← Marketecture: Get Smart. Fast.
Marketecture: Get Smart. Fast.
Programmatic Innovation in Live Sports with Andrew Casale and Megan Pagliuca at Marketecture Live
00:26:48
Loading summary
Transcript33 lines
- [00:00]
A
Hey everyone, it's Ari here. I want to let you know about our upcoming Market Live conference in New York on March 10th and 11th. Our live events last year were smashing successes with sold out standing room only crowds, amazing speakers and the best content you'll get in any setting in the advertising business. This year we've expanded to two days and over a thousand attendees, so it's the must attend event for the doers and thinkers in our business. You're going to learn something at this event. The speaker lineup has just been announced and it's really strong and we're just getting started. So we announced Sophia Kolushi, the CMO of Molson Coors Neil Vogel, the CEO of People Joanna o', Connell, the Chief Intelligence Officer at Omnicom Jeremiah Oweng, the General Partner at Blitzscaling Ventures, he's an expert in AI and Lance Armstrong, the General Partner at Next Ventures. Get your tickets now. Early bird ends soon, so your tickets are available at market, that's markitecturelive.com and we have special deals for brands, agencies and publishers. While tickets last, so we're going to sell out. So you want to get your tickets. It's a two day event so plan ahead. But it's in New York, nice and easy to get to and we're looking forward to seeing you there. This podcast is brought to you by audiohook, the leading independent audio dsp. Audio Hook has direct publisher integrations into all major podcast and streaming radio platforms, providing 40% more inventory than what could be accessed in omnichannel DSPs. What's more, audiobook has full transcripts on more than 90% of all podcast inventory, enabling it advanced contextual targeting and brand suitability. Audio Hook is so confident that in addition to CPM buys, they offer the industry's only pay for performance option where brands can scale audio and podcasting with peace of mind, knowing they are only paying for outcomes. Visit audiohook.com to learn more. That's audiohook.com.
- [02:04]
B
Right? Hello. How we all doing? Let's make some noise. Yeah. We asked for the room to be extra cold just to keep everybody alert and awake. How are you doing today?
- [02:14]
C
I'm good. I'm excited to be here.
- [02:16]
B
Awesome. We have some exciting topics in the cue card, so bear with us as we get through our logistics. I was just talking to Terry about the World Series, so this is a fun first topic, but everybody right now is talking a whole lot about live sports and streaming television. All these magical moments that are so fleeting and so powerful. That's probably a fun topic to kick us off. So Megan, how is OMG and your clients looking at this opportunity right now? How are you guys looking at live and streaming and kind of making the most of all these moments?
- [02:51]
C
Well, live is an area we are particularly passionate about. So we tend to pick an area where we think innovation needs to happen and invest in it. And so live was actually with both our consumer research and our partner program investment area last year that we brought to market in Cannes. And it continues to be an investment area. So think about bringing the power of kind of live sports, live entertainment, with the power of programmatic, so addressability and accountability. So one area, a great example is some of the work we did with Disney where we actually brought into programmatic decisioning what they call magic moments. But it's essentially like high attention or cultural events. So in a sports game, a free throw or a goal. How do you change personal? How do you essentially personalize in live events? And the first phase of it was actually, and I think this audience likes geeky things, so I'll go into that. The first phase of it was just as simple as understanding when a user first turned on and tuned into the programming. So think of it like user session frequency 1. We just didn't have that visibility that wasn't in programmatic decisioning. And so that was kind of the first phase and then we've scaled it significantly of not only understanding when a user first tunes in, but also how to change that experience based on events. So we started that with Disney, but have been focused on how do we bring those signals into programmatic decisioning across nbcu, across Paramount, across others.
- [04:23]
B
That is super cool. So effectively things that you could just never do before, you can now do. And so now it's just like the imagination is upon us.
- [04:30]
C
Yes. I also think measurements, so for live we'll use kind of cross screen measurement. You have providers out there like an ispot or a video amp. We've started to bring in kind of proprietary signals to be able to understand what happens in live events faster. Right. You have such historic delays of when you're seeing what really happens in the Super Bowl. So that is another area. But it's how do you essentially bring the addressability and accountability into live sports? I think one of the reasons this has been successful is because of the change available in technology, both because of the IAB spec and some of the changes in technology. Could you tell us a little bit about that?
