Loading summary
Transcript39 lines
- [00:05]
A
Welcome to the Big Story, a roundtable featuring members of the Ad Exchanger editorial team. Every week we bring you an in depth discussion of key developments in digital marketing and media. What exactly are you buying when you purchase inventory? Programmatically, we have two very different stories today on the podcast that both address this age old question because although programmatic turned the conversation about media from content to audience, context and content still matters. And brands want to be in media environments that perform better. Because as both of these stories show, some content performs well and other places not so much. First up, we're going to hear from our Associate editor, Victoria McNally about the rise of AI slop. Marketers and agencies have long tried to avoid buying against clients Clickbaity slideshows, the kind of content that's purely designed to trap you long enough to serve you as many ads as can be jammed on a single page. But now there's a new foe to contend with, which is AI slop. With a prompt or two, it's possible to churn out reams of low quality content and marketers who are buying the most of this AI slop MFA have the worst media performance. It's a difference between maybe buying 1% MFA and 2% MFA and the best performing marketers are under 1% and the worst performing marketers are over 2%. So small amounts matter. Here she is going to dig in to help all of our listeners better understand this problem and how to steer clear of AI slop. Then in our second segment, for as long as I've been covering Connected tv, I've heard buyers complain about the lack of show level transparency. No one knows what performs better, the nightly news or Love Island. Seriously, the level of granularity that buyers have is often frustratingly general and it hampers their ability to optimize spend. So what to do? Well, one buyer cobbled together a unique solution using multiple ad tech tools in order to figure out where they were buying ads. And here's what happened. Performance improved. Our Senior Editor Anthony Vargas is going to walk us through exactly how you can hack your way into show level transparency. So on this week's pod, we'll talk about how better understanding of your content, whether it's on CTV or hopefully not on aislop, translates to performance. And before we dive into this show, I'm your host Sarah Sluice and I want to talk about programmatic IO New York City, which is where I will be on September 28th and 29th this year. This conference is the established place to talk about disruption. We've seen it before. Ad tech upending the media ecosystem. And this year you can attend to surf the wave of all this new tech and not get sucked under the surge. And as always, our listeners can use the code POD10 to shave an extra 10% off your ticket price. So lock in your plans for September 28th and 29th. We'll see you there. So, Victoria, let's talk about AI slop. Is this the new MFA?
- [03:18]
B
Well, before I start, I do want to point out how funny it is that Anthony and I essentially swapped beats for the week, which is fun. It's like when comedy writers write horror movies. But yeah. So I have to also thank Scott Pierce, who's now at the trade desk, was previously at the IAS when he ran a column with that headline is, you know, AI Slop is the new map of mfa and we need to fight it. I think is something to that extent is what the headline was. But yeah, in regards to that question, not really. But that doesn't mean it's not worth focusing on and being vigilant towards. So to backtrack a little bit, some of the numbers that you mentioned up top, all the numbers actually come from the most recent ANA Quarterly Ad Spend benchmark report. They've been doing it since 2023. And since then, the amount of ad spend, the percentage of ad spend that's been relegated towards mfa, has been pretty much steadily going down as the industry has taken steps to sort of remove it from bitstreams and remove it from their products and, you know, just try to do away with it. And it's mostly successful just to jump in there.
- [04:30]
C
I think that, like, first report found that wasn't it like 21% of, like, ad spend was going to MFA sites. And it was like $13 billion.
