Loading summary
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. This episode is brought to you by Zada. Zeta's AI platform unifies paid, owned and earned media into one powerful growth engine, optimizing performance in real time and driving measurable outcomes with confidence. Don't Hope for higher ROI. Expect it with Zeta. Learn more at zetaglobal.com ADExchanger. If you haven't been paying attention to what's going on with agencies and principal media, now's your chance to pay attention. Two things have happened in recent weeks that are putting agencies principle based buying practices in relief. First, as we discussed on the podcast last month, a lawsuit between WP Media and a former executive who says he was fired for raising concerns about these practices included a number of documents that revealed how the agency was allocating billions of dollars of its clients client's money and it's planned to partner more closely with the places it spent the most. And then this week we learned that Kublasys is no longer recommending the trade desk as its DSP after it refused to answer questions in a process described as an audit and publicist says it found hidden fees added to its contract. Our senior editor James Hersher has been had been working on his story about principal media when this news came out and we are going to bundle it all together to explain what's going on with agency principle based buying and what it means specifically for Programmatic spend. Then let's look to the future. Sell side agents offer the promise for publishers to better optimize their internal processes and even the idea of having an agent that will help sell their inventory to another human in an easier way or or via an AI to AI interaction. We're just at the beginning, but our news editor Andrew Bird was able to talk to a publisher and an ad tech company about their early test to build sell side agents. He'll walk us through what problems people are trying to solve and how it's going so far. I'm Sarah Sluss, Editorial Director at Ad Exchanger and before we get started, anyone who's interested in the Sell side Agent segment should definitely check out Programmatic AI or or conference focused on how this new tech is transforming digital marketing and media. Our listeners can use the code POD10 for an extra discount and sign up by going to the Events tab on our website. Now let's dive in. So James, let's just start at the Absolute Basics Walk me through how principal media actually works in practice. We can hold that in our heads as we have the rest of this discussion.
B
Sure. So at the most basic level, principal media is where agencies negotiate directly with a big inventory owner, which could be a retailer or like a big entertainment company, an online platform, all sorts of, you know, just lots of different TV manufacturer. And for upfront ad investment commitments, they get, they get a sort of a guaranteed set of inventory at a very low rate that they resell to their clients. So it isn't like the normal kind of like they're buying live inventory. It's almost more like ad network ish, where they own like a contractual. Right. They actually own this media for a time. They resell it to their clients and obviously sort of at a markup. It's a higher margin business for agencies.
A
So the idea is they're, they're almost like a wholesaler store in a way. They're buying a bunch of stuff, they're getting it at a bulk discount and then they're selling it at a partial, maybe a partial discount, but partial markup to clients, you know, in theory.
B
And then they're also. The benefit is it's cheaper than the media you're gonna get, but they are still sort of have a handsome markup for themselves.
C
Yeah.
A
James, is it kind of like an interior designer or something where like they get all the couches at 20 off and then they like charge their clients design fees and then maybe like the full price for the couch or like
B
it is, it is that. That definitely is a metaphor. I do remember that, like come to mind. Like, you know, you might have an interior designer. It's like, you know, oh, like, right. I get 10%, I get you a 15% discount. Like, it's all right. That is kind of like part of it. I do think it's different though, because, you know, like, it just changes the fundamental agent relationship. And even like literally when these companies talk about it in their, like deep in their SEC filings, they actually note that like in these instances, we are not acting as the agent, we are acting as a third party, like services vendor, you know, so it's not, it's not just some kind of like quibble. They really are like, not the agent in this scenario.
A
So a decade ago, this was a huge, A huge thing. I, I was actually at the ANA media conference where this guy John Mansell got up on stage and he's a former executive and he's just like going on talking about how agencies are hiding all this money and their profits are going up and marketers don't know that they're being fleeced, essentially. And that led to this whole transparency audit. And I think a lot of marketers in particular took a look at their programmatic spend because these programmatic trading desks, I think in some case were able to take like 70% margins or something just to kind of throw a number out there. But the idea was it was not like a 10% off situation. It was much more so. And then there was kind of this idea that everyone's reviewing their contracts, it's the end of principle based media buying. But then now, over the past couple years, it feels like the pendulum has swung back in the other direction where marketers are kind of like, hey, like, this is kind of what, you know, makes sense. Like, I'm cool with it, like, or this is what, like, my finance, you know, team needs. Like, they won't let me pay this amount for my, you know, for my people. Right. Like, so, sure, they can just, you know, take a little bit of a markup on the media. Then finance doesn't care about that or something. So. So how is the pendulum swinging here?
