1. Podcasts
  2. The Big Story
  3. The Opportunity and Threat Of Google Buyer Direct
The Big Story

7 Aug 2026 · 40m · 15 chapters

Episode artwork

The Opportunity and Threat Of Google Buyer Direct

In short

Google’s BuyerDirect (a beta “automated direct sales” product) and how AI could reshape direct ad buying, pricing/fees, competition with programmatic, and risks like fraud/misuse. It also includes a sponsor segment on how AI trip-planning changes travel-brand marketing.

Guests

Gareth Glazer, Ad Exchanger guest columnist; former AppNexus and Magnite exec; former chair of Prebid; current co-founder of Gamera (inventory quality). He also runs “Gareth Hates Ad Tech” and a weekly podcast, “Ad Tech Ad Talk.” Hosts: Sarah Sluice (Ad Exchanger Editorial Director) and James Herscher (Senior Editor). Sponsor guest: Anna Slater, VP Global Brand Partnerships at Verve.

Key claims

Direct sales still sells 20–60% (and OLV can reach 80–100%); BuyerDirect could automate direct line-item setup across many publishers via Google’s ad server control (GAM). Likely fee: ~10% to buyers, described as “comically reasonable.” BuyerDirect may pressure programmatic to justify middleman fees. Agents should optimize testing/evaluation, not replace real-time auctions.

Notable examples

Coca-Cola-style direct deals booked in GAM with forecasting/reservations; Prebid “programmatic guaranteed” as a bridge; references to the GAM monopolistic-practice trial and possible remedies; AI travel prompts (e.g., 28% discovery prompts; AI inspiration doubled year-over-year).

Written by AI. May contain mistakes.

Chapters

0:29

Introduction to BuyerDirect

Discussion on how AI is changing direct sales and advertising, focusing on Google's BuyerDirect tool.

“On today's podcast, we are going to talk about how AI could overhaul direct sales and advertising.”
1:15

Guest Introduction and Background

Introducing Gareth Glazer and discussing his background in ad tech.

“I'm Sarah Sluice, Editorial Director of AdExchanger.”
2:19

Initial Reactions to BuyerDirect

Gareth shares his initial thoughts and experiences with the BuyerDirect product.

“Yeah, so I first heard about it in early alpha from two places.”
3:21

Impact on Direct Sales

Exploring how BuyerDirect could automate and impact direct sales for publishers.

“I'll have to add to that, that the announcement about header bidding and one of their kind of header bidding killer or whatever products we want to call it, it was a blog post about making mobile pages load faster.”
4:50

Google's Unique Position

Analyzing why Google's integrated approach may give it an edge in the ad market.

“Because we kind of hear a lot about everything going programmatic, like the direct CPMs going away.”
7:12

Potential Industry Disruption

Discussing the potential disruption BuyerDirect may cause in the advertising ecosystem.

“What about that would make it easier or more solvable of a problem compared to an independent SSP or DSP or agency trying to build their own AI agent to connect and negotiate inventory?”
13:19

Legal Considerations and Future Speculations

Speculating on the legal implications of Google's practices and the future of BuyerDirect.

“Like they're going to make that thing just work.”
14:01

The Complexities of Google Buyer Direct

Explore the implications of Google's Buyer Direct and its impact on advertising.

“So I, that was what originally called my attention to this.”
24:04

Fee Structures and Future of Programmatic Advertising

Discuss the fee structures of Google Buyer Direct and potential industry impacts.

“And when you compare 10 % to your average programmatic take rates, it's very, very low.”
28:00

The Challenge of Programmatic Buying

Learn about the limitations of lazy programmatic buying and the need for optimization.

“My, like I work, I do a podcast with someone who does custom algorithms and programmatic.”
Show all 15 chapters
29:08

AI's Role in Advertising

Explore the over-application of AI in advertising and its effective use cases.

“I like that point that will make programmatic better because it will have to compete with a different type of automated buying that also will appeal to lazy buyers.”
30:11

The Limitations of AI Agents

Understand what tasks AI agents can efficiently perform and their limitations in ad auctions.

“a process with agencies or advertisers or people that I'm trying to work with, even publishers, the number one thing that holds them back from trying new things is bandwidth.”
32:18

Concerns Over Direct Sales Automation

Discuss the risks of fraud and manipulation in automated direct sales advertising.

“So anywhere people try to apply agents to things that humans don't do, I immediately go, that's silly.”
35:15

The Future of AI in Advertising

Anticipate the future developments in AI solutions for advertising and agency practices.

