Loading summary
Transcript55 lines
- [00:00]
A
Foreign.
- [00:10]
B
Hello, hello and welcome to another episode of the Digiday Podcast, a show for everyone who needs a debrief on what the hell happened at this year CES. I'm Kameka McCoy, senior marketing reporter here at Digiday.
- [00:21]
A
And I'm Tim Peterson, executive editor of video and audio at Digiday Media. And absolutely someone who needs a debrief on what the hell happened at CES this year. Which is great because we are joined today by Seb Joseph, our executive editor of news, who is fresh back from Vegas. Although Seb, maybe fresh isn't the best.
- [00:41]
C
Word to describe your state of Definitely not. Definitely not tired, jet lagged in set of us.
- [00:48]
A
How was CES this year?
- [00:51]
C
Yeah, first and foremost. Thank you both for kind of having me. Ces, what was it like this year?
- [00:57]
A
Did it feel like very. Cause this was your first ces, right?
- [01:01]
C
Nah, it's the first time I've been there in about a decade now. So, you know, back then it really felt like you had to scrimp and sort of scrounge for kind of advertising stories, whereas this time it's sort of it's hard to avoid them. I guess the kind of main takeaway for me was that was a really intense but kind of necessary reset heading into sort of 2026, kind of with so many, with so many stakeholders in one place, it was useful to kind of get a read on the themes likely to define the year ahead. And what really stood out for me was kind of how consistently pragmatic the kind of outlooks were. This felt less like. And I think we sort of I got at this in the future marketing briefing sort of last week, but it felt like less of an industry chasing the next kind of shiny object and more like one trying to untangle the loose threads it left hanging kind of last year. Particularly, you know, when it came to sort of AI.
- [02:16]
B
I'd be interested, right? Because for the past, I want to say at least two years we've been in this AI hype cycle. But to your point, marketing budgets are once again stretched thin. Right. There's the expectation to do more with less. But everybody, it seemed like for all the coverage that I was reading and that you guys were putting out, everyone there was hawking some AI powered something and if it not AI, then agentic. Right. So what was kind of the media buyers and marketers response to, I don't want to call it just to the hawking of AI products?
- [02:50]
C
Hawkins.
- [02:51]
B
Right.
- [02:54]
C
I think the thing that really stood out for me was like there seemed to be a really clear delineation between the announcements and then kind of what was being said behind closed doors, at least in the conversations I was having. And the clearest way to sort of articulate that, I guess, was when it came to sort of LLMs, right across the kind of buyers agencies, even ad tech engineers that I sort of spoke to, there was a broad agreement that LLMs are reshaping kind of workflows, but not replacing decision making. Right. They are accelerating how people work, but not displacing essentially who controls the work. And I know this came up in some of the reporting we took away from us at the DJ Day Programmatic Marketing Summit last month in New Orleans. But ultimately the consensus seemed to be that LLMs are not yet fit for autonomous buying. Right. They excel at the probabilistic reasoning sort of side of things. But when you think about it, programmatic buying is built on deterministic optimization. They operate through trying to think of the best way to describe it. They operate through abstraction, whereas RTB operates through kind of really hard constraints. And so even though I think those technical gaps will narrow over time and the models are improving sort of quickly, at least for now, there seemed to be a real consensus that, you know, the LLMs, if they are going to sort of touch ad dollars, it will be very kind of small, essentially kind of tests versus anything remotely, you know, kind of of note. But to the kind of point I just made, I think within that, it was less about the technical limitations, more about, I guess, kind of philosophical ones.
- [05:01]
A
What's interesting about that, because coming off the Programmatic Marketing Summit last month, I was of that mind too. Every agency exec, it felt like we talked to at dpms, when we asked, okay, what about AI agents at the point of transaction handling transactions, they're just like, absolutely not. We don't need that. And I have so many questions about that. I don't trust also these AI agents or these large language models. Our language models, they suck at math. I don't want them getting a zero wrong, a decimal point wrong. And then last week, NBCUniversal comes out and says, hey, actually we are absolutely going to be enabling AI agents for buying. We are doing this pilot program with rpa, the ad agency, as well as Freewheel, and we're going to enable agentic buying for traditional TV and digital. And in fact, we. We're going to have an unnamed prominent advertiser that's going to be taking advantage of this for, I believe, an NFL game. I think they said a Professional football game, which maybe they're playing with words a little bit there, but like, that seemed to fly in the face of what I was expecting for this year of AI agents not touching transactions, where NBCU is here saying, we. No, we are absolutely having AI agents touching transactions and potentially big transactions.