- [04:37]
B
Like, it was between, I think, 15 and 21. I should have it written down here. And I don't apologies, but it was. It was a lot. It was a huge amount, you know, and millions and millions of dollars to your point. And so now it's closer to like 1.1% in the most recent report, but it's also technically higher because the previous quarter or the previous year, I think it hovered between 0.4 and 0.6. So in some respects it technically doubled, but it also doubled from like half to one. So still not something to like, really freak out about just yet, which we can get into in a little bit, but something to be aware of and sort of. There's always this kind of constant vigilance. You need to have with bad faith actors, you know, it's like playing whack a mole. You get too complacent, they come up with a new way to get around your stuff and then you have to pivot from there. It's hard not to be reactive in this kind of a situation. And so the ana, when they noted this, suggested in their write up of the report that like, you know, obviously advertisers have to be aware and vigilant, but then also, like, it's possible that the rise of AI Slop is contributing to this rise. So that was what I investigated and what I found is there is some overlap between MFA and AI Slop. You know, it's easy to use AI tools to create things that are typically defined as mfa, but it's not quite a. Like, it's almost more of a square and rectangle situation where AI SLAPP can be mfa, but it's not always automatically mfa, because the proper definition of MFA requires low quality content, really high ad load and paid traffic. And you need all three for most experts to really say this is mfa. So AI slob doesn't always fit those criteria. Obviously it's like very low quality. That's the one that it fits pretty comfortably into, at least as far as slop goes. We can have a debate about whether AI generated content is slop, but as far as ad load, sometimes it is lower, sometimes it's higher, sometimes it's lower on the homepage, and then once you get into the site, it gets worse. And then the paid traffic is also a big part of it that it doesn't always have. So it's still worth, you know, investigating. But it's also, it's not quite the same.
- [07:00]
C
Yeah, and like, just to kind of.
- [07:02]
A
Oh, go ahead.
- [07:02]
C
Sorry, I was just gonna say just to jump in there just because like, I've covered like a bunch of like the attempts by the industry to define like MFA and all those different things and like how they kind of like landed on those three main criteria. That was like the product of like a lot of like, you know, back and forth with the ANA and like a lot of like, you know, other like vendors in the space and stuff like that. And I think the main reason why, like, you know, like, like you had a lot of sources saying like, it's not really like a one to one correlation between MFA and AI Slop is because as a result of those conversations and like fears that like, you know, legit publishers were getting caught up in the MFA dragnet unfairly, the industry really decided to, like, focus mainly on, like, is this mostly coming from, like, paid traffic? And that's why they kind of changes from, like, made for advertising to made for arbitrage, because it's more about, like, that arbitrage model of, like, can we make more in ad revenue and than we paid to acquire the traffic? And so, like, you know, when it comes to, like, AI slop, like, a lot of times, like, they're not really relying on, like, paid traffic sources, although sometimes, like, the. The content lives on, like, some of those platforms that are major sources of paid traffic, like social media platforms like Facebook and Instagram. So, you know, it's, it's, it's. I think it's just one of those things where, like, since AI slop is, like, relatively new, the industry just has to, like, nail down its own definition for that. And so, like, things aren't going to necessarily fit perfectly.
- [08:23]
A
I'm going to. I'm going to muddy the waters a bit, Victoria, because one thing I didn't notice in this story and kind of in what the ANA is talking about is how much AI slop exists on social media. And I feel like these platforms shouldn't get a free pass because people's ads are occurring between this AI slop. And one thing I've noticed, much to my frustration, is often my dwell time or my engagement level is sometimes higher on AI slot because I'm like, what is this? This is clearly a fake video of a whale attacking a ship or something, literally, just to name one example. Or, you know, and then there's, I mean, there's also a whole, whole realm of like, for example, like, fake gender reveal videos or like, fake skits. Like, there's a lot of, like, kind of fake, Fake outrage, fake texts between people that you're like, this is outrageous. That could be, you know, AI generated or not augmented. But people have kind of realized that this fake. They can make fake content that performs well. That's often imitating, like, you know, an amazing, you know, whale encounter that did actually happen. And then there's like, fake versions of that on top of that. So I didn't see that address. And I'm just going to say that I feel like this is something that is going to become more of an issue and something people are more aware about.