“The big opportunity with AI is maybe enabling more testing, more evaluating new vendors, things that people don't have the bandwidth to do, but then it's not going to replace like a machine task.”
36:50

Building the Future of Ad Tech

Learn about the potential for open ecosystems and collaborative testing in ad tech.

“Okay, so let's think about like, I feel like I'm always wondering like what will happen next or like what milestones you might be looking out for.”
A man walking through an airport, asking his phone a question

Ask about this episode

Get the gist, explore an idea or ask about something said.

Claude

On The Big Story, what are the key points of this episode?

Searching your listening

Transcript

Automatic transcript. May contain errors.

0:04Gareth Glaser: 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.

0:21Gareth Glaser: This episode is sponsored by Verve, an ad solution that helps brands activate consumer intent in real time across platforms. As AI reshapes how people plan vacations, Verve helps travel brands show up when it counts.

0:39Gareth Glaser: Hello, everyone. On today's podcast, we are going to talk about how AI could overhaul direct sales and advertising. There are a number of initiatives out there right now for buyers and sellers to use AI to negotiate ad inventory. But according to an AdExchanger guest columnist who is on our podcast today, there's one tool that might solve this problem before anyone else can get to it. A beta product from Google dubbed BuyerDirect. We are going to talk about the opportunity this product presents to ad tech and how it might undercut competition on this week's podcast. I'm Sarah Sluice, Editorial Director of AdExchanger.

1:18Gareth Glaser: With me today is our Senior Editor, James Herscher, and our guest, Gareth Glazer, who wrote a column analyzing this product. It's been taking off on our website. He is a vet of AppNexus, Magnite, a former chair of Prebid, and the current co-founder of Gamera, an inventory quality company. He also has a blog, Gareth Hates Ad Tech, and a weekly podcast, Ad Tech Ad Talk. Welcome, Gareth.

1:45Ad Tech Ad Talk: Thank you so much for having me, Sarah. Cool.

1:48Gareth Glaser: And then one quick note before we get started, we have Programmatic I.O. New York City coming up on September 28th and 29th. This conference is the established place to talk about disruption. We've seen it before, ad tech upending the media ecosystem. I feel like that's what we're talking about today. So attend this year and you can use the code POD10 to shave an extra 10 % off your ticket price. Got to say that and give some love to our listeners. So there, now let's dive in. When did you hear about this product and why did it catch your attention?

2:24Ad Tech Ad Talk: Yeah, so I first heard about it in early alpha from two places. So just one, it was making its circles in the premium publisher, like chats, where a few select premium publishers were part of the early conversations about this. And they were chatting about it. And I heard an inkling, like there's a there's an automated direct sales product coming down the pipe. And then I saw the documentation really early. It got shared through like some exact channels where people were like, oh, here's the here's the buyer direct stuff. And it's it's coming. And then lo and behold, you know, a few months later, I saw Paul Gubbins's post on LinkedIn, which was a Mia focused and like interestingly went under the radar, by the way.

3:07Ad Tech Ad Talk: Like I've noticed that like the pickup on that was like, OK. And then even like Google's landing page for it was didn't work properly, which also I'm like, what's going on here? And but since then, right, it's it's it's a hot topic for me. It's it's very much on my mind. it's tinfoil hat wise too that's so like google by the way just to like you know pull us aside but to like slide i remember when they were getting rid of ip addresses for google analytics and they slid it into like the bottom of paragraph 14 of some like blog posts it was like okay all right like a good thing someone like found that at the bottom it's so funny there's a hundred percent a track record of this right where the google the things that google is the loudest about are like the most benign and the things that are like very quietly done are super impactful.

3:59Ad Tech Ad Talk: They absolutely play it this way. It can't be an accident.

4:03Gareth Glaser: I'll have to add to that, that the announcement about header bidding and one of their kind of header bidding killer or whatever products we want to call it, it was a blog post about making mobile pages load faster. I read it like five times and I'm like, wait, wait, this is about Ed or me.

4:26Ad Tech Ad Talk: 100%. That's how they do it. They know what they're doing. They're very clever.

4:30Gareth Glaser: Yeah. Okay. So this has been something that people are picking up at. And I'm going to kind of ask a dumb question or maybe an ad exchange-ery question because I feel like we focus a lot on programmatic. And I feel like publish, I guess I'm wondering like, How many direct sales could this even automate? Because we kind of hear a lot about everything going programmatic, like the direct CPMs going away. So how big of a deal is this for US publishers? Maybe a little bit more in EMEA if I had to guess.