- [06:30]
B
Yeah.
- [06:30]
C
It comes back, I think, to the kind of point about this not really being a technical issue. Right. Like, I think ultimately, you know, whilst you'll have certain players launching agents, I think it's. You'd have to look at it more in the round. Right. And see whether or not you're seeing more activity from the kind of buy side as well. Right. Or other sellers, you know, kind of starting to sort of build their own agents in order to kind of handle that autonomous sort of decisioning. Because. And this was a really interesting point that came up in one of the conversations I had. But, you know, ultimately, you know, for LLMs to replace humans in that sort of process and, you know, some of the existing sort of tech that's already there, they would need to be trained on the programmatic supply chain itself. Right. A system that's widely acknowledged to be noisy, incomplete and adversarial in sort of many respects. And so teaching AI on that. On that substrate would not make advertising necessarily smarter. It would sort of make its blind spots more permanent and more like. And faster and sort of like, more expensive. It would just compound a lot of the kind of the worst aspects of sort of, you know, kind of online advertising.
- [08:04]
A
So because you're not handling AI agent, You're not handing AI agents like a very clear, predefined rule book. You're handing them like a notebook with a bunch of pages ripped out and posted, sanitating things.
- [08:19]
B
That's it.
- [08:19]
C
Right. And I think it was interesting because a couple of buyers actually said sort of, you know, kind of garbage in, garbage out. Right. That came up in most of the conversations I had with buyers. And I sort of brought this point, raised a kind of point to them, you know, and it was almost like it was a kind of warning. So I saw the. You know, I think it wasn't even just nbcu, was it? I think kind of Pubmatic had a. Had a kind of test campaign out there where they were sort of purporting that agents were sort of very involved in that set of trade. And I think that's kind of fine. I just wonder.
- [08:55]
A
There's been a lot of questions in that campaign to.
- [08:58]
C
Of course. No, no. Yeah. And we'll like. We've got a story that comes out later on this week that will unpick some of that.
- [09:05]
B
Right?
- [09:05]
C
Because I think to the earlier point that is more of a workflow thing that they're sort of, that they're claiming than sort of real agent to agent sort of transactions. And they have sort of clarified that following some questions from us. But you know, I do think that whilst it'll be one to watch, right, and whilst the NBCU kind of agent was interested and I sort of thought that sounded pretty cool, you know, I'm yet to see any of the kind of main advertisers or hulkos look to sort of spin up anything even remotely sort of close to that on the sort of buy side. And just kind of one other thing to kind of flag within all of that as well. I think we're forgetting that for these things to kind of these solutions to be sustainable in market, they have to sort of exist within a broader commercial model, right. The whole kind of outcomes based kind of shift that many sort of CEOs whole co CEOs talk about. And from a conversation I had with Sir Martin Sorrell, he basically said that there are no clients really actively looking to make that call.
- [10:21]
B
Right.
- [10:21]
C
He said often what normally happens is they talk to a CMO who is interested in at least sort of trialing that and it gets to the procurement team and they're like nah, it's too expensive.
- [10:31]
A
I wonder how much jockeying there is going on though on that point. Sir Martin Sorrell, famously a phenomenal salesman, knows how to like talk the market in certain directions. I, I'm not aware of like what S4 capital is up to on the agentic front, but I don't, I imagine they're doing things on there. I believe there was like an earnings report that they had last year where he made some comments respect to, with respect to AI. But the other major agencies have been very loud on the agentic front. The start of last week WPP announced its agent hub where they're going to have a whole bunch of agents within the open platform. Omnicom last week introduced the next generation of its Omni platform. Now that it has closed the IPG acquisition, it has a lot more data and so I wonder if some of this jockeying is legitimate and if some of it is throwing a bit of cold water on what the other agencies are doing. Similarly like time will tell, right?