- [09:58]
B
Absolutely. And yeah, I agree. The fake whale attacks, that's so unfair to all the hardworking orcas out there that are actually capsizing. Billionaire. Yes. They're doing a lot of great work. No, I'm kidding. Yeah. I mean, because we you know, I sort of went down this rabbit hole because of the ANA's report. The focus was very much on the type of sites that typically constitute mfa, which is not something you really hear about in relation to social media. But I do agree it's obviously a big problem. It's big enough that a lot of these companies, while they're developing their AI video tools, are also talking about trying to cut down on slop. You know, YouTube spoken out about it. Meta, I think, has also removed a bunch of spammy content. I think X just changed their video editor to try to convince people to like, repost, rather like through, you know, the app itself, rather than like steal somebody's content whole cloth. So, like slop and spam and regurgitated content are all things that plague a lot of these platforms for sure. And interestingly, I couldn't fit this into the article that I wrote, but when I was talking to Scott Pierce in particular, he told me that one of the things that he saw as a turning point for clients starting to ask more questions about AI slop was a Last Week Tonight with John Oliver segment where he talked about all the shrimp Jesuses that you see on Facebook. And his focus in that segment was very much on the social media angle of, yeah, you've got all this weird fake art of wood carvings of lumberjacks or whatever, just whatever's going on over there. And it's such a big thing on social media especially. And none of those companies are doing enough, it feels like. So that's definitely something I want to keep thinking through and reporting on in the future. But yeah, for, for Now, I mean, MFA, MFAs and low quality sites have been around a lot longer than we've had definitions for it. Obviously. You know, like, the thing with AI is it just makes it faster and quicker and in some ways that makes it actually harder for them to monetize because they're spinning it up so quickly and then abandoning it so quickly that they might not, they might not have time to sort of build the same kind of infrastructure that a lot of those major MFA sites did, which is the silver lining, I guess, but. And I wouldn't be surprised if those people all move to TikTok and start doing Fruit Love Island.
- [12:25]
C
Well, yeah, and to that, and to that point, like, I'll also take this opportunity to shout out another recent guest column that we had on this topic. Rich Radden, the CEO of Zephyr, he. He wrote a column titled AI is Bringing MFA to Social Media. Here's how advertisers can avoid it. And like what Rich was kind of saying was exactly like to your point Victoria around like one of the big concerns around MFA for the longest time is like you know, the ease with which like these kind of sites can be like created. So it's like if you can just like get you know, and like how AI kind of like enables that. So if you can just like scrape legitimate sites, steal their content, spin up like an imposter site that either resembles the original site or even doesn't just takes its content and then just like overload it with ads. Like that's kind of the old like MFA website model. But then, and you know, like I, it makes it easier to make hundreds and hundreds and thousands of these kinds of copycat sites or like content stealing sites or aggregation sites. And what Rich's point was like in a similar fashion to like how like you know, you look at like these like video platforms like TikTok and increasingly like Instagram is becoming like a video first platform you're seeing like, you know, when you think about like we all kind of know this from like having experimented with filming our own social media content, that stuff is actually more time consuming to produce than you might expect. But like because of like you know, new AI video generation tools, Rich was saying like now like people can like spin like can generate like thousands of videos a day and like put them out onto these platforms and you know, get like their pennies from the share of whatever like revenue. Like you know, these platforms are like doling out but like the more content you blast out there, I guess the more impressions you get, the more revenue you get. So it's just like it's such an obvious playbook for how this can just be like applied on the video and too like it had been already on the website.
- [14:12]
A
And when I think of mfa, I often think of like are, are there interests that are aligned to prevent it or are the interests at odds with each other? And it's often quite complex and that the people creating the AI slop benefit from creating more slop. The SSPs or the you know, TikTok Facebook of the world have kind of a short term benefit as well because they're sharing in that revenue but more of a long term harm in terms of trust in the platform and long term spend. And then I think marketers are really pretty much never winning unless perhaps the you know, that really low cost CPM ends up converting. But I think as we've seen the low quality content is, does not perform as well yeah, that's.
- [15:03]
B
That's really true. And I think part of what's happened over the past couple of years is the industry is really aligned on those sort of complex factors that would make someone actively try to seek out mfa. It's sort of the industry sort of thinking more about performance and thinking more about quality overall. And part of that is because the ANA sort of highlighted how much MFA spend actually gets wasted. Part of that is just people are thinking more about how to do more with less rather than thinking about cheap reach. Obviously that still happens. But, you know, if the idea is you're trying to get as much bang for your buck and you're also trying to perform at the same time, then, like, reach doesn't really do anything. You need to be able to prove outcomes. The magic word that we all love to say over and over again. And I think that that trend has been overall good for the industry. Whether or not it's actually effective is a different thing that depends on measurement and depends on where the money actually goes. But I do think it is interesting that when I talked to a couple of SSPs about what's the difference between MFA and AI slope and I guess the midpoint, which is MFA slop, they all kind of said like, well, we don't necessarily. We're not as precious about the distinction because it's all low quality. So it all has to go.