5:04Ad Tech Ad Talk: I think that's a super fair assumption. So it's funny. So one of the products that my company sells is a direct sales product. So it's a product to help publishers with direct sales. So I like have my finger to the pulse of the direct sales market pretty well. Depending on the publisher, direct sales is still a thing. So once you're into local news, once you're into the top tier of publishers, you will see sell through rates of anywhere between 20 and 60 percent. With publishers who have like really high desirable OLV, so online video that will sell 80, 90, 100 percent. sometimes. It's not unheard of.

5:38Ad Tech Ad Talk: To the point where there are entire businesses built around publisher direct sales teams facilitating programmatically trafficked campaigns because they can't satisfy the size of the IO on their internal properties. So they extend them out to other. This is audience extension. So it's still a thing. Programmatic has not killed direct sales, but of course it has declined from where it once was, which is most of the market for mid to high tier and high tier publishers. So I would say that direct sales is still very much kicking. The hope for publishers when they hear all of this, and you're going to see a lot of publishers leaning into these agentic discussions and leaning into discussions about revitalizing direct sales, is that it goes back to the glory days where it's like, oh, we're selling 60, 70 percent.

6:29Ad Tech Ad Talk: Programmatic is cleaning up at the bottom. This is wonderful. We're back. We're back to where we were. Revenue is back to where it was. Traffic is back to where it was. We've turned back the clock and we've made publishing back into what it was in 2010. Will that happen? Yeah.

6:46Gareth Glaser: Yeah, that is a really interesting comparison point because, yeah, programmatic is definitely the rise of lower priced inventory. And then so the idea of a new technology actually allowing a return to higher priced inventory is obviously more appealing to publishers than what happened the first wave of automation. So let's talk specifically about the Google piece here. When Google tries to solve for this problem, it has an ad server, it has an SSP. What about that would make it easier or more solvable of a problem compared to an independent SSP or DSP or agency trying to build their own AI agent to connect and negotiate inventory?

7:37Ad Tech Ad Talk: Oh, it's such a good question. So I'm going to get a little nerdy for a minute and I'm going to go hard on this one. So my background here is when I was the chair of Prebid.js, we devoted significant resources to developing programmatic guaranteed inside of Prebid. And so programmatic guaranteed is a bridge between direct sales and programmatic transactions. It actually has nothing to do with like programmatic, like DSP, even though in theory we were building this for DSPs. But it is here is a way to set up a guaranteed campaign and execute it via programmatic pipes where there is a price, there's a volume guarantee.

8:15Ad Tech Ad Talk: There are all of these things. So I want to take a step back and think about how direct sales actually works for most publishers. A direct sold campaign, historically, as publisher shakes hands with Coca-Cola, goes, Coca-Cola goes, we're going to buy 100 ,000 impressions at$5. Publisher goes, great, lovely. They have inside of their ad server, right, which is now almost always GAM, they have something called forecasting. And forecasting tells a publisher, hey, we expect you to have a million impressions over the next 30 days. And forecasting systems also have reservations. So they'll literally go in and set up, okay, Coca-Cola of the million you expect me to have.

8:52Ad Tech Ad Talk: I'm going to book out 100 ,000 for Coca-Cola. And then the ad server, as ad impressions accrue on the publisher's website, will say, okay, we have a direct sold campaign. It's for Coca-Cola. We have to deliver 100 ,000 over the course of the month. This gets priority. It's paying us$5, but we don't really care about price because we have to deliver this. We're going to shoot it in. Everything I just described is very contained. That ad server has a tremendous amount of control and leeway in how it does that. And it is still a form of auction. So when we're thinking about what we have to do here, if we're going to introduce third parties, we have to bolt things on to that.

9:33Ad Tech Ad Talk: And that's hard. This is where it got really hard in pre-bid because if there are other systems doing bidding, doing pacing, doing all of these different things involved in delivering a campaign. And now we have to go into this controlled environment that has its own will, right? Like the, the internal ad server has its own priorities, its own way of thinking. It's Google. It's not exactly open source. Like we don't really know how this thing works. It's really hard to do. And many people have tried over the past 15 years. And ad CP is the latest attempt to slay this dragon, right? Where ad CP automating agentic direct sales is like, okay, maybe we can get it this time where when we're building things inside of Google, if it's not humans doing it, if it's agents doing it, it'll work better.

10:23Ad Tech Ad Talk: And that's kind of the thesis. That's the bet. And then of course, there's the whole selection portion, but the delivery of direct sales is there. I feel like that's such a good example too of how, you know, buyer direct, you have this sort of situation of the whole industry trying to assemble this framework and you know get a million people all in the canoe rowing together at the same time and you have to like you know you've got to like shoot an arrow through a million moving hoops and then like google could just walk in and like flip on a light switch it's like the easiest thing in the world and it's just like oh it's done like we've got it everything's in place bingo that's what got me about this one man because when you think about all the things i just said that contained ecosystem is owned by Google.