B
It definitely has. I do think that, I mean, a. Like, there was, you know, 10 years ago too, there was like, rebates. I do think we can kind of like separate rebates out from this. But it also is a useful because, you know, it is sort of part of how this like, thing whole. This whole thing just like morphs these business models. I definitely think principal media buying sounds a lot like rebates. But instead of getting like, just a cash kickback, you get ads that you get to resell. And. And yeah, I think that is part of how this will just sort of evolve. Like, you know, things get tweaked. That WPP filing, it documented how, you know, there are these. This person is aggrieved that he brought up rebates. Like, he's, he's sort of presenting himself as a whistleblower. WPP is saying, like, look, your proposal included all these other ways that we could just sort of kind of like recreate rebates. Oh, like, we do upfront commitments with, you know, for advertising or content production with like A and E or TikTok, and we get like ad credits or first look or something like that. And it's like we're, we're like, you know, it's really rebatesy. It's principled. Me, it's kind of getting at the same idea, but, like, maybe sometimes like slightly different. Ways by which you get paid or like how the transaction unfolds. So I think that is part of it now. You know, it just, they keep kind of like cloaking themselves in more complex models for this. Although definitely truly like you know, rebates, you might think like oh that's terrible. In the US it's, it's like flatly not allowed but where it is legal like they are still totally commonplace and principal media like has a very like sour taste bad association. But is $1 billion business line for all the big holdcos is like a huge business which they kind of have to like not forthrightly acknowledge. So, so yeah it isn't, it is very common. It's a big business. But I, it isn't exactly like accepted
A
so she sheep in different clothing perhaps like the idea gets rebranded in ways that are considered more or less acceptable over time like rebates. Principle based buying. There's kind of nuances involved in these practices, but seems like what hasn't changed is that agencies are, are taking some, some money off the billions of dollars in media that they're spending. I want to talk about the intersection of principle based buying and programmatic. We, we know kind of historically trading desks were a big source of, of revenue. Brian Lesser who's on his like return boomeranged back to WPP and he was kind of a big part of the whole initial trading desk rise as one example. But you know we've, we've seen stuff go on with the trade desk here. We saw that Dentsu and WPP media in fact left open path over the fee structure, you know, quote. And then just earlier this week Adage broke the story that publicis is no longer recommending the trade desk as a dsp. It says there's these fees that maybe were not disclosed in part of its contract or hidden fees. The trade desk has clapped back saying that's not true at all. But what's going on here with programmatic and principle based buying.
B
So really why this is so particularly messy is that principle based buying, as I said before, it sort of is these direct reserved deals and often when they're doing it like the, you know, the advertiser has kind of like opted in to participating but they're not really like aware of what's going on and you know, maybe sort of don't understand like what, what inventory is coming through principal media and what isn't. And so like you know, I think the trade desk sort of feels like it's sort of like hey, like what's going on here. Like we're, you know, we're going to like shine a light on this. And also Open Path has kind of, you know, has blown this whole thing up because the trade desk is now going direct to publishers. And you know, Open Path is big in like CTV retail. Like a lot of, you know, these are the purveyors of principal media on the supply side, often who are integrated with the trade desk. And I think from the trade desk point of view, like the agency perspective is like, hey, stop shining a light here. Like this is, you know, this is like an ad network essentially that the agencies operate and they don't want, you know, like they don't need all of these like log files and the transparency. You know, I think that, and there's a competition now because they are, they're, they're both sort of trespassing in this new territory of going direct to media. And, and right. Like having, having these sort of direct integrations. And like they mentioned, oh, we can't disclose the auditing data because of confidential confidentiality agreements. That was what the trade desk said. Why they couldn't, they didn't fail this audit, they sort of didn't participate. And I'm curious, like, I think that those, those may have been sell side agreements. Like they, not necessarily that the advertisers like log files, they couldn't go hand over. But it was, it was sort of publisher side data that was being asked for.
A
I kind of was wondering if it was other agencies that they were like, we can't sell other agencies. But you're right, it's very unclear whether the trade desk not disclosing this data was reasonable or unreasonable. I think based on the coverage that's been out so far in my opinion.