- [11:42]
C
Like I think time will definitely sort of tell when yes, you know, everyone is trying to sort of create a narrative to fuel their own sort of commercial intentions. But like I don't know about you, but I'm yet to see any kind of business and subsequently marketing team make any major shifts on the AI front. Right. Like everything that we write about, everything that we see in the press is their tests or projects with intentions to do X, Y and Z in years, not months, not kind of weeks. So I think everyone is jockeying to position themselves at the kind of vanguard of whatever kind of future this is. But you know, when that kind of future comes, I don't know and I don't think anyone kind of does. I don't think there is any real driving economic force. Right. Which is always the thing that kind of forces a lot of these changes to push businesses to deviate from the operating playbooks that they sort of currently have. I think most forecasts, economic forecasts that you see, from what I was hearing kind of on the ground in Vegas last year, seem to suggest that the market is okay at the moment, despite how chaotic the macroeconomic sort of climate may sort of look. Actually kind of consumer spending and subsequently sort of advertising has held up okay in Q4 and looks okay to do so, at least for the first six months of the year. So until I think some of that starts to look more dicey, it's hard to see kind of CFOs ordering any kind of major shifts around sort of AI. I think the Holocaust have to do it because they are advertising funded advertising based businesses. But P and G Unilever isn't that right? So they have to take a more pragmatic kind of look at, at these situations.
- [13:49]
B
I'm actually super curious to kind of keep pulling on that thread. Right. Because Walmart are.
- [13:55]
C
Please don't.
- [14:00]
B
There's. There was also some retail media networks that, that showed up and they had a lot to say. Right. So like Walmart Connect, but I wrote about last week has released some agentic AI capabilities, especially around like sponsored search campaigns for like bidding keywords and billing and whatnot. I talked to one commerce expert and I was like, does any of this seem useful, this agentic AI offering? And he said no, I don't know who's asking for this. So like on the ground at ces, did you get any sense of what advertisers are asking for from AI? It seems to be there's one conversation happening on stage and then another one happening to your point behind closed doors.
- [14:43]
C
Yeah, I think it's everything that we've written about over the last. It's workflow stuff. Right. Everyone wants to do things faster, cheaper, better but ultimately that is about everything up until an ad is kind of bought. You know, from sort of planning to sort of. And after, I guess from planning to sort of optimization. Those types of things, the sort of creative side of it, like, you know, but not, I think it's that. And I think that will be the thing that will really drive growth for marketing services businesses, the ones that can really nail that and I guess differentiate their sort of products through kind of proprietary data and expertise, the creative expertise, strategic expertise that sits on top of those workflow optimization tools. But it sounds boring and almost contrarian to the narrative around AI. But ultimately, when I'm talking to buyers and the execs that are closest to them, that's the thing that, that CMOs want at the moment. No one can really afford to make major changes to their organization at this point. There's just too much uncertainty around to do that.
- [16:00]
A
And it seems like there's also uncertainty when it comes to the financials because I haven't talked to anyone who has a sense on what the regular recurring costs for AI would be. Especially like right now if anyone's tried to buy a memory card or a hard drive at the moment, like, God save you, like that's not going to go well for you if you can even find that GPUs. But then also just like token costs, which is something. Seb, we had spoken with you and Michael Berge back in August about of kind of the cost of using these large language models and how agencies would be able to assume afford those costs, especially at a certain scale. Did you get a sense on to what extent that is playing a factor here? The fact that just these costs associated with AI seem to be going up and it's unclear what their ceiling may be.
- [16:59]
C
Well, Tim, my inimitable colleague, you're in luck because that was one of the kind of focuses of a conversation I had with Wes Deha, who's the chief AI officer at S4 Capital. He was sort of talking about this cost of inference, so cost of compute, essentially. And that is an active part of the conversations that they are having with partners, including Nvidia at the moment. So that he was reluctant to talk it up too much because of those earlier points about the adoption of AI. Right. But he said that as it becomes a bigger component of how that business operates, it will naturally have to become a bigger component of the deals it brokers both with those partners. But then increasingly then how it starts to sort of charge CMOs as well. And so it was funny because that was a conversation that he said they're having in parallel with all of this, you know, the outcome based sort of kind of conversations that they're having with sort of marketers at the moment. So this idea of like cost of compute being something that agencies have to figure out and then therefore, you know, whether they sort of have those cloud costs be derived from sort of on premise side of things or sort of up in the cloud is I thought was one of the more interesting observations, you know, to sort of come from my sort of time out in Las Vegas last week.