11:10Ad Tech Ad Talk: So, and the thesis here for buyer direct is a simple one. It's, oh, okay. People want a UI where they can log in, or they want to talk to an agent and go set up a direct sold line item on a hundred publishers, an act that would require a third party to log in to 100 different publishers, gamma counts, send 100 different API calls to 100 different things and build these items and then monitor them on an ongoing basis using API reporting requests to 100 different gamma counts. Whereas Google has it all in one place. It's just there. Yeah, it's slick. They've got an advantage, most certainly.

11:55Gareth Glaser: You're kind of saying too that this is something that could kind of stop a lot of these other efforts before they even start. You just brought up ADCP. I know there's a lot of publishers, especially at CAN, that were announcing that they're doing different tests of these direct agent mediated buys. So who do you think should be scared or poking out of their hole and paying attention?

12:19Ad Tech Ad Talk: Yeah, I mean, so I think that there are a few components, right? So I think that the notion that Agentic is going to automate the operations part is thin. I don't think that goes the distance. However, where this could succeed is the programmatic supply chain is historically so opaque that there's a non-zero chance that an agent doing 10 direct buys across publishers who are well known with like good stuff packaged in there could outperform a really sloppily trafficked programmatic campaign. That is perfectly possible. And I think that that's where they have to hang their hat. They have to hang their hat on, hey, the programmatic supply chain is so opaque and no one is actually doing SPO and no one actually knows what's going on with curation.

13:08Ad Tech Ad Talk: A direct sold deal will perform better by virtue of the fact that 90 % of the dollars are going to the publisher. That's a possibility. It's a possibility. However, I would say that Google being able to aggregate all the publishers in one place, if I'm an agency, there's just no way that it delivers as effectively through third-party pipes into Google specifically as it would through Google's product. Like they're going to make that thing just work. And we have precedent for this. If you ask a publisher, hey, have you ever run programmatic guaranteed in anything other than Google? They will say, no Google is the only company that was able to make PG work and they could make it work because they they owned all the different parts of it so yeah that's uh this came up during the trial too a little bit people were talking about in regards to pre-bid too like with Google's reticence to sort of allow this and you know would Google did they technically like there was a lot of sort of outsiders arguing over and misunderstanding of what like the difference between direct and programmatic direct and all these, but like, there's a lot of these terms that are just getting used in the courtroom.

14:25Ad Tech Ad Talk: But I do think it's sort of like interesting to speculate on how this could shake out, like how, you know, depending on what the remedies in that case look like is, you know, is this, could this, could the court upend this? So I, that was what originally called my attention to this. And I've gotten a lot of LinkedIn stalking from Google people since we published that article, by the way. But what really called my attention to this, and Adam and I called our podcast, right, Fruit of the Poisonous Tree, the way that Google established GAM as the ad server for 95 % of publishers was tying, was ruled to be a monopolistic practice.

15:05Ad Tech Ad Talk: The appeal of buyer direct is that it's on 95 % of publishers. So that's a really important consideration when we're like thinking about, oh, if this becomes the dominant way to run direct sold campaigns, if this becomes how buyers do it, they port it over into all of their other ad products, they port it into all of these things. The reason they're able to do that was because of a behavior that was determined to be monopolistic.

15:33Gareth Glaser: Is the buyer direct product tied to the ad server or the SSP? Like technically, what does it need to connect to?

15:41Ad Tech Ad Talk: The ad server. That's the astounding part about it because the remedies, and actually, Alan Chappell corrected me on this. There is a GAM remedy where one of the proposed remedies was you sell GAM. GAM is no longer part of Google. That would obviate all of this. But if the remedy they land on is simply, we're going to separate ADX and GAM, those are the two things that were tied together to establish the monopoly. If buyer direct exclusively lives in GAM, now GAM has used its 95 % market share to reestablish itself as something that you have to work with because all of the buyers use it to traffic all of their directs.

16:19Gareth Glaser: I feel like GAM gets a little more exciting as a standalone product if it has buyer direct.

16:25Ad Tech Ad Talk: The ad tech nerd in me, like just thinking about building stuff, it's really cool. Like it's a great product. like here buyer here's a menu of every publisher who uses scam there's no middlemen right there's no hops there's there's massively reduced latency there's tremendous transparency because the ad server really knows what's going on like of all the of all the things in the ecosystem that know where the bodies are buried google's ad server knows where the bodies are buried um and yeah it It is objectively a very compelling product to do what it does. Also from the courtroom discussion of this, where A, another confusing bit is it's still referred to as DFP throughout the – like in the trial and like in all the cut like paper.