B
And their curation services is now sort of what these agencies do. Like, you know, they, they might frame this as like curation, their principal media when they're, when they're talking to a client. Like I can easily see that there. Another thing that shows up in that filings is these very like, you know, euphemistic terms that come up. Like nobody wants to actually call it principal media or like rebates. So there's like risk purchase media and like proprietary trading media. Like you know, you got to keep track of like a million different terms in the, in different filings. They report it in very like strange categories, seem innocuous but like, okay, that's actually principal media. So yeah, it's hard to untangle part
A
of the hypothesis here, which I think is maybe a Little bit ironic is that there was a lot of talk about how many fees there were in Programmatic. Every step of the way you have the DSP fees, the SSP fees, the ad server fees, the data fees, all of that. So. And lack of transparency where the agency couldn't, could only ask or demand information from the dsp, not anyone else further down the chain. Right. So coming out of that we get these agency SPO relationships where SSPs and agencies are forging all these partnerships. We've covered a whole bunch of them at Ad Exchanger and it's like oh, we got like a better path to supply, a lower fee path to supply and like kind of other you know, add ons. And I think part of perhaps what's being alleged here, I think is that those agency to SSP relationships didn't reduce the fees. They merely had the fees accrue to the agency instead of, you know, the agency, dsp, SSP publisher maybe. And then and you know, and the trade desk is saying well are that. And that's why they pulled out because the open path fees were lower and it kind of expanded, exposed, you know, these like fee structures that weren't advantageous for you know, the marketer or the publisher. It.
B
Yeah, I think, I mean it is like for anyone who's all these agencies and the trade desk like they're sort of saying like oh this is our like fee reduction. That's the whole idea. Like oh, greater or roas we're reducing fees. Well that fee is like somebody else's lunch, you know, like there's no like it has to come from somebody. So they're all. There's not, there's no like you know the whole industry is so has is lean and mean. Like it's anything is just marrow deep cuts. Like there's no like there's no more like fat. There's no easy pick ins. So. So yeah, when you say like oh I'm making like a big like big cut to people's fees. Like okay, like no wonder SSPs are screaming like you're talking about SSPs when they're talking about like cutting out like buy side like fees and transparency. It's like okay, like that's the agency's launch. So there's just no way to like to make that next step to like okay, greater roas reduce fees without it hurting someone in your supply chain.
A
You talk to the ANA for this piece and they kind of say this is buyer, buyer beware. Right. And I think that's good advice. I also Think from having this. Sarah Caputo on our podcast, Great one, definitely listen to it. She talks about looking at these contracts where people thought they were putting together a great contract only to find that there were these fees that were cloaked in ways that they couldn't understand or like allowed the trade desk to charge more fees than they had thought. So how do you have buyer beware when maybe the buyers don't even know what to beware of? Like how do, how do agencies and marketers figure out what's being charge whether it's principal based buying or you know, an ad tech fee?
B
Well, I mean I think that might be like what you're seeing like this rise of sort of advisors and consultants and the sort of audit that caused this whole like imbroglio between publicists and trade desk is you know, these agencies and brands themselves. I have felt for a long time like that agency buyers and you know, the brands buyers who they represent are shockingly distant from the ad tech. The like the react like the nuances and like how their ad tech works. And so you know, maybe this is just sort of like part of that like introduction. This is happening in a lot of places. People are sort of demanding transparency that you know, the Internet has been like saying for a while and they say like they're not against principal media but they say like have a clear contract and have auditing rights. You know I think I'm trying to like ponder because there's an ANA thing that is embargoed for. I'm pretty sure when this is going to come out. Like we're going to be, we're going to come out after an A thing has already been announced by the way. And so there is like a whole, you know, there's even just in the next week there is a lot of like principal media focus that is going to happen and sort of new products and new like stats that are released. We haven't really had an update in like a good two and a half years from the A and A on this topic in which time like it has probably gotten a lot bigger and I think sort of become more of a problem. Yeah, like, and that's also like my point, like I feel like principal media is just like bad. Like this is not, this is just like wrong for agencies. Like this is just sort of like not a good, just fundamentally wrong and why they are kind of very like shame faced and shuffling about it. But yeah, it is coming for sure.