- [16:23]
C
Yeah, yeah. And I think like, you know, on the SSP side, when that first ANA report came out in 2023, it put it like, you know, there was basically like this like big anti MFA crusade that was kind of kicked off at the time. And it put a ton of pressure on the SSPs because as everybody was kind of like pointing fingers, I'm like, why is this, like, stuff even included in these programmatic deals if everybody agrees that it's drunk and it doesn' drive performance? And the reason why it doesn't drive performance is because it's usually overloaded with ads. So, like, your ad is competing with, you know, dozens of other ads on the page and, you know, it's usually low engagement on the end users part. They're just clicking that link to see, like, how gross like the celebrity looks like today that used to be hot back in the day. And they're just, you know, clicking through the slideshow to get to that particular one. They're not really paying attention to what's going on. So like, you know, it's all this stuff about like, oh, yeah, this stuff doesn't really, like, perform, but Then it's like, oh, yeah, like, you know who's actually like, selling it? Oh, it's the SSPs that are including it as cheap reach for like, you know, audience extension for campaigns or various different reasons. So it put a ton of pressure on the SSPs to kind of clean up their supply. And as a result, that's why, like, the SSPs are probably taking a more proactive stance with the, the AI slop stuff too.
- [17:34]
B
So just as an aside, you know what I miss about the pre AI slop? You know, you, when you would get those websites that like looked like they'd been filtered through eight different language translators and then back into English and you had no idea what was being said, it would be like, does Bruno Mars is gay? I miss that. That is the one thing I'll say about AI slop today. It's like, oh, it's, it's a little too clean.
- [17:56]
C
Yeah, not sloppy enough.
- [17:58]
A
Too bland. Generic, Recycled.
- [18:01]
C
And then I would also say just like, when it comes to. Because like, this is another thing that like, occurred to me when we were talking about like the transition of like this, like MFA and like Playbook through like AI slop into social media. Because, like some of the big concerns around performance that you brought up, Sarah, were like, having to do with like ad overload on those MFA sites. As this stuff, as this Playbook kind of transitions some more to like these vertical video platforms, it's harder to overload those like videos with a ton of different ads because, like, the platforms will only let you know so many like pre roll or mid roll ads, like into there. It's not like the actual like user can be like, oh, I want like, you know, X amount of ads on it like you can when you design your own MFA site. So, you know, maybe it's going to be the case that like the performance thing isn't as bad when it comes to like slop video because, like, you know, anecdotally, like a lot of people who actually like really like these like slot videos, they like recognize that it's like AI and they're like, I like it anyway. Like, I know a couple of people who are like that and like, maybe they're engaging with the ads they see in this stuff to the same degree that they would with like more premium content because they like the content.
- [19:07]
A
And I think, you know, marketing is becoming more and more creative too. I've even seen on TikTok some of these. I think it was right, right now I'm getting served a lot of like, Text exchanges, like outrageous text exchanges that are clearly, now that I'm consuming more of them, I'm getting more of the faked ones. And then there was one account doing this and I look and it's actually like I can't even remember what it was selling some kind of like energy drink or supplement or something because they kind of like, you know, 10 slides in, they start mentioning it or something. I'm like, okay. So they're actually like grabbing attention in kind of a weird way and then hawking their products like in a low key way. I'm like, this is, you know, at some point this will be like the Dr. Oz, acai berry, Oprah thing like that that we all can instantly recognize. But for now the novelty is actually working. So there's a lot of banana stuff out there. And I'll just also give like a shout out, like our coverage of like clipping, which was also new to me. You gotta be following our daily newsletter because our team is all over that and it's been a, that's been fascinating to watch as well. So a lot more to come here. It seems like the industry is kind of like recognizing the threat early, but it's still an issue and we know AI is kind of very fast at morphing. So I do not think this will be the last time we cover this. We're going to take a quick break and then talk about CTV show level transparency. And we're back. So Anthony, who normally covers MFA and not CTV show level transparency, took the reins on this story, which is about how one buyer figured out how to get that show level transparency. And then when they actually decisioned on it, they had better results. So pull open the hood, tell us how everyone can design their own show level transparency solution.