17:18Ad Tech Ad Talk: So you always have to like in your mind, you're like, wait, that's like, okay, so it's GAM. That was like it used to be DFP. I tried to hold out saying DFP as long as I could, by the way. I really did. Nobody can help themselves. um but like they kind of got all the like the ad tech folks they had like uh you know keville james avi of keville they had like a solid from index exchange had like a couple other people and so they all had them on and they were like if dfp were available like would you be interested would you be a buyer and they're like yeah like it's looks pretty great like i think we'd have to like if they were if it were being divested and so then google came back around and all the like the final arguments were like see all these are just interested parties who want us to spin this out so that they could buy it um which i thought that's such a sneaky thing to do objectively if you're an ad tech company of course you would want to buy the thing that's on 95 percent of publishers yeah they're like well you know we responsibly would have to evaluate it i think like they're like we'd bring in we'd have to bring in some investors i'm sure like it's much bigger than we are um but yeah we all we all joked about it when we were at when we were at pre-bid one of the ideas was sort of was what if someone bought gammon open sourced it really nice um but yeah it's it is such an asset now right it is it is so deeply woven into the fabric of how publishers do their day-to-day jobs um that it's it's really a hard thing to break i'm it would take i always feel like the ad server the like such a great like sort of you know image like symbol of programmatic is how all these names have changed but they're all still there in the ad server like you could find it's still like double click it's still there's like the app nexus it's actually the one that was before app nexus is now like microsoft but they know none of them trade desk has it i'm blanking on like they're like cookie pool identifier like and no one could ever change it because like you just need the continuity i guess like you can control f for history magnite only very recently switched over from from rubicon right like it was it was rubicon

19:26Gareth Glaser: project.com and all their urls forever all those names are yeah so hard to change we're gonna take a very quick break um and then hear from our sponsor and come back and i want to find out what the fee structure is when we come back so stay tuned

19:45Gareth Glaser: Hello listeners, I'm Sarah Sluice, Editorial Director of Ad Exchanger, and I have with me today Anna Slater, VP of Global Brand Partnerships at Verve. Welcome Anna.

19:55Ad Tech Ad Talk: Thanks so much, Sarah. It's great to be here.

19:57Gareth Glaser: So we know that AI is changing how people plan trips. I've heard the anecdotes from my friends, I've done it myself. So can you give some insight into traveler behavior right now?

20:10Ad Tech Ad Talk: Yeah, absolutely. I really don't want to date myself, but I've been in the travel industry now for almost 20 years. And having worked so long in one particular industry, it's been amazing to see the overall progression. And I really feel now that the adoption of AI and this trip planning process is the most impactful transformation we've seen over that time. And looking at our own LLM data here at Verve, consumers are using it for general discovery. Around 28 % of the prompts that we're seeing relate to these types of inquiries. And this is growing. And AI use for travel inspiration has doubled in the past year, especially amongst the 25 to 34 year old audience, which has grown to around 18 % adoption.

20:51Ad Tech Ad Talk: And we're also noticing there's a shift away from keyword searches like cheap flights to Spain and treating AI now more like a travel agent, which is adapting in real time. However, there's still an element of nervousness around utilizing AI to book, you know, since the prompts that we're seeing here are certainly dominated by that discovery, the timing, the itinerary building. But I'm sure if we're going to have this conversation this time next year, that will have all changed.

21:18Gareth Glaser: So if I'm a brand and I know that AI is owning discovery, people are using it as a travel agent, but not for the booking part of the equation, where does that leave brands who are trying to reach travelers?

21:31Ad Tech Ad Talk: Good question. Travellers are using AI to navigate every stage of their holiday planning. But the biggest challenge with this new way of researching is having on brands is the pace at which it takes consumers from dreaming to consideration. And prompts get incredibly specific and in a very fast way. Someone can go from give me trip ideas straight to a shortlist without ever touching a brand's website. And we're also seeing really heavy vendor comparison with travelers using AI to sort of audit those travel providers and then benchmark prices. So by the time the brand actually shows up in the conversation, the shortlist may already be set.

22:08Ad Tech Ad Talk: So it is a very difficult moment for travel brands at this time. And then furthermore, that intent is scattered across more and more devices and platforms. But these conversations, they're carrying some of the richest signals that we have ever seen on intent. And AI is forcing the conversation to accelerate the decision making. And if they can tap in to use these signals to target, they're going to get better performance overall and have a chance at winning that conversion of the traveler's itinerary.