- [18:38]
B
Given that this is your first time back to CES and in a decade, I'd be curious who were the key players that you saw show up this year and who kind of made the biggest impression. I guess.
- [18:53]
C
Everyone is kind of there. So I think the main stakeholders, ad tech, advertisers, agencies, platform was OpenAI there. Not that I saw. And no one I spoke to had said that they'd had any sort of conversations. It sounds like given how last year sort of ended that, you know, they are still, there's a lot of work to kind of be done there in terms of the, you know, how they take whatever proposition that they're sort of cooking up to sort of market. So can seems like the one that, you know, if they are going to sort of make us have a sort of coming out party to the ad industry, it will be there. I thought it was interesting to hear more people in the ad industry talk about Nvidia as a company that is kind of doing the rounds, not just sort of in ad tech, but on the agency side as well. So that's going to be something that we look into as a team a.
- [19:49]
A
Lot more this month, like as a brand. Nvidia looking for agencies?
- [19:53]
C
No, like as an infrastructure sort of player. Right.
- [19:58]
A
In terms of how some more circular spending.
- [20:02]
C
Oldies but goldies. So yeah, I thought that was interesting. Stavisi has been very much focused on their work with ad tech players, particularly on the SSP side. But it sounds like there's a lot more overtures being made on the, on the agency kind of side as well, increasingly. So you know that that was the real kind of standout. I guess for me the most kind of fruitful conversations were the ones that I had with the kind of buyers because I'm always just, I like those conversations. I'm able to kind of get a bit of a deep dive in terms of where the kind of money is sort of flowing. And so I think, you know, it was interesting to Kind of get a handle on how Q4 sort of ended up, but then also getting a steer on the sort of outlook for this year, which seems to be in line with the kind of forecast from, you know, kind of both WPP Media, but also kind of Madison and Wool, but then also the engineering, the engineers that I, you know, because AI is such a kind of big part of our sort of COVID our reporting base at the moment, just speaking to engineers about how they see these trends sort of shaking out or not, I thought was interesting, which comes back to the earlier point about the autonomous sort of buying side of things, because even these guys, the ones that are at the bleeding edge of some of these innovations, are hesitant to sort of talk it up too much at the moment. You know, the capability is there, but it does seem like it's more of a, I guess, a behavioral thing. And, you know, also I think, you know, turkeys don't vote for Thanksgiving, right? So I think there was a component of that, like the industry is just mindful about how quickly it embraces some of this technology so as not to sort of automate its way into oblivion. At least not too quickly.
- [22:04]
A
What about AI ads? OpenAI wasn't at CES. There still seems to be a ton of interest at the moment in what is this ad product or products that OpenAI is going to roll out. Kimiko, you mentioned Walmart. Walmart last week announced that it's putting ads in. I learned this name, Sparky, which it's AI chatbot. Over the weekend, the Financial Times had a story on Google putting ads in AI overviews, which is something Google announced last year. It was testing. When we had Dan Taylor from Google on the show back in May, we talked about that. It seems like that's expanding. So ads are coming to these AI chatbots. I would group AI overviews as a chatbot experience, even though it's not Gemini, but it's pretty damn close. It's really hard to tell the difference between the two at this point. Zep, how much talk was there around ads and AI chatbots? To what extent money is moving in that direction, what these ads should look like?
- [23:07]
C
Not much, which was kind of surprising, I guess, because it's still really early days. The industry, whilst kind of keeping one eye on it, seems very much focused on issues sort of elsewhere. You know, I think when they are thinking about these chatbots, a lot of the attention, you know, as you'd expect, is more about how they can show up in them organically, you know, and thinking about, you know what they need to sort of change about, you know, how their brands exist online in order to be able to influence those LLMs. But in terms of the actual ad experience in and of itself within them, you know, yes, obviously it's a kind of big deal whenever it does happen, but because there is no real kind of timeline on any of that just yet, you know, it felt like the industry is kind of at least sort of content for now to sort of worry about it later on down the line when it becomes a bigger sort of a bigger issue.