- [05:07]
B
Absolutely. We're going to go back and forth just to Have a little bit of fun. I don't want Megan to have all the fun. So thanks for the tee up. So I like to say a lot that there's all these plumbing things that really matter. I see some tech lab representation here, but that's been really true in live and in streaming. So the first big one was the work that we did with the tech lab in potting and the 2.6 spec. We're three years into that now and it's been really incredible to just go from a market that was taking a pod and firing requests out into the Programmatic pipe to one that speaks time now, where if there's a three minute or five minute pod, we can just send that to a dsp. We can tell them the durations of time that are available. We can massively cut down the request load. These standards have worked. They're hitting great scale. Live has a very unique. These cards are attacking me. Sorry, my apologies. Live has a very unique version of this problem as well though, which is if you, if you kind of go back to that World Series example for a second, imagine, you know, we're, we're, we're finishing the, the first inning, we go to commercial break, we cut to the commercial breaker or the pod. In that exact moment, Programmatic does something really unique. Every single device asks for an ad at the same time. Every LG Smart tv, every Samsung Smart tv, every Roku stick, every Fire stick, they all say, we want an ad. And on one hand, that's super exciting because it makes it very addressable. You can associate it with data, but it also creates these unbelievable surges or spikes. And these spikes are punishing. So much so that they often get throttled or they get errored out or this most, this unbelievable moment in time, this live event that is fleeting and if you lose it, you can't get it back. It's almost like a travesty that we can't always address it. The standard that's being put into motion though, is going to attack this problem. And I believe it might be in public comment or nearing public comment. Yep, yep. So there. I'm not wrong. And so I'd encourage everybody to take a look at it. But the whole premise behind it is why wait for the commercial break to send these requests? Send them in advance, start federating the request out in advance of the commercial break because we know the inning is going to end. So there's never a spike, there's never a loss opportunity, there's never throttling. Um, so that, that's like some of the standards work that's underway that will just make this even more scalable and will get us to a point where it won't matter how big the broadcast is, it won't matter that it's millions or tens of millions of fans. We'll always be able to handle everything. There's a fun metaphor to this problem from linear. Some of you may have heard this one before, but I think in the 70s in the UK they had something called kettle spikes, which was when daytime soaps would go to commercial break, everybody would run to their kitchen and start making tea, and so they would turn on their kettles. And it was such a phenomenal spike that it would actually tax the grid. And so eventually the grid caught up and solved the problem. But they temporarily implemented a solve to offset the broadcast just a little bit so that people would go to their kettles just a little slower to avoid the spike. So we're kind of dealing with a modern version of that right now. Kind of boring, geeky stuff, but I think also demonstrates the real critical importance of standards, which are well underway.
- [08:25]
C
So one of the biggest shifts we're also seeing is in sell side decisioning. So why is that happening now and what has changed to enable this to be such an opportunity?
- [08:36]
B
Sell side decisioning has been a phenomenal event over the last two years. In particular, if you haven't heard the term before, you've probably heard curation, which is really where it all started. And it's really just the premise that the sell side of the transaction can get into the business of adding value, organizing it, curating it, associating it with data, applying decisioning instead of just waiting for the DSP to make the decision. And the reason why we've kind of gone from curation, which was largely just organizing domains, to something far more sophisticated in the last few years is we've recognized the sell side has a pretty cool opportunity to leverage its compute in a way that we haven't before. And the way to think about that is if you go back to the earliest days of programmatic and RTB, you know, 15 years ago or so, we were all high five when we ran an auction in 200, 250 milliseconds. Because at the time it was absolutely technically incredible. It was really hard to do, and it was really impressive. The thing is, computers don't stop advancing. And what used to be hard becomes easy very quickly. And today, 250 milliseconds is actually an eternity. And in a relatively short amount of time, I'm talking 10 milliseconds. We can do just about as much work today as we were able to do 15 years ago with 200. Because of that, we're now in a position to start to apply new value and new capabilities on the sell side before we kick off an auction to a dsp. So imagine now, historically you built a lot of the value in the 200 millisecond window after the bid request. Imagine instead in the 10 milliseconds before the bid request, you add almost the same amount of functional capacity to associate an impression with data, to apply a decision to call a model, to do all kinds of new and novel things. And so we don't really know where this goes now other than we've opened up a new pocket of innovation. And the reason why I think that's really important is it's not necessarily to compete with DSPs or to challenge them. I actually think it's giving a new home to innovative ideas that we're stuck and trapped in what I've often referred to as like roadmap hell, where the market has built itself largely around the dsp. If the DSP doesn't prioritize a problem, and more often than not, they don't typically prioritize problems on the sell side as the top of the list. They just got stuck in this like limbo abyss. And I think that's where a lot of the early innovation started. But now we're applying it to a whole host of new capabilities that we just hadn't previously thought were possible. So it's kind of like an interesting, exciting emerging time relatively early into this one.
