Loading summary
Transcript53 lines
- [00:00]
A
Foreign.
- [00:11]
B
Hello, hello and welcome to another episode of the Digiday Podcast, a show for anyone wondering when AI bots are going to start leading upfront negotiations. I'm Kamika McCoy, senior marketing reporter here at Digiday.
- [00:23]
A
And I'm Tim Peterson, executive editor of video and audio Digiday Media. Kimiko, when do you think Claude chatgpt Gemini are going to be handling upfront negotiations?
- [00:34]
B
I think they're already pretty deeply involved in the process as far as handling. And that last where we've talked about before, like that last 10%, right. I don't know, maybe this time next year. Where are your bets at
- [00:49]
A
leading the negotiations? I think never, if buyers and sellers are smart and want to keep their job. That being said, for all I know, the chief investment officers at agencies, chief ad sales officers at TV networks and streamers may be surreptitiously having Claude chatgpt so maybe like a kitchen cabinet of AI agents involved. But I want to bring in our media buying expert, our senior media buying editor, Michael Berge, who's joining us to talk about this. Michael, we're when do you think AI agents are going to be leading upfront negotiations?
- [01:31]
C
Hi guys. Thanks for having me here. I think it's going to be a couple of years. It does feel like it's rolling in really, really fast, but I do think there's kind of more kind of smoke than actual fire when there comes to this. There's a lot of talk of bots being kind of incorporated, but, but I look at the holding companies, the agency holding companies, and they're still kind of developing their systems and I just, I don't see it taking root in any significant way for at least another two years. But who knows? We'll see. It is coming in fast. Yeah.
- [02:12]
A
It's also interesting because like maybe the agency holding companies take a little bit more time, but maybe individual clients or other agencies don't. I mean, Kimiko, we just had Ellie Uberto from Duluth Trading Company on the show recently and she talked about they're already having AI agents be involved in their bidding, so they're buying. So it just feels like everything just keeps speeding up quickly and quickly. We'll get to the agentic AI part of the upfront in a bit. But another way in which changes in the upfront have spent sped up has been on the programmatic advertising side of things. So, Michael, you and I have been reporting on the upfront for years and I feel like each of us, at least once, if not more than that, have written a Story along the lines of the upfront model is changing, or the upfront model is not long for this world.
- [03:07]
C
Right, right. I think I've even said that. But that was about 15 years ago and it's not dead yet, as Monty Python so wisely declared.
- [03:16]
A
Yeah. So the upfront model has changed, though. Or maybe the upfront model hasn't changed so much as the. What the upfront has meant for how ad dollars are flowing is what's changing. I remember talking to folks buy side and sell side even, you know, just six years ago, and asking, okay, programmatic though, like, where does programmatic fit in the upfront? Does it fit in the upfront? And six years ago, even five years ago is like, no, not really. Like, that's a whole other thing. This is a very traditional marketplace.
- [03:54]
C
Right.
- [03:55]
A
Three, four years ago, that conversation started changing a bit. Like, well, there's. It's kind of related to it, but it's a separate thing more or less than about three years ago. 2. Three years ago it was. Yeah, we're starting to talk about actually having our programmatic spend count towards these upfront commitments. And now this year, I've been having conversations with folks on the sell side, and I spoke with Jamie Power at Disney. She said that for their. Disney's biddable inventory. So the inventory it puts up for sale and programmatic auctions, 70% of the advertiser demand is coming from upfront advertisers. So that's not quite the same as, like, people in the upfront committing to spend a certain portion of money programmatically, but it shows how the upfront has become a majority of Disney's biddable ad sales.
- [04:52]
C
Yeah, exactly. You can't say that 70% of the upfront is being bought programmatically, but the Disney example is a really good one. I mean, I think of the impact that the pandemic had on everyone needing to become digitally savvy in their business. And then the effect of that has been being able to then buy and invest programmatically as a result of everything being digital. And as you know, one buyer said to me, by investing programmatically, you just get a much clearer story of what the audience is you're buying. It allows for more clarity. But what it does take away from is, and you and I remember this historically, the relationship part of the business. And you and I both speak to some of the major kind of chief investment officers who've been doing this for a long time when it was just tv. And that ends up being a little bit sacrificed But I think those chief investment officers, until they retire, aren't going to give up too much more ground to that, because I think relationships still have a part of this business, if
- [06:06]
A
only because it's easier to haggle with another person. We were just at the Programmatic Marketing Summit. I feel like one thing that came up in a few of the conversations was, okay, cool, I'm fine one day having AI agents handle buys for me, but I would still like to, at the end of the day, if I'm not getting the deal that I want, be able to call someone up.