A
Okay, so we're, we'll sum up this segment by saying there There's a lot more to come so definitely pay attention to it. As a publisher, some of your, your money might be kind of siphoned off by these, you know, fees or you might be pressured to lower, lower your costs as part of the, these principle based buying agreements. Marketers and might not even know what they don't know.
B
It's interesting, wpp, one of the things that, that came up in that filing is that WPP's top 20 clients, none of them participate in WPP's principal media. And many of those are Google, Sony, Uber, Amazon, Paramount, which you know, they work with them as advertisers but as publishers, those companies sit sell principal media. Like these are the people who know what this is and they all explicitly do not opt in. So like, oh, it's like, oh, that kind of makes you think.
A
I thought the, the counter argument to that is that oh, they already have enough buying power on their own that they don't need to do that.
B
That should mean that they derive greater benefit. It's a pure volume thing. So they should be the most incentivized by like the sort of structure of principal media. But they do not participate. Yeah.
A
And I think the lesson to go back to one thing you said for anyone that's in the ad tech space is like, you know, are you eating a lunch that someone else wants to eat? You know, do you share half your sandwich? Is your sandwich stolen? And I think some of these, some of what's going on between the trade desk and agency is purportedly over, you know, who is taking this, this fee, whether it's a principal based fee or an ad tech fee or something. And that's all shifting around. So stay with us. I can't wait to talk about sell side agents. I can't wait to talk about, who knows, 10 years from now sell side agents and principal based buying. But for now they're separate. So be right back. I'm Sarah Sluice, Editorial director of Ad Exchanger and I have with me here the chief Growth officer of Zeta, Ed C. Who leads the charge in helping businesses and CMOs achieve measurable, high impact marketing outcomes. Thanks for joining us, Ed.
D
Sarah, great to see you. Thanks for taking the time with me today.
A
Yeah. So what's different in the year 2026 about how brands and agencies are approaching ad tech?
D
So Sarah, I think it's a really, really interesting time. For years we've all been talking about ad tech and Martech and all these different things, but at the end of the day, as a marketer what am I responsible for doing? Helping people make marketing decisions, helping them make buying decisions through a series of touch points. AdTech and MarTech are actually artifacts of our history of how we bought Meteor and how we managed our own touch points. What's finally happened with today's technology is we can release some of those artifacts and bring ad Tech and MarTech together. I think that's going to be a huge thing. We're seeing more and more people doing that, and it's releasing marketers to have the freedom to really say, how do I reach Sarah? How do I recognize Sarah? How do I reach Sarah? How do I offer something relevant? And how can I see the results of the communications I'm having with Sarah?
A
So, yeah, let's talk more about that first freedom that comes from blurring the lines between ad Tech and MarTech. What else happens when you kind of unify things and blur them, bring them together?
D
Well, a couple of things happen, Sarah. Instead of saying, how much did this channel produce? We actually started saying, how profitable is Sarah? To us, it changes from measuring a channel to doing what marketers are really responsible for, creating a profitable customer and saying, what did each touch point, what did each offer, what did each thing actually contribute to the profitability of that particular customer and allows the marketer to actually say, I'm managing a supply chain, and I'm managing the most important supply chain to a company, the customer supply chain.
A
So how does that unified customer view then empower marketing teams and agencies once they know about this profitable customer?
D
Well, one of the big things here is it lets them do the really unsexy side of marketing, waste management. And that is incredibly important. Making sure that you're able to stop the wasted impression. The extra touch point. We have all probably in the last two weeks bought something and then got another impression trying to sell you the thing that you just bought. It helps take those things out, things that you would never buy. It allows us as marketers to say, I am being able to recognize a pocket of opportunity with precision. I can reach that pocket of opportunity again with precision. I can be more informed of what relevant items I can bring to that person. And I can leave enough data to actually say what's working, what's not, and see the results. It's incredibly empowering. It's incredibly freeing. And as a marketer, it helps you have a better conversation with your cfo, which some of those conversations haven't always been so positive.
A
Okay, so less waste, more profit, better meetings with your cfo. Thank you, Ed, for Weighing it all down for us.
D
And it slices bread along the way too.
A
Wonderful. Thank you to Ed and thanks to Zeta for supporting our podcast.
D
Thank you, Sarah.