- [21:14]
C
Yeah. So basically like what we're going to be talking about here relies on contextual ID integrations with CTV publishers. So Pier 39, they're a contextual advertising platform. They have like a bunch of like these like ID integrations with various different CTV partners. And through those integrations, Pier 39 can create these kind of like anonymized content IDs that correspond to whatever like show it was that a person was watching when they were exposed to an ad for a given campaign. And then the publisher can pass that anonymized like content ID back through its SSP without exposing what the individual viewers are watching. And that's like a big piece of this because it helps avoid VPPA privacy concerns. And then the DSP in this instance was Pontiac Intelligence they're able to ingest that Pier 39 content ID as a signal within their DSP. Then Pontiac just does like, the usual, like, performance measurement measurement thing of like placing a tracking pixel on the brand's website to see if people who are exposed to the ad then end up taking some kind of action after they visit that site. And then Pontiac can use that kind of data, that data to calculate the cost per acquisition when running ads on different individual shows. And then they can also break it down at like, the network or genre level. And then the advertiser, who in this case was the agency, I should say, who in this case was intermediate advertising, they can then take all that information or look at it in the dashboard, like in their dsp, and decide how best to optimize the campaign mid flight or like, you know, future subsequent campaigns. And it's also important to note that all of this is happening through an open auction. So it's not a PMP or direct deal where it'd be easier to get this kind of, like, specific, like, data back from publishers. Arguably, it's all just happening through open auction through, like, programmatic pipes that, you know, everybody's using for all kinds of different things. So, you know, it's kind of a. It's, it's a way that the publisher still needs to be involved and still needs to, like, proactively send the data, the, the ID back. But it's doing it in a way that, like, you know, I guess the publisher feels a little bit more comfortable about sharing that data.
- [23:24]
A
So the vppa, the Video Privacy Protection act, also known as the Blockbuster Law, it's basically saying, like, it could be embarrassing, you know, to me if someone finds out what I've been watching. So therefore it's, it's private. It feels that's something that I think a lot of buyers feel is kind of a smokescreen, a privacy smokescreen for publishers who are programmers who really don't want to share because of channel conflict. Like, they don't want to pass that information if it's being sold through a partner or they want to use it as leverage to get someone to buy directly. So what's your take on this and what does this mean for this whole kind of channel conflict part of the debate?
- [24:09]
C
Yeah, so before I touch on the channel conflict part, like, like, just like more generally, like, I guess, like, you know, publisher concerns around, like, providing this, like, level of transparency. One of the big ones you always heard was like, you know, they don't like the idea of, like, advertisers cherry picking, you know, different content where they want their ads to run which is also like, you know, ties into like why they won't let you know, show level targeting happen also. So there's like a lot of like concerns that it's like oh if like average, if like buyers are like able to like reverse engineer, okay, this content performs better than that content, then a lot of like that backlog content that they have that like maybe isn't as like trendy or like you know, isn't as new is going to be like monetized less and like all the demand is going to flow to like whatever is like new and trendy. That's like really driving a lot of engagement. So that was like one big publisher side concern around not sharing this data with buyers. And I see you nodding your head Victoria. So I think I'm on the right track.
- [25:03]
B
It sounds very reminiscent to sort of the way that people talk about how reality TV became really popular because it's cheap to make and advertisers like it and viewers like it and so then it ends up dominating the entire schedule and you can't do as much like high level, like you know, weird dramas for 10 people kind of stuff which obviously TV shows, they want to be as popular as possible but like it makes it harder for those mid range or like smaller productions to even get made in the first place I guess.
- [25:34]
C
And then also. Well yeah, go ahead.
- [25:37]
A
Well, I just want to comment on that, that piece because I guess in my mind, you know, I guess for years even pre programmatic there'd be kind of like a run of network like that was the, the thing that would get you the 3am Judge Judy or whatever like that type of stuff. But the CPMs were always lower on that than prime time. And maybe there's like hidden pockets that like some 2am show performs really well. Like so I just, I just wonder, you know, I, I just wonder about the validity of the cherry picking argument or if it's more like a fear of change or like kind of the devil, you know, versus the devil you don't. Because I do feel like it could be good. Like maybe pricing recalibrates but it feels like it would have to end up at the same place. Maybe that's me being naive. So.