- [24:17]
A
So that'll be something we'll have to stay on top of. But we talked a ton about AI. Was there talk about anything else during the csi?
- [24:27]
B
I was literally just about to ask.
- [24:29]
A
Were, like, creators a big thing at ces? I know that's been the case in the past at CES as well as at Cannes. I feel like I heard less about creators this ces.
- [24:40]
C
Yeah, let me. What did. I'm trying to think. Yeah, like, it didn't. It wasn't a major. A major sort of focal kind of point in. But I wonder how much of that was because of the conversations or the people that I was kind of speaking to. What was interesting, though, was the. It feels as though this year will be the year where creative measurement kind of finds its footing because, you know, kind of platforms need TV dollars, brands are no longer willing to sort of spend exorbitant amounts of money on sort of creatives without some. Without some robust sort of guardrails kind of there. And so from a couple of conversations I had, people seem to be sort of suggesting that don't be surprised if by Cannes in the summer, you see launches from the likes of kind of Nielsen and potentially the likes of Publicis, which has made obviously a big investment in that space in terms of their own currencies. And I know Tim and Com Store.
- [26:04]
A
Had some news last week around measurement.
- [26:06]
C
For influencer content, so I think that's. Whilst I didn't hear loads about it, the brief conversations I had were very much focused on this kind of almost being the year of sort of measurement in that regard. So it's something that our new creator reporter Alyssa, is kind of looking into at the moment. And it will definitely be something I think, like, is one of the kind of big kind of reporting lines for that sort of part of our coverage this year.
- [26:47]
A
Okay, so while all the, like, consumer electronics folks are over at the Las Vegas Convention center gagging over robot dogs and whatever the ad industry is at Aria talking infrastructure, it sounds like that sounds.
- [27:02]
C
Yeah, that's it. It really felt. Yeah, it kind of felt like an in the weeds sort of event. Right. Like, I was quite surprised at how kind of hardcore the conversations were. There was like there wasn't a lot of kind of fat to everything. Everything felt like, at least from, again, from the conversations I had, they were more substantive in a way that kind of threw me a sort of little bit.
- [27:29]
B
So, yeah, I think it also lights up interestingly for CAN when that comes this year, because it seems like a lot of the conversations that you had sub were kind of like breadcrumbing to lead up to that. So maybe the chips were stacked, you know, at CES and we'll see how they fall later this year. Yeah.
- [27:46]
C
On the crater point, I thought it was interesting that so much of the issue or the bottleneck seem to be with the platforms. Right. In terms of allowing that data to kind of flow into the sort of market. And again, from the execs that I were talking to, they had noticed in recent sort of months a kind of shift there, which I guess is a precursor to whatever currencies are sort of coming down the line. I think the other thing I thought was interesting, though, is that whether or.
- [28:18]
B
Not.
- [28:20]
C
It sounds like Publicis is cooking something up, but whether or not the other Holcos attempt to do something similar, and then we just have more fragmentation. So many different sort of currencies in the market purporting to sort of do the kind of same thing. So I thought that part of the conversation I was having around the crater, the economy, was also interesting. It gave me kind of pulse for thought in terms of how that might sort of shake out later on this year.
- [28:49]
A
Certainly. Yeah, a lot of questions around how all the conversations that you had during CES are going to shake out through the end of this year. But I guess that's what makes CES so great, is it does in some ways set the agenda for the year, in other ways, kind of temperature checks the agenda, which I think is like the phrasing you would use SEB in your Future of Marketing briefing, recapping takeaways from ces. So for anyone listening who wants more of what Seb's experience with CES was like, the conversations he had, his takeaways. Check out the latest Future of marketing briefing on digitad.com, seb, thanks so much for after CES. Still dealing with the Vegas hangover. Not an actual hangover, I don't think, but CES is its own kind of hangover.
- [29:35]
C
Thank you for having me, guys. Appreciate it.
- [29:43]
A
Thanks for listening to this episode of the JJ Podcast. If you enjoyed it, please leave us a rating and a review on Apple Podcasts, Spotify or wherever you're list listening. Get more from Digiday with our daily newsletter sent out each weekday morning. Visit digiday. Com newsletters to sign up.