22:37Gareth Glaser: So you just described a very complicated, complex, scattered and rapidly changing set of consumer behavior. So what steps do you recommend to brands who are trying to capture this consumer intent at this at this moment in time?

22:55Ad Tech Ad Talk: Yeah, absolutely. And this is conversations that we're having with our clients daily at the moment. And the answer really is twofold. Firstly, travel brands need to see the intent and then also follow up and ensure that they show up in it. But there are a lot of brands that are still missing out on both right now. So on the seeing side, you know, that's LLM data, the conversational signals that we're pulling from AI chat tools. We layer them on top of actually search and zero party signals. And when you combine this, it provides a bigger, fuller picture of consumer intent. And importantly, you know, the LLM data shouldn't be seen as a replacement.

23:30Ad Tech Ad Talk: it's a real complimentary intent signal. So obviously we're encouraging brands to sort of see what's out there in market. And then on the showing up side, it means being present inside the AI conversation itself and not just hoping that the LLM happens to mention you in the brand.

23:48Gareth Glaser: Well, thank you, Anna, for those concrete steps and recommendations. And thank you to Verve for supporting Ad Exchange or podcasts. Thanks so much, Sarah.

24:04Gareth Glaser: and we're back so gareth earlier in the first half of the segment you mentioned something around 90 does that mean there's a 10 fee uh for google buyer direct like what are what are people saying and have you heard anything definitively yeah so i i can't know for certain because i'm

24:22Ad Tech Ad Talk: not a participant in buyer direct right now but everything i've heard from everyone is that going to charge a 10 % fee for this. And when you compare 10 % to your average programmatic take rates, it's very, very low. Now, there are things that cannot be done with direct sales that can be done in programmatic, right? Impression level optimization, all of the fancy things that are bringing programmatic to its next era, they cannot be done via this product. However, if you were asking me to price like, hey, what's an end to end, what's a reasonable number for an end to facilitation tool and really deal sourcing.

24:56Ad Tech Ad Talk: Because this is Google saying we are going to point agency demand and buyer demand to you publishers. 10 % is eminently reasonable, like a comically reasonable number.

25:08Gareth Glaser: And what about the other kind of bells and whistles that you get often with programmatic, but are sometimes also available with direct deals like contextual or like certain audiences, certain types of profiles, is all of that available as well?

25:23Ad Tech Ad Talk: Yeah, so some will be. So Google has integrated some data providers into GAM. That does exist. However, there's nowhere near the flexibility, right? These are Google integration cues. They are really long. Google doesn't like doing new integrations with people. Even if you look at their curation product, it's really bare bones. It can't target a lot of stuff. So I think that this will be a relatively true to form direct sales product where publishers will be able to layer stuff in if they're working with a company like mine or working with their own DMP, something like that. Publishers can layer it in.

26:00Ad Tech Ad Talk: But if you think about all of the really fancy infrastructure that exists in programmatic, right? So custom decisioning, custom targeting profiles generated through like all this crazy user syncing through live ramp and stuff like that. I'm not sure how much of that'll be in this. Now there of course is the opportunity for them to build a giant clean room here like that, that exists too. But my instinct is the initial version of it is going to just be, Hey, this is direct sales. There'll be packages of formats, packages of audiences using a very small set of data providers, and you'll buy a package, right?

26:36Ad Tech Ad Talk: You're going to buy a million impressions at this price. It works or it doesn't work. And that's it. Whereas in programmatic, right, it's much more iterative.

26:45Gareth Glaser: The name too, buyer direct, assuming it very straightforwardly means you don't need a DSP or anything else. So really the 10 % fee is the 10 % fee, right? If there's not another 10 % from something else because it's going direct to the buyer. That's the name.

27:02Ad Tech Ad Talk: That was my 90 % comment. It's going 90 % to the pub. Pretty crazy.

27:10Gareth Glaser: So, I mean, what do you think then if we're talking about, let's say that this Google product gains traction, maybe there's some other direct buying tools that gain traction. How does that change what the digital advertising environment looks like two years from now, five years from now, is that going to put pressure on companies like the trade desk with the DSP or Google's own DSP or other types of programmatic buying?

27:35Ad Tech Ad Talk: Yeah. So I think one really nice side effect of this, one positive externality of this is it will put a ton of pressure on programmatic to outperform direct sold strategies with very low middleman margins, which means that programmatic kind of needs to earn what it's doing. And I think programmatic can. So I'm a programmatic zealot. Like I just believe it. I, my company is the data set to make programmatic work better for buyers. My, like I work, I do a podcast with someone who does custom algorithms and programmatic. Like I am a believer in programmatic, but in being a believer in programmatic, I butt up against all the time.