- [11:14]
C
So it's been a long time that the best practice in the industry has been consolidate as much of a buy as possible through a single DSP, only add other DSPs when necessary for unique data, for unique inventory. We're also at a time where you see a lot of the SSPs, not you guys, but others kind of offering buy side interfaces and then see kind of the trade desk with open path, kind of direct to publishers. So with the changes, where do you see the ecosystem going? How do you see that evolving?
- [11:45]
B
Yeah, we are in a pretty wild time of change. We haven't really seen a chapter like this since either the original emergence of real time bidding and programmatic as sort of the contrast to the ad networks at the time, or at least for the sell side, specifically the emergence of header bidding and what it did to the traditional waterfall. I would say that this event is on the same level of that and that's why you're seeing some of these strategic changes that are taking place. The way I would frame it is there's two large camps and there's certainly many others that are potentially in the middle. One of them is the premise of saying, why do we have a buy side bifurcated from the sell side? One side of the market can do the same work as the other. So you don't need a DSP if you're, if, if you're an ssp, you could be a dsp. If you're a dsp, you don't need an ssp. You can be, you know, the sell side too. Like, you can kind of have each side of the platforms do everything. The challenge that I see with that long term is we rip the market in half for a reason. It becomes really hard to be an agent of the buyer and the seller simultaneously. Eventually conflict creeps in, but, but perhaps this time around it will be different. So I would largely characterize it as like, one strategy that's in motion right now. Have platforms all do everything. The other one that's in motion that we are certainly advocating for and I think is gaining a lot of ground right now is the idea of unbundling all of the value creation, unbundling capabilities, modularizing them so that you kind of remove the whole notion of power from every platform and you put platforms on a path to just extreme commoditization to effectively say, like, why do the modern equivalents in programmatic of like the credit card terminal take 20 points on a transaction? It may have made sense 15 years ago when this market was immature, lack standards. And we were just trying to figure this out. But we're at a point now where this is not new technology and these are not new protocols. One of the reasons for it is at least if you look at a modern DSP today, it kind of does everything. It associates the impression to data, it runs the algorithm, it finds performance, it does attribution, and it applies the creative. It organizes supply. You name it, it does it well in an, in an unbundled world where all that value is modularized, you're left with the bidder that is effectively going to come closer to commodity, much like the sell side. The sell side has been on the path to commoditization for, quite frankly, far longer. And I think if anything, we might start to see the DSP look a lot closer to that on the, on the other side of this trend. So I guess to summarize it, we don't really know where this is going to go from here. We are at a fork in the road. It's either one platform does everything and solves every problem, or we massively unbundle all value, modularize it, and then we make it so swappable. You can replace any piece in that chain quite seamlessly and you can also drive each piece of that chain to a far more efficient business model. We'll see which one ends or wins in the end. I'm not biasing it at all, but it'll obviously be the second vision. But maybe I'm wrong.
- [14:50]
C
It's fine with this evolution. I ask this question, which is a little loaded of a question and I think it's good to balance it right. You'll consistently hear the question of should there's not enough dollars go into working media? I think when you ask that question, you have to remember that if you buy on a direct IO basis and consolidate Programmatic, it always works better. So where there is significantly more fees in Programmatic, you actually do see significantly more value. So with the changes you're talking about, what do you think the impact is to working media?
