there, there are certain things to keep in mind. One in ad ops, unlike creating like a media plan or an RFI or communicating with your data via natural language, in adopts, precision is paramount. An adopts error is a costly mistake. And so you're not trying to use AI or natural language or generative AI to override a human decision. Your goal is to Keep a human in the loop at all points in time such that a human is consenting to the scaled workflow that is being executed. Because if the machine is now making decisions on your behalf and the machine makes a bad decision, well, that's going to reflect poorly on Swivel and it's going to reflect poorly on the ad ops person. So yes, AI is critical. AI can improve human behavior, but it doesn't mean that the human is no longer involved in the decision making process. The human can now just do an unlimited amount of ad operations.
- [23:12]
B
Thanks for touching up on that because there is a lot of nervousness in the market that a lot of the roles that are hands on keyboard are going to be replaced by AI. And even though, sure, maybe you don't need 25 AD operations people, you just need a less number of them as you scale your business. This is truly a tool to help optimize and improve performance, not necessarily to replace the human element of it which is controlling, watching, monitoring what's happening and making and suggesting changes.
- [23:43]
C
Yeah, I think that's, that's very accurate. And adopts is not unique in this space. Operations in general across all sectors are, you know, they're being augmented by artificial intelligence. And I think the pattern that I see is like if you're an operator and you ignore the trend, I think that's more dangerous than kind of adopting it and becoming an expert in it just because it's hard to kind of hold on forever to what I believe is kind of a slowly dying workflow modal, which is the manual part of things like data entry and pushing buttons. So I would encourage people to adopt these technologies and learn how to use them. Because unless everyone's wrong and everyone might be, these things aren't just going to disappear overnight.
- [24:41]
B
No, I think we've proven that it's not going to disappear. I think how how much we utilize it is still in question. I think for a solution like yours it makes total sense in other aspects. I think there's going to be other AI implemented solutions in market that may not scale as much. But based on what I'm hearing and from understanding the ADOPTS process to some degree, like I could see how a lot of that work was pulling XMLs, making suggestions, re uploading XMLs, putting tweaks into the ad server to improve performance, all that stuff was so manual and such a pain to do and it wasted so much time only to potentially pivot the wrong column and make an error and find out that you killed your campaign or over delivered or ran on the wrong allow list. I think all this makes sense to me. So Joe, I really appreciate you coming out and being my guest on the pod. I hope I see you at Sushi soon. It would be a word from our sponsors every day. Your digital campaigns are missing nearly 40% of their target audiences due to identity fragmentation. That means that nearly half of your audience is invisible, unmeasured and unmonetized. Reach out to find your 40@adform.com that's the number 40. To see what that missing 40% could mean for your revenue, reach out to us again. That's findyour40adform.com.
- [26:12]
C
My pleasure.
- [26:13]
D
Hey party people, It's Amelia from Architecture. Kick off Advertising Week in style on Monday, October 6th at the Madison Ave Returns Party. Over 300 agencies, brands and leaders all under one roof at Virgin Hotels, New York City. Want your brand front and center? Sponsors and VIPs get priority access. Lock it in now at edtechgod events.com and stick around. The Refresh with Kate is coming coming up next. Hello and welcome back to the Refresh, your weekly download on what went down in advertising. I'm Kate with Marketexture and today is Tuesday, September 16th. This week we're talking about the eyebrow raising mentions in Google's ad tech remedies filing, Magnite's acquisition of streamer AI, and the tune in for YouTube's first live broadcast of the NFL season. Before we kick things off, shout out to our sponsor Adform for their support. 40% of Internet users are unreachable on cookieless browsers due to identity fragmentation, causing wasted impressions, inflated CPAs and revenue loss. Reclaim your audience and find your 40 with Adform. Now let's get into it. Is the Open Web in decline or doing just fine? If you ask Google, it seems to be a little situational. Statements made in their court filing for their AdTech antitrust case revealed two major red flags for the programmatic advertising ecosystem. The first, internal data cited in the filing show that only 11% of display advertising impressions purchased through AdWords were for open web display in January 2020. That's a pretty drastic drop compared to the 40% that were transacted in January 2019, according to Google. The other red flag this statement that expands on these data points. The fact is that today the Open Web is already in rapid decline and plaintiff's divestiture proposal would only accelerate that decline, harming publishers who currently rely on Open Web display advertising revenue. For obvious reasons, these two statements in their filings set off a flurry of commentary or maybe distress across the ad industry, prompting Google to issue a follow up statement and correction that clarified they meant open web display advertising when referring to the open web. And in my opinion this is really just a dance around semantics, since most advertisers still associate the open web with open web display and video. These statements in their filing seem to be an effort to absolve themselves of the severest of remedies in their adtech antitrust trial. Given the previous quote was immediately followed by this statement as the law makes clear, the last thing a court should do is intervene to reshape an industry that is already in the midst of being reshaped by market forces. The irony here, or maybe it's not that ironic at all depending on how you look at it, is that Google is arguably one of the biggest market forces reshaping said industry, and they have been for years Now. Even before OpenAI stormed in with ChatGPT, Google also cited the shift of investments to other channels like CTV and retail media as a source of blame for pulling ad spend away from open web display and negatively impact affecting the channel to that point. Expansion of the digital ad ecosystem into other channels and platforms is a natural evolution that has been gradually happening for a while now. But the open web in particular is seeing much more dramatic shifts, especially when you look at drops in referral traffic as a result of zero click search or search that starts and