28:16Ad Tech Ad Talk: People who are just doing programmatic lazily. It is very common who they do not want to take the next steps to optimize things the way that they should. They are just clicking buttons. And those people could be wooed away from programmatic pipes with a really slick buyer direct product. So I would say that programmatic is going to need to start to sell really well what the benefits of it actually are for buyers to justify the tremendous middleman fees. Because if you were to ask me, like, what is the most indicting part of programmatic today is how much money goes to middlemen, not to publishers, without corresponding increases in outcomes for advertisers.

29:02That is what plagues us.

29:05Ad Tech Ad Talk: and this should put pressure on that. I hope that it does.

29:09Gareth Glaser: I like that point that will make programmatic better because it will have to compete with a different type of automated buying that also will appeal to lazy buyers. So that's a great point. So another thing I want to touch on that you mentioned in your column was that AI is really useful, but it's really being over applied. I think we see that all the time at Exchange or in terms of the companies we look at, or even our own internal processes where we're looking to put AI on things. And you draw that, you know, like blockchain, like blockchain was everyone, everything was going to be blockchain for a while.

29:46Gareth Glaser: And now I'm pretty sure almost nothing is blockchain in the ad ecosystem. So that was definitely the epitome of a trend that did not happen. So tell me a little bit about what you think are the AI use cases that you're really excited about beyond maybe this direct sales piece and what ones you just would give total side eye to?

30:10Ad Tech Ad Talk: Yeah. Number one that I love is as a programmatic vendor, whenever I go through a process with agencies or advertisers or people that I'm trying to work with, even publishers, the number one thing that holds them back from trying new things is bandwidth. Like it's people have to be very discerning with their attention. And like, I can only test so many things. I think what really moves the needle for us is when the buy side agentically evaluates their various strategies. And I think that is a perfect use of agents because that is inserting agents where there were humans, right? That is, that is where agents thrive.

30:49Ad Tech Ad Talk: Agents thrive doing things that people did faster at larger scale in ways that humans could not do. Right. So that's my favorite use case is agencies now test a whole bunch of new vendors. And in the course of doing so, massively increase their results for their clients and everybody gets more money and then programmatic starts doing better and better and all that stuff just works. And I think that's wonderful. Where I give total side eye is when the original dialogue about a lot of the agentic solutions was that they were replacing programmatic and that per impression decisioning was going to go away.

31:28Ad Tech Ad Talk: And we were going to go to an entirely direct sold ecosystem. Agents are good at many things. Holding a programmatic auction is not one of them. this is because ask any exchange they'll tell you we listen to four million queries per second how would you do that with agents you wouldn't it would either be astronomically expensive or non-performant or it's just it's just not a use case for agents which are based on text right like these are llms they're large language models they operate in our image like like they're built in our words. They can only do things that are based on our thoughts. No human could auction a million ads a second.

32:14Ad Tech Ad Talk: Like we couldn't do that in our heads. We couldn't write that down on paper. It's not a thing. So anywhere people try to apply agents to things that humans don't do, I immediately go, that's silly. That doesn't make any sense. So that's where when people we're talking about programmatic going away i'm like hey i don't think that's the case i i like

32:36Gareth Glaser: that i think that's an important nuance and distinction that it's competing with something different versus being able to replace it by doing the same thing if it's if it is replacing anything it's because it's competing against programmatic and it's better something is better are served by a buyer direct, for example. So this is something I asked a recent guest as well, which is that, you know, every time there's some new digital ad thing, there's always like fraud and people feeling, figuring out how to game the system. And I just imagine like people just, you know, hacking some agent to do some, you know, weird negotiation.

33:20Gareth Glaser: Like I'm wondering what you think is the potential there, like both broadly and then I guess, you know, for Biodirect, like the fact that it's like in this ad server somehow makes me feel like there might be a few more fail safes that are already built in. But what do you think is the potential for misuse here?

33:39Ad Tech Ad Talk: Yeah. So I know that the agendic frameworks have a bunch of like governance stuff built into them. So they are thinking about this, like how do we like verify this? But at the same time, once an advertiser's creative is fired off into the ether somewhere, like the likelihood that there can be a manipulation of some system somewhere to create the appearance of something is very real. Um, and it scares me, right? Like that's, that's the sort of transaction that's very manual historically involving a lot of emails back and forth between people. Um, and then even then there's a tremendous amount of trust.