- [26:26]
C
Well like I'll, I'll get, I'll offer like maybe like an anecdotal example here. So like you know, like because like a lot of this information has to be obtained directly from like the publisher or the, you know, the, the the broadcast rights owner or whatever it is in a lot of situations. David Nuremberg at Intermedia Advertising, he was telling me that like the best case scenario that you're going to get when a publisher is providing this is like maybe 10% to 15% of the impressions for a campaign or are going to have some kind of like show level data or even genre level data attached to it. And so like you know, as a result of that, even when like publishers are self reporting stuff, they sometimes like kind of like hide like what the inventory actually is in the self reporting. And Mario Diaz, the CEO at Pier 39, he was telling me like in Q4, Pier 39 was noticing a lot of demand flowing towards like this, this content that was being labeled by the publisher as documentary programming. And then they dug into it a little bit more and it turned out that what was actually, what it actually was was the Yule log. And like, because like so much demand was like the pricing was like going up and up and up. And so like all these advertisers are competing for this inventory. They didn't really know what it was. Turns out they're competing to bid on the Yule log as if it was like. And like the CPMs for it were like comparable to like some like live sports stuff you're saying. So like you can see how like some of this like you know, muddiness can benefit publishers when they're trying to sell something that maybe isn't super valuable. But that being said, you know, you can make the case that like people are having that Yule log on all day, you know, in their living rooms,
- [27:55]
A
like they're staring at the Yule log. I mean.
- [27:59]
C
Yeah, but on the pricing piece though, another concern going back to like your, your point earlier about like, you know, the carriage deals that like resellers and like the vm forgetting the. I had it written down, there it
- [28:14]
B
is
- [28:16]
C
when they're like, you know, they have the rights to sell ad inventory on behalf of like other broadcasters. There could be some concerns about like those resellers undercutting direct deals and like especially like, you know, if it's like done programmatically, it's like maybe this is like, you know, I'm charging like more of a premium when it's like a PMP or a direct deal than I'm getting for like the backfill inventory when it's sold like programmatically. So there's like also like some concerns from the publisher side about like are we letting too much pricing information into the bid stream? That can then harm our direct sales efforts.
- [28:48]
A
So I'm really into the Yule log versus live sports example. Like what are some other examples of what advertisers are doing who have this information through this integration?
- [28:59]
C
Well I think like another thing that like I've always heard that like buyers really hate is just like when platforms send back just like unknown like values for like content. And then it's like because especially because like there's so much like off platform audience extension going on. So like, like even if like you're doing a direct deal and like you know, with like a reput suitable publisher and you get back a lot of stuff that's like unknown, there's always concerns that it's like is this just like junk like is this just like stuff that's like never going to perform for me that I'm just like, you know, it's like the pitch is like oh well it's cheap reach and it's like bringing down your overall CPMs and keeping your, you know, your budget like allocations in check. But like is it ever going to perform? Is any like body who sees any of these impressions going to actually buy the brand's product or you know, sign up on their website or something like that. So like whenever there's like just like those complete black boxy aspects to this where like no visibility whatsoever, that was like always the major concern. I think like the stuff of like publishers willfully like obfuscating what content is like that's also a concern but maybe less pressing than like just like the I don't even know what this is
- [30:03]
A
concern which, which on linear you do get a report after I believe of where your content ran. Even if it's the run of network and you can, you know, and people see, they see the 1am spikes to their, their website after a TV runs. I mean TV should give itself a, a little more credit. So one thing I'm wondering about too is like so Pier 39 has to have their information on the website in order for advertisers to do this? Like do. Or do you think there's publishers who are like jk, you can't put your tag on here anymore. Or do you think that it's in their best interest to share this information?