- [06:29]
B
Yeah, absolutely. That's one of the conversations that I had, I think, on stage, where there's still a value to being able to just call someone and be like, hey, I actually need a quick favor. What can you do for me here? And a bot simply can't do that. It was actually brought up on stage at least twice in the conversations that I had. I'm curious though, when you guys are talking about the, as I, you know, sit on the brand marketing side of things, hearing this conversation is so interesting. But what is the actual catalyst behind this happening? I mean, part of it. Yeah, I would imagine the pandemic and moving into like a digital space, but I feel like that's maybe the underpinning and there's more layers here.
- [07:07]
C
Well, I, I would say I would go back to the, the fact that you can kind of get a clearer idea of the audience that you're buying. So I think there's, there's kind of metrics related reasons to be buying programmatically, although you, there's a little bit less clarity of what exactly you are buying. There is better, a better idea of the audiences that you're trying to reach and that you can buy through programmatic investment. And I, I, I imagine I've never been a programmatic buyer, so I don't exactly know, but I imagine it might be a little bit easier doing this by keystroke than by phone calls and iOS being emailed one way or another. So I think there might be a little bit more simplicity to it. But Tim, correct me if I'm wrong about that.
- [07:53]
A
No, that's about right. Yeah. Like, it'd be super easy back in the day to be like, hey, I want to advertise in Sunday Night Football on NBC. A human seller on the other end could be like, cool, I'll put you in Sunday Night Football. Your ad will run. And that's like one spot that gets shown to everyone. But the thing programmatic really introduced and popularized Was I don't want to reach everyone watching Sunday Night Football necessarily. I'd actually rather just reach people who have a certain profile who are watching Sunday Night Football. Now, a human seller theoretically could sit there and put all the rules in place and plug the ad and make sure it runs when it can. But who wants to do that kind of thing? So instead we're just going to have a computer program automate that, set up the rules and then have it be the one to be able to recognize, okay, someone fits this profile, we're going to target them with that ad, and then we're going to handle the delivery and the measurement and attribution and all, all the things that like a normal person maybe could handle and maybe tear their hair out a bit. But at the scale of something like Sunday Night Football that's being watched by millions, tens of millions of people, doing that at scale could require just as many people as watching the.
- [09:21]
C
And also the idea of that being closer and closer to real time. We've been, you know, Sam Bradley's written a couple of stories and I think others have written stories about how using kind of programmatic means allows for that to happen within like hours or even minutes based on kind of contextual guides that allow that to happen in a much shorter timeframe. There aren't that many humans sitting at their desk Sunday night for a Sunday Night Football game that are going to say, oh yeah, let me get that in. So. And then finally, I would even add, as much as I hate to admit it, you know, there is cost savings for an agency or a brand to not have three or four people trafficking when it can be done through computer.
- [10:10]
A
And also, like, you know, when it comes to programmatic, especially if you're an advertiser and you have your preferred demand side platform, or maybe, you know, a few of them, then you can have the ad tech manage your buy across different streaming services and increasingly TV networks as well. And so that becomes your kind of programmatic proxy for you in a way. So there's been, yeah, a big push on the buy side to I want to get more targeted, I want to get smarter, I want to get more efficient with my buying programmatic lens to that. The sell side had been more reticent in a way of the whole point of the upfront for TV networks and streaming services is I'm going to sign these upfront commitments so that advertisers, agency holding companies are committing to spend X millions of dollars with me over the next year. Great. Now I know I have Revenue coming for the next year. Traditionally, Programmatic has been all about in the moment, spending and so it's not really assured. So there was kind of a phase to this where you had Programmatic Guaranteed, which was basically just automating direct iOS, automating traditional buys and having computers do the execution. But all of that money was very much wrapped up like a traditional upfront commitment. And that was a big sticking point. Bergi, I'm sure you've reported on this too, where the sellers, the TV networks and the streaming services like Programmatic guaranteed because they were in control and that money was guaranteed as the name implies. Whereas you had the buyers, the agencies and the brands who preferred private marketplace deals where they control the spending. They can kind of come in and out as they wish. And this was a tension for a while. For a few years there, it seems like things have shifted and there's kind of a relaxation. I think the sellers would still like to do PG deals. But I spoke with Jill Steinhauser from Warner Brothers Discovery ahead of this year's upfront presentations and she was saying that she's seen the pendulum swing where PG is still in the mix, but the growth towards PMP is overtaking it.