A
And we are back. So, Andrew, you've been doing a lot of research into sell side agents since last month when pre ADCP or the Ad Context Protocol donated their code to pre bed. Kind of opening the door, I think, for publishers to do some experimentation. And you looked into it and found some stuff that publishers are doing. So what is a sell side agent and why are people talking about them?
C
Yeah, I mean, I think you did a good job of summarizing it in the intro. A sell side agent is basically a AI system that kind of acts like an agentic salesperson for the publisher. Instead of optimizing things behind the scenes like floor prices or maybe which ad wins, this AI can actually help decide how inventory gets packaged, how it's priced and sold in the auction. So instead of like a buyer sending an RFP and waiting for a human to respond, the agent can understand what the buyer wants, match it to available inventory and audiences, and propose deal or campaign, and sometimes even start setting up that deal as well. And I think in the long run, people are looking for a sell side agent and a buy side agent to kind of interact with each other and kind of talk through those things. Instead of a human being in the loop. I think a human might be there and sometimes when like kinks work out. But based on what I've talked to people, like, I don't think they've gotten quite there yet where a sell side and a buy side agent are kind of doing all these processes themselves. Like a human is kind of looking over it, but it's essentially like being a salesperson for the publisher in a way.
A
And you talked to News Core, which I think is kind of just getting started and figuring out what it wants to do with AI. And then you also talk to the weather company, which is a bit further along, and testing things out. So why don't you give us a few examples of what problems they're trying to solve with their sell side agent.
C
Yeah, I mean, right now, as I said, it's a little bit early, so I know one thing that the weather company kind of talked about is there's like some premium inventory that might be overlooked. So like the sell side agent could kind of distill that to see like maybe some inventory that buyers are missing or some inventory that sellers are packaging in a way that buyers might want to buy. So as far as testing it, they kind of Started off with like a small set of like top performing ad formats. They kind of defined what a premium audience looks like and then sent it to a limited set of buyers to see how that kind of worked. Again it's still kind of a little bit early and a human is still kind of checking over it. But I think they're just trying to look for ways that you can kind of distill some of that preference pre premium inventory and kind of starting with small testing pools in that way.
A
And I open this segment by talking about AD cp. We also know there's an IB has an agentic roadmap and there's been some, some tension I think between these, these two different frameworks that are being developed. So of the people you talked to were people using one versus the other? Did they have opinions on how they wanted kind of the industry to change, move forward around these different frameworks?
C
Yeah, I think it's a little bit of both. Some people are using the framework and some people aren't. We mentioned the Weather company earlier. They are using ADCP's source code to kind of help build their sell side agent. They kind of work with Scope three who I know is a part of adcp with their, they worked with their ad product team, their ad tech team and their adopts team to kind of build that in themselves. And again I said they kind of tested small segments on the other end I talked to the data clean room company Optimal who is also a part of Prebid and kind of said like as they worked with the source code, prebid is still trying to figure out if they are going to start from scratch or kind of incrementally build from there. But I think even though Optimal is a part of pre bid, they kind of decided to build the sell side agent based on their own platform just because they had an existing system that worked within their platform. They have an audience agent which basically powers their sell side agent. And with the audience agent a salesperson can kind of upload an RFP into Claude Anthropic's tool and they can kind of put in a brief which is connected to Optimal's data environment and from there they can like map systems with like relevant audience segments and like mix them with like new and existing audience segments and that kind of powers like the data that will build the tool for their sell side agents to potentially make the deal.
A
Essentially I will just give a small, a small round of applause to Optimal for just being like hey, we're using Claude and like this is the interface because I Feel like so many people are like, we use all the LLMs, like they just kind of try to white label like, like make it go away, like whatever platforms they're using. But it kind of underscores, you know, maybe not the level of simplicity, but just like if you've ever used Claude, you're like, okay, I could imagine what like the optimal version of Claude is or what that interface would look like to help me, you know, build my, my data segments. So what's like one of the coolest things that you, that, you know, people were super psyched about, like, is there some kind of, you know, killer app that something people are going to be using AI for like, and they just like can't go back.