34:18Ad Tech Ad Talk: Like to get reporting back on your direct sold campaigns, you have your ad server report, which is, this is where we think we serve, but a lot of the time we don't know for certain. And then you have the reporting you get from the publisher you bought it from. That is inherently a super high trust environment. And I agree with you. I share your fears about direct sales automation being gameable if it becomes something that's more automated than it is today because of the inherent trust that exists in that ecosystem. But I do know that these guys spend a lot of time thinking about that stuff. And I would hope that any agency or any large body who's implementing something that's deciding where to send a lot of money thinks real hard about it and double checks everything before they run any checks to anyone.

35:09Gareth Glaser: Okay, so I'll sum that up. Okay, so fraud is a real concern. We're on the same page about that. The big opportunity with AI is maybe enabling more testing, more evaluating new vendors, things that people don't have the bandwidth to do, but then it's not going to replace like a machine task. It's going to replace the emails. but man I have to say I'm already sick of reading AI generated emails and strategy docs I'm like

35:36Ad Tech Ad Talk: come on come on I yell at people on my team when they send long AI responses I'm like do not do that everyone knows that's AI and you didn't write that page email everyone knows that like you saved

35:48Gareth Glaser: like 30 minutes of thinking so now I have to read a 3 ,497 word response that you know or whatever Or like, it's just, I mean, sometimes it's good analysis as well. But I feel like, yeah, I have my thoughts on it.

36:04Ad Tech Ad Talk: I totally agree. It gets me on LinkedIn too. I'll be reading something on LinkedIn. I'll get three sentences in and I'll be like, this sounds fine, but a human did not write this and I'm not going to finish reading this.

36:15Gareth Glaser: Allison and our team, I think she was saying like AI text, or maybe this was a tweet. I don't remember. But just like this idea that like sometimes AI text is very slippery. Like it's almost like so clean. You can't grasp and synthesize it. This is, of course, us as writers, not wanting to be disintermediated as writers.

36:33Ad Tech Ad Talk: But there's truth to it, though, because AI will also write things that sound reasonable when you say them. But then you think about them for a second and you're like, wait, that doesn't make sense.

36:45Gareth Glaser: Cool. So, yeah. Any closing thoughts here now? Okay, so let's think about like, I feel like I'm always wondering like what will happen next or like what milestones you might be looking out for. Like what are things that you're waiting and paying attention to that you feel like will show the direction that this AI buying will go?

37:10Ad Tech Ad Talk: Yeah, so I can't wait to see agencies adopting new solutions and sending people to frameworks where they like to test things. So I think that that's the next big frontier that's coming in ad tech is pooled areas where people add solutions and an agency's agent can go in and deploy their campaign and adopt various solutions. That will be when we have crossed the Rubicon. That will be when we are there. And there are a few people building this sort of stuff right now. Like Amazon Fabric has the bones of this. Index Exchange is building stuff like this. Kind of sounds like an app marketplace or something.

37:54Ad Tech Ad Talk: It does. It's very similar to an app marketplace, except instead of being applications, it's co-located infrastructure. That's all that it is. But you're absolutely right. It is the entire programmatic stack needs to app marketplace-ify itself. And once that happens, and that was kind of the promise of DSPs, but DSPs are individual companies with individual integration queues. this is an open standard. So just if you look at prebit, there are 400 prebit adapters. There are more prebit adapters by far than there are DSPs. It's crazy. So yeah, I think that's when we're going to know when it's there.

38:33Ad Tech Ad Talk: That's what I'm looking for. I'm keeping my eye on those open ecosystems. I think they're really important. And then I'm keeping my eye on agencies being like, yes, we will test you. We like to test things via this means. Those are the two things I'm I'm watching.

38:48Gareth Glaser: Okay. So the, um, to borrow the, uh, like the league, is it not, it's not a league of our own. It's the, if you build it, they will come. So you're looking for people to build it. And you're looking for the people. Field of dreams,

38:59Ad Tech Ad Talk: by the way.

38:59Gareth Glaser: Field of dreams, field of dreams.

39:02Ad Tech Ad Talk: Stage zero field of dreams. Yeah.

39:04Gareth Glaser: We got to build the field and we got to get the agencies to play ball. So that's a great place to lead to end. Thank you so much for your insights, Gareth. I learned a lot today and, um, we're going to be looking out for the, the next milestones.

39:16Ad Tech Ad Talk: Thank you so much, guys.

39:46Bye. Bye.

From the publisher

AI agents could replace direct sales. And a beta product tied to Google's ad server, dubbed Buyer Direct, could solve this problem better than anyone else, according to our guest Gareth Glaser, co-founder of Gamera and co-host of the podcast Ad Tech Ad Talk.

More from The Big Story

All episodes