- [12:36]
C
Well, I think that's all really interesting and I think it kind of relates to that need for flexibility. If there's one thing, if there's one kind of theme we've seen in the last few years, especially as there's economic uncertainty, brands and their agencies want to be sure that although they're committing long term dollars, that there are exit ramps along the way. It sounds like what you described kind of enables that, but still enables kind of, you know, automated investment. So I think those two are kind of related to each other.
- [13:14]
A
Yeah, there's also this pressure coming from the tech giants Amazon and YouTube. Yes, YouTube, which is part of Google specifically, where they're already automating a lot of streaming inventory. Like Amazon has Prime Video now, which is ad supported, but it also has the Fire TV platform in it. It gets a share of inventory from major streaming services as well as long tail streaming services. And so it's conceivable for an advertiser to just get that inventory through Amazon's DSP. Similarly, if you want to be buying YouTube programmatically, you have to go through YouTube or through Google's DSP DV360. And DV360 also plugs into other streaming services. And so it's been really interesting to read like the reporting. Our executive editor of New Seb Joseph and our senior ad tech reporter, Ronan Shields have been doing this year around how Amazon and YouTube are kind of approaching the upfront. And I think an Amazon exec even told Seb as much of, like, they're looking at the upfront as, like, enterprise tech deals, more so than traditional media deals.
- [14:31]
C
Yeah. And, and I think that's. That's inevitable as Amazon and YouTube, their content becomes a bigger part of kind of where advertisers want to put their money. Now, the chief investment officers that I was referring to before certainly try to find their ways to push back on that so that they're not ceding too much control to that. But, yeah, I mean, we've seen Amazon push so hard on its DSP for the last two years. It. It's little surprise that they're making that connection between, you know, their, their inventory on prime with, you know, needing to go through their DSP to make to. To get the really good stuff.
- [15:11]
A
Especially because Amazon doesn't only have to sell CTV inventory in the upfront.
- [15:18]
C
Right.
- [15:18]
A
You know, Kameka, they're the biggest retail media network out there.
- [15:22]
B
Yeah, they are massive. One of the running jokes and conversations that I have is like, there's Amazon and there's Walmart and then there's everyone else. So I would. I can see the case to be made for, like, a large share of voice there. But in, in hearing this conversation that you guys are having, the thing that keeps coming up for me is so, like, it seems like there's a massive shift of power dynamics happening here. I hear the case to be made for buyers. I hear the case to be made for the tech behemoths. If I was a seller, I'd be feeling a little slighted right now.
- [15:57]
A
Why is that?
- [15:58]
B
Because there's not the commitment that there was. There's more flexibility that caters to others and not necessarily me.
- [16:07]
A
Well, and that's why they're trying to figure out how to thread that needle. So for things like, you know, like what we were talking about at the start of this conversation of the sellers are still doing these big upfront commitments, you're saying, like, look, if you want Sunday Night Football, if you want the super bowl, you're going to need to be in the upfront for that. Maybe, like, we'll make that inventory available later on, but first dibs will come from the upfront. But if, then, if you're making these big commitments, if you want to, like, spend that money programmatically, like, if you want to in Q4, have I keep coming Back to Sunday Night Football. But like have Sunday Night Football, have what other ever other shows on NBC or Peacock. And you also just want to be like picking up NBC inventory on the go. Whether it's, you know, on Peacock or on the digital properties. If you're buying that through a private marketplace, that'll count towards whatever money you're on the hook to be spending in Q4.
- [17:16]
B
So this all boats rise with the tide or whatever that saying is type
- [17:20]
A
of deal I don't know about. I mean this is business. I'm sure there are somebody that are sinking or taking on someone's losing, someone's losing here. But I think it's also just like media has been trending more and more towards automation and programmatic is a big part of that. It's like what Bergi was saying. How much do you really need humans in the loop for certain kinds of buys? For some things, absolutely you do, and especially the bigger deals. But when it comes to execution or more of spots and dots type buying that can be handed over to ad tech firms and eventually AI agents.
- [18:05]
C
Yeah, large scale integrations, you know, where you're talking about like deep seated like sponsorships and stuff like that, I don't really see a future where that's going to be done programmatically. That's still kind of relationship part of the business. But I don't want to jump the gun. But I would also say that maybe this is why some of the sellers are pushing so hard on offering agentic means as a way to kind of secure a stronger position in the buy sell negotiation. But we don't have to go there yet if you're not ready.