C
Yeah, I mean, I think based on that, it's kind of building on like what Optable was already doing. One thing when I was talking to, I talked to Bosco from Optable and he kind of gave me like a demo like through their dashboard platform that has like all these different types of tenants of products like based on like CVTCV or it could go into even more granular. So like maybe health and wellness segments. And he kind of did like a sample buyer exchange. And so like a buyer, he described like a buyer's brand, well, audience. And they kind of had like a $1 million budget. And so based on that, the system can come back with like publishers to find products that optimize that for like CTV package from first party, first party audience segments. They can configure pricing, packaging in which buyers can access that data. Again, as I said, a buyer submits a brief and they can come back with like brand, audience, media preference, budget and put that into Claude and then that can help set up the deal. And then the agent generates a plan showing relevant sellers, different networks, data and media costs and how campaigns can be executed. And then you can even go more granular. You can say, we only want this into ctv and so you can. The seller, the sales agent can make a plan based on that. And so like, I think just generally putting those processes into more granular things is kind of what people are looking to. And then I kind of talked about a little bit earlier about how the weather company wants to like surface discoveral data that maybe sellers aren't seeing or buyers aren't seeing in that kind of vein. So it's kind of just like an extension of that. Outside of that, it's not like anything like new or special that's like, oh, this is going to be like the new next big thing. That AI is doing. I think it's kind of just automating a lot of the processes that they're doing and using the sales agents to do it in that way.
A
So responding to an rfp, which is definitely like, kind of like a junior publisher role, can be kind of kind of manual, but also can be quite strategic depending on, you know, depending on what, what's at stake there, what they want. And then I think it also sounds like some of it's just building maybe a broader media plan overall, perhaps as a potential use case. So I'm wondering like, you know, the responding to an RFP feels a bit more like a direct deal maybe versus something that's programmatic. You know, programmatic is, you know, it's like, wait, wasn't this already all. All machine learning and using AI to automate everything? So what do you think it will be used for? Is it going to be more like direct deals or maybe helping small buyers who they wouldn't normally just be able to do the hand holding for? Or is it also about maybe putting together a PMP or something that's a little bit more on the programmatic side?
C
From what I've talked about, especially with news corps, I think they want to kind of sit in both lanes. I think some of the confusion comes from just how early it is and in the process. And even publishers are still trying to test and figure out what they want to do. So I could see it happening in both lanes where like it helps with programmatic deals and direct deals. But as again, it's just like they're still trying to test and figure out to see exactly how that works. I mean, even talking to a lot of publishers, I talked to one yesterday who I won't name just because I wasn't really talking to them about this on the record, but they were just saying like, you know, we're looking to test it, but it's so early, like maybe even two years down the line, this could even be flipped on its head. And the framework that ADCP has now is not even what we're going to be using over the next two years. So I just think that's where some of the confusion comes from. But I think in general they're looking for sales agents to work with indirect and in programmatic.
A
Okay. And it. And also could happen slowly or all at once. And maybe just to draw one more connection before we close out the podcast, which is, you know, we publishers have been in a really tough, challenging situation spot also a lot of adopts functions feel very Much like something that an AI agent could do. And I think there's, there's a good amount of outsourcing that has been going on and then maybe will continue in that direction. So what's kind of your forecast for how AI agents will disrupt these different publisher jobs and functions?
C
Yeah, I mean, I think we kind of talked about it. It could be like the manual salesperson for them. They can like handle small budget clients, maybe some more niche things that like publishers, ads, people can't necessarily pay attention to, like execute highly customized deals, package niche audience in that way. You can build manual proposals in that way. Back and forth, RSPS deal setup. So kind of the things we kind of talked about right now, I think a lot of the people, I think when a lot of people talk about AI, they always talk about how a human is always going to be in the loop. And I think that's what people are kind of saying right now is still a little too early to kind of remove the human from the process just because it's so new, you know, I don't think anywhere near replacing human capabilities as far as adopts workers.
A
But you know, okay, well we have that as a prediction. We'll see. Thanks for tuning in and we will see you next week. Till then,
D
Foreign.
A
This episode was brought to you by Zeta. Zeta's AI platform unifies paid, owned and earned media into one powerful growth engine, optimizing performance in real time and driving measurable outcomes with confidence. Don't hope for a higher ROI. Expect it with SEDA. Learn more at sedaglobal.com adexchanger. It.