ends within a search interface like Google's AI overviews. For example, the percentage of zero click news related searches increased from 56% in May of 2024 when AI overviews launched to 69% in May of 20. Even if other platforms and ad tech partners continue to drive advertiser spend to open web channels like a retail media network that leverages off site media would, Google still has an incredible amount of leverage when it comes to directing consumer traffic to publisher websites or not. And at the end of the day, referral traffic is a critical variable in publishers ability to attract ad spend in the first place. Right now we see Google walking a pretty thin line between presenting a search business that's thriving within a healthy ecosystem while also trying to avoid the harshest remedies in their ad tech antitrust key days. On the search side, Google maintains that publisher referral traffic is doing just fine and its AI search products continue to drive billions of clicks per day to publisher websites per Google search head Liz Reid. But on the ad tech side, it's in Google's best interest to make themselves appear weaker or disadvantaged in court positioning. A potential sale of Adex is harmful to the publishers that rely on the technology to run their businesses effectively amid an open web that is vulnerable and in decline. We'll see what Judge Brincoma thinks in a couple of weeks. Moving on to Magnite, the largest independent SSP in the market, has announced that it's acquiring Streamer AI, a generative AI platform that enables production of creative assets for CTV ads. This announcement marks one of the more significant bits of M and A activity we've seen in months. Excluding ongoing activity from major holding companies like Omnicom and IPG and TV conglomerates like Paramount's merger with Skydance, Streamer AI's technology will be offered to Magnite's ecosystem partners, including agencies, retail media networks, publishers who operate buyer marketplaces, and DSPs. The move comes as momentum continues to build behind efforts to bring more small to mid sized businesses or SMBs into the CTV space. And so far this year we've seen some major moves to push this effort forward. Comcast launched Universal Ads and several platforms like Roku Ads Manager have rolled out features or self serve platforms that make it easier for advertisers of all sizes to tap into the CTV opportunity. Advertiser Perceptions reported there were 6.2 million SMBs in the US in 2023, representing a collective $99 billion in annual ad spend. It's no surprise then, that many see SMBs as the next avenue for revenue growth, and Magnite is no exception. Their press release announcing the acquisition of Streamer AI made their ambitions clear. This acquisition is specifically geared towards overcoming one of the primary hurdles that SMBs face when trying to break into CTV creative production. As for delivering this creative to consumers, magnet also has access to some of the most highly sought after streaming platforms, including direct relationships with Netflix, Disney, Roku and Warner Bros. As well as preferred integrations with over 90% of its CTV supply partners. All in all, the CTV opportunity is rapidly expanding. Nielsen reported that 3/4 of TV viewing is ad supported, with streaming, stealing ad supported share from both broadcast and cable. Meanwhile, eMarketer estimates that CTV ad spend will reach about $33 billion in 2025. The chance to increase revenues by opening up CTV access to a larger pool of advertisers is a legitimate opportunity. But there is a catch the measurement conundrum Measurement remains one of the biggest hurdles in the CTV space and the reason that channels like social and retail media still see outsized budget allocations. These channels can more immediately and directly demonstrate Performance growth within the small to mid sized business sector may be difficult to sustain if these advertisers, who are often operating on more restricted budgets, can't consistently identify and communicate performance. Finally, we have YouTube, whose first exclusive NFL broadcast aired Sept. 5 to the tune of 17.3 million global viewers. This includes 16.2 million U.S. viewers and 1.1 million viewers outside of the U.S. the matchup between the Kansas City Chiefs and the Los Angeles Chargers marked the first time a streaming platform has live streamed an NFL game in its entirety for free. To give a bit of a comparison point, Amazon's Thursday Night Football on prime video averaged 13.2 million U.S. viewers per game during the 2024 season. Over on Netflix, the games they hosted on Christmas Day 2024 averaged 30 million to just over 31 million viewers for the two games that aired that day. Last Friday's game served as a freebie to get viewers excited about YouTube's exclusive rights to Sunday Ticket in the U.S. an honor previously held by DirecTV. Viewers can watch Sunday Ticket on YouTube or YouTube TV, but they will have to subscribe. Both plans offer either an 8 month subscription for $34.50 per month or a month to month plan that can be canceled at any time, but it will run you $85 per month. I'm skeptical that subsequent NFL games on YouTube will rack up as many viewers as this free game. YouTube TV had just over 9 million subscribers as of April 2025, but time will tell. Personally, I'm not a huge football fan. I'm more of a hockey gal, to be honest. But I think this NFL season will be fascinating to observe from an advertising perspective given the sheer volume of live sports that have shifted to streaming environments since last season, particularly ESPN's new Direct to consumer streaming platform, which gives subscribers access to all of their linear content as well as access to content that comes as a result of the recently inked deal with the NFL. Alongside this, ESPN subscribers also have the option to bundle their subscription with Fox's Fox one platform. Another major player in the live Sports Mix is NBCUniversal, who has access to a ton of live sports content this year, including the 2026 Winter Olympics, Sunday Night Football and the 2026 FIFA World Cup. Another variable that will be interesting to observe is the ongoing beef that's happening with Nielsen over the accuracy of its TV ratings. The NFL is now involved and has recently been vocal about Nielsen undercounting viewership, particularly co viewership numbers. And earlier this year, just ahead of Cannes, Nielsen sat down with the Video Advertising Bureau to address concerns that its machine learning based household demographic assessment model was inaccurately capturing demographic information leading to wide swings in audience demographic composition. So whether you're a sports fan or not, this season has huge implications for advertisers. And this is definitely not the last you'll be hearing from me on it. That's all for this week. Thanks for joining us for the refresh and we'll catch you next week.