- [18:37]
A
No, let's go there. Kimiko and I had Ryan McConville from NBCUniversal on the show earlier this year and they had done a big test where with RPA and Newton Research, they actually used AI agents to to sell ad inventory in a live NFL playoff game. So they kind of came out of the gates swinging. And what was really interesting in that conversation that we had with Ryan was he talked about how AI agents enable what he termed premium automation. The idea that like in a traditional TV ad buy, they're are aspects of deals that aren't suited whatsoever to the OpenRTB standards that govern programmatic. You just don't have the rules in place to be able to handle parts of the negotiation that you would normally need a human for. But with AI agents that are able to reason, they're able to be a lot more agile than programmatic pipes that Are, as the term indicates, pipes. Like, you can't have a pipe change direction as it needs to. It's going to be. Whereas an agent can make decisions and complete tasks. That's the whole thing about agentic AI. Kimiko, am I remembering that conversation right?
- [20:08]
B
Yep, absolutely remembering that conversation. Right. And yeah, and I think it kind of points. And we had that conversation, was that earlier this year that we had that conversation. And things have just kind of been moving at 100 miles per hour ever since, I fear. And I think that kind of maybe shortens to me, shortens the time gap between, like when some of these processes in media buying and whatnot and programmatics start to become even more automated, especially as you start to, like, spend more into these automated pipes. So they. They learn and become, to your point, Tim, more agile and responsive.
- [20:44]
A
Yeah. Because Bergi, I think of it as, like, so often when we're talking to folks on the buy side and sell side about agentic AI and their advertising workflows, the first thing people, or one of the first things people talk about is like, well, we could use them for planning. They can actually, like, be really helpful for planning.
- [21:03]
C
Yeah.
- [21:03]
A
And I mean, as much as the upfront is a buying cycle, there's so much planning that revolves around the upfront.
- [21:12]
C
Yeah. And it feels like that is a natural place for that kind of smarts to be applied because there's no doubting that AI can be very smart so long as you train it properly. And planning is a way of setting out kind of your plan of attack, but without the actual investment part. Sam Bradley, my colleague, did. Did a story a few months ago, really small agency that tried actually implementing investment through an agent. And the agent found like one vein of investment that it was like, oh, yeah, this really works. And it went and took like 10 times the budget that they had put aside for that client and was like, here, we're going to put it here, put it here, put it here, put it here. And the thing went out of control. And 24 hours later, the agency had to pull the plug on it because it was smart but had no context of limitation or, well, don't put all your eggs in this basket. So I think that I found that to be very telling. Now it's a very small agency with four people who I think just tested this out. But to me, it's a little bit of a canary in the coal mine of what the risks might be. So I think for planning, I think we see a lot of application of AI on the creative side too. That's not, that's less a part of the upfront, but is still an area where I think agencies and brands can, can learn how AI works and thinks and how they need to train it so that then they can put the right guardrails in place. I don't think they're there. But Kimiko, I grudgingly admit you're probably right about this being here next year in much larger, in a much larger presence than what. What I think is going to happen.
- [23:03]
B
Well, yeah, I think because a lot of people are thinking about guard rails now there's some experimentation happening which gives me the confidence that says, you know, this time next year more of the kinks will be worked out and being willing to hand over the reigns increasingly to automated bots. I will say though, as long as the more and more that we had this conversation, the more and more I'm like maybe a year and a half. I don't know.
- [23:23]
A
I, I don't know because I talked to. So I talked with folks at Paramount who has a seller agent that they're in a closed beta test on. Disney's been testing agentic Warner Brothers Discovery's been testing around this and trying to figure things out. Also AMC Global Media and Televisa Univision are trying to figure this out too. And I think it was John Kozak at Televisa Univision who is saying give us six months to a year. We're going to use this upfront cycle to have the conversations with the buyers to understand what are their buyer agents, what do we need to be doing with our seller agents, what do we need to put in place and then we'll see what the conversation or like the actual deals are a year from now. But it feels like they're moving pretty quickly on this because there is a clear opportunity. I wonder how much of that also comes back to the experience with the programmatic adoption or kind of that fusion of programmatic in the upfront. If that laid a lot of the groundwork where now they can be applying AI agents again. Still with humans in the loop and doing oversight but talked with the co ad sales heads at Warren Bros. Discovery and they talked about using AI agents to go through their inventory to be able to like help with supply forecasting. So and I mean you're using it for supply forecasting. Well, what are you doing those supply forecasts for? So that you can come up, come into the upfront conversations with brands and agencies and say here's the inventory I'm going to have available for you now let's haggle.