- [15:25]
B
I think like the, the ultimate loser in the end of this trend is ad tech and margins for ad tech platforms. And the winner is the marketer and their working media going farther and the media company compressing the value chain in the middle. If you take either fork, just to be very objective, if the strategy of put it all into one platform ends up being the winning strategy, you should conceivably find an economy of scale come out of that. That should lead to more money, goes through the mousetrap and gets to the other side, the publisher's wallet, which is a win for the market, it's a win for the edge. Same is true on the other side. If we get to a place where we continue to commoditize the SSP as we have been for over a decade, we begin to commoditize what we consider to be a DSP or the bidder. And we allow all value to be modularized and to be very easily swapped in and out to find the most efficient, you know, piece of value. You should also get to a place where the dollar from left to right is far more efficient. So I think whatever, whatever version of the market that we see next being the dominant version, we are going to see efficiency continue to rise and we are going to see more ad spend get to working media. Well, maybe to turn it around on you, Megan, OMG was among the first to embrace sell side decisioning. At least from what we've been doing here at Index. And so just curious, from your vantage point, how are you rethink either your strategies or your customer strategies to get to even more effective and efficient outcomes.
- [17:00]
C
So we have for many years we've always cut working media fees for our clients. So we have a small number of DSPs where we have best market rates. We then have the same thing on the SSP side. So working with a kind of limited number, ensuring we have the most efficiency for our clients and then also a huge focus on kind of supply path path optimization and inclusion list only approach. So I think that has been in place for some time. I think what's changed is that more capability in terms of supply side data signals and optimizations.
- [17:36]
B
Right.
- [17:36]
C
Curation seats haven't been around for that long. Also some of the evolution of AI that allows us to actually bring way more data signals into these algorithms and also then deploy the algorithms on the supply side. So I think where we've provided some of that efficiency for some time, there's so much more we're doing now because of the advancements in technology.
- [17:59]
B
I think it's a really important point and I think it's something that might not always be really well understood, which is one of the values of something like an algorithm or a model on the sell side is the sell side doesn't really have the option to throttle its access to supply. When we or any of my peers on the sell side plug into a publisher, whether it be a paramount in streaming or a weathering app or Yahoo on the web or whomever, we see all their scale, we see all their supply. And when you point that at a model, especially in models that are underpinned by neural nets, they're thirsty for data and scale. The way to make a neural network better is A give it really reliable data, but B give it more data, give it all the data. And so it almost supercharges the ability for those models to do next level things by moving it away from an optimization process. Because typically when we're interfacing with a dsp we have to constrain that scale in the form of QPS caps, which reduces their view of the Internet and I'd argue over time reduces the ability for these models to become next level in their, in their ability to output value. But with all that said, I'm. I'm a disaster here. I think we can move on. I'm never using cards again. This is awful. Apologies for that. Stick to computers, Andrew. But why don't we Pivot to another question because we're already running out of time. AI, AI. AI. What's. What's your AI right now within OMG or with what you're doing with your clients?
- [19:39]
C
So I think there's a lot of. Well, first, obviously there's a lot of talk about AI, but it's like what is actually happening, what is real, what has changed. So you did already touch on the changes within Algos and essentially the portability and then more data within them. So we see that more powerful, particularly on the supply side, but also inventory curation. I'm going to talk through a couple use cases that are very real and have been very impactful with AI already. So one example in our inclusion lists, we historically would look at and have humans manually review based on consumer experience, metrics based on. Then we'd have data just on fraud, on kind of table stakes, things we want to exclude, but it would be a human that's following that framework. Now we've created thresholds and we have kind of gen AI scanning the Internet, whether it's domains, apps or CTV to ensure those can be part of our inclusion list. So feel so much or that's a huge improvement in terms of also the frequency we can update that versus what humans could do and just the objectivity around that. So I think that's one piece. The other very impactful piece is around catastrophic event targeting. So, you know, has historically been a challenge and something we've been talking about for like 15 years when we talk about semantic.
- [21:01]
B
Right.
- [21:02]
C
But it just, the technology wasn't able to do that. So if a brand doesn't want to be next to. There's a hurricane, doesn't want to be next to hurricane news. We may have blocked entire news outlets. Now we can more semantically understand is this that hurricane or is this the Miami Hurricanes or is this the drink? The hurricane. Right. So like, you know, we talked about that for many years, but the capability wasn't there. So I think there's like a lot of concrete, specific examples that we're using it today.
- [21:37]
B
That's exciting. Did you have a question for me by chance?
- [21:40]
C
Yes.
- [21:43]
B
We'Re doing our best here.
- [21:45]
C
I chose my phone instead of the card a little better shape than you. So where are you seeing meaningful applications beyond the hype? Right?