Date: March 19, 2026
Host: Sarah Sluis (Editorial Director, AdExchanger)
Guests: James Hercher (Senior Editor), Andrew Bird (News Editor), Ed C. (Chief Growth Officer, Zeta)
This episode examines urgent headlines in digital marketing, focusing on two hot topics: the resurgence and controversy surrounding agency "principal media" buying practices, and early experiments with AI-powered "sell side agents" in publisher ad sales. With in-depth input from AdExchanger’s editorial team and perspectives from industry leaders, the show unpacks evolving business models, the implications for programmatic media buying, and how AI is shaping the future of advertising operations for both buyers and sellers.
How it Works:
"It isn't like the normal kind of like they're buying live inventory. It's almost more like ad network-ish, where they own a contractual right... They resell it to their clients and obviously sort of at a markup. It's a higher margin business for agencies." [03:08]
Analogy:
"They're almost like a wholesaler store... In theory." [04:02]
"It just changes the fundamental agent relationship... we are not acting as the agent, we are acting as a third party services vendor." [04:41]
"Sheep in different clothing perhaps—rebates, principal-based buying, kind of nuances involved in these practices, but seems like what hasn't changed is that agencies are taking some money off the billions of dollars in media that they're spending." -- Sarah Sluis [08:44]
"It just... they keep kind of like cloaking themselves in more complex models for this." [06:48]
Principal-based buying coexists with programmatic, but buyers are often unaware what portion of inventory is principal media.
Tensions have arisen as DSPs (like The Trade Desk) and agencies both pursue direct publisher relationships, increasing competition and scrutiny of fees.
James Hercher:
"There's a competition now because they are... trespassing in this new territory of going direct to media." [10:09]
"Fee reduction" claims come with tradeoffs: every cut means someone in the chain loses margin ("not your fat, someone else's lunch").
James Hercher:
"When you say like, oh, I'm making a big cut to people's fees—okay, like, no wonder SSPs are screaming..." [14:36]
Agencies may present "curation," "risk purchase media," or "proprietary trading media" as euphemisms for principal media, complicating transparency.
Buy-side and sell-side now need increased vigilance; calls for clear contracts and auditability rise.
Sarah Sluis:
"How do you have buyer beware when maybe the buyers don't even know what to beware of?" [15:33]
"These are the people who know what this is and they all explicitly do not opt in. So like, oh, it kind of makes you think." [18:38]
Guest: Ed C., Chief Growth Officer, Zeta
"What’s finally happened with today's technology is we can release some of those artifacts and bring ad tech and martech together... how profitable is Sarah to us?" [21:57]
"It slices bread along the way too." -- Ed C., joking about the all-in-one capabilities of unified tech. [23:57]
Guest Segment: Andrew Bird (News Editor)
"A sell side agent is basically a AI system that kind of acts like an agentic salesperson for the publisher..." [24:41]
"A salesperson can kind of upload an RFP into Claude Anthropic's tool and... map systems with relevant audience segments, and like, mix them with new and existing audience segments..." [27:08]
"I will just give a small, a small round of applause to Optimal for just being like hey, we're using Claude and like this is the interface because… so many people are like, we use all the LLMs, like they just kind of try to white label like, make it go away..." [28:22]
"I think when a lot of people talk about AI, they always talk about how a human is always going to be in the loop... I don't think anywhere near replacing human capabilities as far as ad ops workers." [34:02]
“It just changes the fundamental agent relationship… we are not acting as the agent, we are acting as a third party services vendor.”
— James Hercher, [04:41]
“There's no way to reduce fees without it hurting someone else in your supply chain.”
— James Hercher, [14:36]
“It changes from measuring a channel to doing what marketers are really responsible for: creating a profitable customer.”
— Ed C., [21:57]
“It slices bread along the way too.”
— Ed C., [23:57]
“A sell side agent is basically an AI system that acts like an agentic salesperson for the publisher.”
— Andrew Bird, [24:41]
“Right now, a lot of what's being automated are things like responding to RFPs, which junior publishers or ad ops would usually spend time on.”
— Paraphrased, Andrew Bird, [30:49]
This episode underscores the ongoing evolution (and recurring controversies) in agency-client relationships around principal media and the shifting landscape of programmatic transparency. Meanwhile, AI-driven sell side agents offer a glimpse of a more automated future for publisher ad sales, though current reality remains human-guided and experimental.
Stay tuned for further industry developments—both on principal media's future and the mainstreaming of AI-powered ad operations.