- [25:14]
C
Yeah, yeah. One, one thought I have too is if, if the bots really do kind of become more able to do what, what you know, kind of buyers and sellers want, does this advance the idea of in housing more where brands can take control of this themselves? Maybe you don't need an agent if you're an agency. Rather, I'm sorry, I got to be careful with the, the words here. Maybe you don't need an agency because you have an agent working for you. And if sellers are so aggressively putting these kind of these platforms and opportunities out there, can they work directly with brands? I don't know. I'm asking the question. I don't know the answer.
- [25:58]
A
I think the answer will come down to the next question which would be, well, how much are you going to spend with us? Because the agency holding companies, the big brands, they still have the budgets, they're going to get the pole position. So it's great for smaller agencies and smaller brands to be agile that you know, there is the opportunity for them to in house. But if they're not spending as much money, there's less incentive for the sellers to want to work with them. It's similar to Bergy. We were at the Programmatic Marketing Summit in December of last year and there was a lot of conversation around like log level data. And I remember there was, you know, smaller agencies who were talking about, oh, I want to get log level data. I'm really frustrated that I'm not getting log level data. And then I believe it was, we had publicists on stage, he said, oh, we get log level data, it's great.
- [26:53]
C
Yeah, we get whatever we want.
- [26:55]
A
The reason was it's publicist, they have a lot of money to spend and so one of the perks of all of that spending is you get things like log level data and you can
- [27:05]
C
be sure the holding companies will continue to wield that clout and that scale to their advantage as much as they can. Because the last thing they want to see is more and more brands going in house with stuff and you know, in housing had kind of in my mind had slowed down a little bit at the very beginning of kind of the incursion of AI into the buy sell process. But as it gets more sophisticated there could be a risk there.
- [27:35]
A
But yeah, and that comes back to kind of the revenue assurance incentive where that's why the upfront exists because the sellers want to bank that inventory or bank that money ahead of time and so they're willing to offer discounts. Or other perks to entice buyers to make those annual commitments. And I mean we even see an ad tech which doesn't necessarily need an upfront model, but an upfront model exists. We've already talked about with like Amazon DSP and what Google slash, YouTube are doing with DV360. But you know, Kimiko, we were at the Programmatic Marketing Summit in May of this year. There was a lot of conversation around the upfront type deals that the trade desk signs with advertisers and agencies where there's minimum monthly spend requirements.
- [28:28]
B
Yeah, I think I'm almost wondering like the last time that brands tried to bring programmatic inhouse, what then happened? There were frustrations around like spending commitments. There were frustrations around who had time to do hands on keyboards and who was going to be like the institutional knowledge from things like that. And like if those issues weren't resolved, you know, does Agentix stand to resolve them for that conversation to then happen again?
- [28:54]
A
So that's interesting because I feel like Birgi, we're going to keep writing the story about the upfront model is changing, but as much as it's changing, with programmatic becoming more and more incorporated into the upfront or the upfront incorporated into programmatic businesses and now with AI agents coming into the mix, it still feels like the upfront model survives. It's just getting more and more automated.
- [29:23]
C
I agree completely. And I think to your point that you've made a few times that, you know, sellers need to know they have some degree of guaranteed revenue coming in and if they don't have that, how, you know, how on earth can a Paramount or a WPD soon to become one company, you know, know exactly what to expect and shareholders have expectations. So I think the upfront model. Yeah, is it, it's going to be here, but yeah, it's, it's getting more and more different every year and still fun to cover. I, this is in, in a weird way, this is my favorite time of year because you just kind of get, I mean, I still love the relationship part of this too. And the, the haggling and the, oh, these guys did this. Oh, that was so underhanded. Those guys did that. I love that part of it. But that might be a smaller and smaller part of it as we go into the future. Because you're not really haggling with agents, are you?
- [30:22]
A
Right. Yeah, we move into that forever. Upfront model.
- [30:26]
C
Yeah.
- [30:28]
A
Well, Michael, thanks so much for coming on the show and talking with Kumiko.
- [30:31]
C
Now a real pleasure.
- [30:32]
A
Thanks guys. Thanks for listening to this episode of the JJA podcast. If you enjoyed it, please leave us a rating and a review on Apple Podcasts, Spotify or wherever you're listening. Get more from Digiday with our daily newsletter sent out each weekday morning. Visit digiday.comnewsletters to sign up.