- [30:45]
C
Well, so Pier 39's pitch around this is like by like giving buyers this kind of transparency, it's actually like beneficial for public for like especially premium publishers who like maybe their inventory is like being like devalued in the marketplace because like you know, there's so much, so many questions about what buyers are actually buying. So they're more incentivized to like try and keep those CPMs as low as possible. But like, you know, if, if there's like more transparency around like proving that the content actually performs and drives like actual business outcomes for the, for the brand, then maybe agencies are more willing to like kind of play ball on the pricing. It's like, you know, I'll pay you a little bit more for like these proven impressions and that'll actually like help me keep my, my budget allocations in check because I can orient more of that spend to stuff that actually works instead of doing this kind of like broad net thing where it's like, I don't know like what 85% of the stuff I'm even buying is, but maybe it's going to perform.
- [31:44]
A
So do you think this technology will be a forcing factor for, for more show level transparency? Especially if other people kind of come up with their own. I'm calling them hacks.
- [31:57]
B
My favorite show.
- [31:58]
C
Yeah, yeah. What I'm, what I'm hearing, you know, David Nuremberg at Intermedia, he was telling me like this isn't just like a Pier 39 like Pontiac thing, like they like they worked with them on that. But he said he's also seen stuff from Van Urstv and beeswax that it's also like oriented around embedding this kind of like content level data directly into like the buy side decisioning. And that was like another piece that like David was really excited about here because it's not just like you know, like you were saying Sarah before, like advertisers for years have been able to get this stuff from like linear TV buys from post campaign measurement which like can then inform subsequent campaigns. Sure. But the real kind of like big change here is like now they're getting this stuff like mid flight, they're getting it back like directly like and it's ingested into the dsp. So like when you're kind of like bringing that stuff into the buy side platform, it can go into that decisioning layer. So where it's like I can make mid flight adjustments to orient my spend more towards inventory that's performing really well for that particular brand instead of having to run a whole campaign and then optimize after the fact and like you know, for subsequent campaigns. So the actionability like the piece that like, you know that digital media is always kind of like pitched on or it's like we can do this stuff in real time and you get all kinds of like, insights that you wouldn't get from like, you know, traditional media channels. It's actually bringing us that, that value to fruition here for buyers.
- [33:20]
B
So that's also add, I'll also add on the publisher side, there's definitely a much stronger concentration on the idea of contextual ad targeting, not just on a show level, but on a really granular level. I talked to a lot of vendors and a lot of people on the publishing side. We're working with those vendors who say things like, oh, if there's a scene that takes place in a kitchen, we want the next ad to be for kitchen appliances. And I think that sort of changes the, you know, the arithmetic of like, how valuable is this particular show in this context? Because at a certain point if it's a kitchen scene, those happen in a lot of shows. So it might not necessarily matter how popular the show is if that kitchen scene is what's performing now, whether or not they're being super transparent with the buyers about where those ads are ending up and they're not just saying like, trust us, it's in a kitchen scene. That's a different conversation obviously. But I think that sort of changes the dynamic a lot as well.
- [34:13]
C
Just from a practical standpoint, in order to actually get an ad placed into like a contextual kitchen scene, you need some kind of like data signal that's coming in that can be decisioned on by the buying platform. You can't just like apply that stuff retroactively and just take the publisher's word for it. And like, you know, it's not just like content level stuff either. We're also seeing all kinds of stuff around like media quality signals, like attention metrics. Like the trade desk is in integrating attention metrics into its pre bid decisioning. Amazon DSP is doing the same thing. TriggerDesk is also doing all kinds of stuff with, you know, sincerity since they acquired them around media quality signals being folded into buy side decisioning. So it's a whole big trend around like making this stuff more actionable pre bid instead of just applying a post campaign.
- [34:56]
A
Yeah, I, I love that, Anthony. I think I'm going to, I'm going to tie it all together and echo your thoughts, which is, I think buying through DSP has always been viewed as like a performance medium and it's always been by buying by audience instead of content. But I think now we're seeing this pendulum swing back where media quality, whether it's the context, the content, the attention, the not mfa, not AI sloppiness, of it all is also being taken into consideration and I think an acknowledgment that audience alone does not translate to performance, which is great news I think for any high quality media publishers, programmers out there. Maybe the Yule log is high quality, maybe it's not. As well we can. That's the open question. So I'll leave everyone with that thought and we'll see you next week.