- [21:57]
B
Yeah. And so I would say that there's a lot of hype, but I'm happy to comment on that as well. But I think right now we are seeing some really exciting applications of AI directly where appropriate within the business. We've touched on it a little bit. But when we're deploying models at scale, at the edge of the impression, whether that be with some of the work that we're doing with you guys, or whether it be through some of the work we're doing with a whole lot of new upstarts, whether it be companies like Chalice or Empowered or Cybids or Fenestra, a lot of them are building neural nets and those neural nets are really what underpin the LLMs that we see today day. They're very hungry for data and they're doing things that frankly humans can't always understand. They get to the outcome in the end. That's highly performant. They're trained on massive amounts of data, but how they achieve it is sort of otherworldly. But that's like a very direct application that we're excited by because it feels like the next version of the performance algorithm. It feels like our response to pmax or Advantage plus as well. For the open Internet in the creative space, we're also seeing a ton of really exciting work in Genai and specifically applying it to reducing barriers to entry to get new advertisers, particularly in things like television and streaming. Programmatica has done an incredible thing at democratizing access to tv. It used to be priced out, especially if you were a local advertiser. There's never been more options for you to get into the channel and also do it at scale at the same time. You can do it with creative that's affordable as well because the production costs of TV creatives are also huge inhibitors. Gen is doing phenomenal things where you can give a URL to a platform and outspits a pretty impressive creative relatively quickly. I think I see a representative from Open Ads here as well. There's also some really cool new DSPs emerging that are focused on applying creative in real time to the pages context as well. To literally tell a story while you're on a page or an app that fuses with the brand. Those are great applications of it. We see applications of it as well in the workflow arena. You know, as much as we want to talk about agentic and the agentic future, there's a bunch of boring problems that we haven't really solved yet. Like we still today. I see Janet nodding her head. We still today deal with discrepancies like literally on a daily basis. They still exist. We want to ignore them, but they're there. It would be really great to apply AI to solve a problem like that that no human wants to deal with. Deals are still a sore spot for troubleshooting, which is application to bring a bunch of data to bear to find the broken piece so the human doesn't need to. And so I'd argue there's a bunch of workflow applications that we're seeing. They're getting excited to find like the broken bit faster than the human would have, but they're all in the realm of augmentation versus something revolutionary. I'm sure something revolutionary is coming. I'm just not so sure it's here yet with proof.
- [25:00]
C
Yeah, it's your turn.
- [25:06]
B
I'll never be asked back here. Looking ahead, this is our closing question, so we're almost through it. What's one area of programmatic innovation that you think will matter the most looking ahead to 2026?
- [25:19]
C
Well, I think also kind of looking at the word more broadly, I think influencer is an area we're seeing huge innovation. We've taken a lot of the principles of programming and honestly, Influencer used to be very PR driven. You decide influencers you work with based on the number of followers. We're doing things now where we're having more programmatic decisioning around it. We're essentially looking at our audience and then deciding influencers based on the followers match to that audience. I think that's a big area. The other big area of innovation has to be around like Generative Engine optimization, which now is Geo. So no more SEO. Geo is, is the new language. But that has to innovate this year because it's such a challenge for brands. We're not. The consumers aren't going to the web, their websites as much. They're going to chat GPT. And then how are they. How they're using Rufus when selling products on Amazon and how are they using Gemini? So you have some, I'd say like immature startups in this area, which we're kind of like partnering with everyone. But there has to be more innovation in that space because it's such a change in consumer behavior. Behavior.
- [26:29]
B
Well, I don't have a lot of time, so I'm just going to pretend like you asked me the same question and respond by saying for 2026, I think it's going to be a continued acceleration to efficiency. It's like frankly, the biggest thing that I'd say is happening right now is I've never seen so much change in the air. The air. The air strategies were set for the last decade and people were just consolidating around a few. But we're really seeing expand out. We're seeing people test in new ways, new platforms, new capabilities. It's exciting. I think it's in service of a more efficient future, and I expect a lot more of that next year.
- [27:09]
C
Yeah.
- [27:09]
B
Thank you all for bearing with us, and particularly this. Thank you, Megha.
- [27:26]
A
Thank you for listening to the marketecture podcast. New episodes come out every Friday and an insightful vendor interview is published each Monday. You can subscribe to our library of hundreds of executive interviews at marketecture tv. You can also sign up for free for our weekly newsletter with my original strategic insights on the week's news at News Market tv. And if you're feeling social, we operate a vibrant Slack community that you can apply to join@adtechgod.com.