In Defense of Defensibility

0:00 Dallas and Camera owning the day. Adam drops gems, he's breaking it down how the buy side spins dollars around. Garrett steps up with that data finesse. Selling it right, no room for a guess. The mics are hot, the beats hit hard, fans throw their hands high in the yard. Deals and impressions like a verbal sprint. Breaking it down with an ad tech tint. >> The state had the defendant on tape selling with marijuana to an a female undercover police officer on several different occasions. He had no problem whatsoever hitting on the undercover officer on the audio tape trying to pick her up and saying, "Girl, you're thicker than a bowl of oatmeal."

0:45 >> >> Wow, that's incredible. Amazing. >> >> That's an HR violation as well. Hello everybody, welcome to Ad Tech Ad Talk. We're back. I'm Garrett and I represent the sell side. >> I'm Adam, I represent the buy side. I just took a 4-hour trip back from a week's vacation in Fire Island and made it just in time. That's why I'm eating. >> That's amazing, man. How was Fire Island?

1:17 >> Really good, the ocean was enormous. >> Oh, really? Like in size or like waves? >> Size and scope. >> >> Waves, yeah, we had some really We had some days of really rough waves and some really smooth days, too. It was perfect. >> Oh, it's amazing. Yeah, I know I've I've been going out to to Long Beach a lot and last weekend we were there, the waves were huge. Like >> Yeah. >> swimming was like intense.

1:45 it's really fun. It's really fun on the ocean. >> Great Odin's raven. >> Does your wife have Long Island experience? Cuz no one really goes to Long Island except people from there. I'm not from there, either. >> Yeah, so my father-in-law's from Long Beach. so we go to a very specific beach called Lido Beach, which is super duper nice. I love it. so it's just like one past Atlantic Beach, if you think about it that way. And so my father-in-law grew up Long grew up there. My wife grew up in Woodbury, Long Island.

2:11 >> There you go. >> Yeah, Long Island. >> I'm a Jersey side person, but the houses are much cheaper and better on the Long Island beaches. Quieter. >> Oh, it's cra- it's crazy. cuz what Going out to the Hamptons is a nightmare, right? It's super duper far. and it costs a bajillion dollars for a mediocre house. It's $10 million. but Long Beach, well, it is changing cuz people are starting to to wise up to this. it is I can be there in 45 minutes. So I can leave Saturday morning. I can be there. I can be there all day, stay a night, and stay overnight, and then drive back on Sunday. And it does not feel schleppy, right? It feels like a perfectly reasonable thing to do. Whereas the Hamptons, you could not do that. You can't drive to the Hamptons Saturday morning and come back on Sunday. It sucks.

2:54 >> Having a vacation house that's like a Manhattan scene is a nightmare. I can't believe people do that. >> That's the other thing that confuses me about it is like, oh, that's so that's so taxing. Like that it's just as performative out there. And Long Beach is not like that at all. Like it is super laid back. and you know, you're you're you're close to the city. You can see it. >> Do you think it's rewarding at all being interested in status? Do you ever feel Do you think those status people ever feel like, my status is high enough? I feel I feel really good about myself.

3:27 >> No. >> It just seems like you're always striving for slightly higher status. There's always another velvet rope. >> Yeah, I think my explanation for it is that they're they're not self-aware enough to know that they're striving for status. They just know that they have like some sort of insecurity. >> It's like an algorithm. >> It's like an algorithm, 100%. Their brain is just processing. They don't step back and be like, wait, do I actually give a about this? Like does this matter?

3:53 >> >> Does it matter what these people think of me? They don't do that. No, they just worry. >> That's very human, you know, it's a tribal thing, right? You get closer to the alpha status. Gorilla culture, you know, we have it in all of us. >> Yeah. Everyone's Everyone's out there trying to mog everyone else. I was talking about mogging yesterday. >> Yeah, I was at the zoo I was at the zoo with my son a couple weeks ago and and they were giving a We're at the baboons, like there's a whole seating area at the baboons, you can watch them.

4:24 This is the zoo in Brooklyn, it's great. And in front of the glass where the baboon tribe lives, a woman was giving, you know, a a zoo professional was giving guidance and she's like, this one there's this big baboon sitting on a rock. She's like, that's the alpha male, he has five babies in there and this person woman kind of hippie gray hair and the little woman goes, oh, do they mate for life? It's like, you think he's in there with a nuclear family? Five babies, like like restraining himself around the other moms? Like this is the biggest baboon in the tribe. Like >> He didn't fight another baboon to the death to become the alpha.

5:01 >> >> Five babies by by one gorilla one baboon really stuck to despite the bright red butts of all these other female baboons in the tribe. So I was really surprised she didn't know that baboons do not mate for life. >> Oh, it's very funny. I haven't seen a good baboon in a long time, man. They're crazy animals, they're tough. Like they're >> Yeah. >> When we When I was in Africa, we saw like, you know, like wild baboons out and they they scrap. Like they they fight constantly. They're like and it seems very stressful.

5:35 >> I think they thought chimps were peaceful when they first started observing them and then they found they have wars like people that every generation they have massive wars and one tribe completely wipes out another. >> I I that's 100%. I think that's like relatively unique to chimpanzees. They they go to large-scale war. there was God it happens like at scale large enough that it can span multiple countries. Like they there are huge chimp battles. >> >> It's wild out there, man.

6:04 Animals are nuts. >> I heard we're closer to bonobos. I haven't looked up this I haven't looked this up to confirm, but I heard humans are a little closer to bonobos and they're not so warlike and they don't have the alpha male structure. They're like matriarchal. I don't know. I'll come back with more info on that next week or maybe someone will >> I can't visualize a bonobo. I only have bonobo pants. >> I know the pants has totally colonized the name.

6:27 >> Yeah, they really have. Oh, and we're not supposed to talk about bonobos. Marley, you can't you can't post something like that and not explain why. Yeah. >> No comment. No comment. All right. >> there's something that the younger people know, Adam, that we don't. >> I know. >> Mhm. >> so, what have you been up to lately? >> Yeah, man. we I've been A getting into it on Twitter a lot, which is really fun. >> Oh, yeah.

6:56 >> I for the first time I got accused of posting AI content to Twitter, which I I I I have unresolved feelings about it. I'm not going to lie. >> Really? >> Cuz having some Yeah. I mean I I tried to fight back. so so yeah. >> Load-bearing? Was there a load-bearing right? >> Oh, people thought it might have been I used I think I used an m dash when I was talking, but I've been using m dashes for a really long time, so I I'm not stopping. Like it's not happening. But I even reread my post and I was like the grammar in this post is mediocre at best. So like and somebody but somebody tagged an AI like analyzer and said, "Hey, analyze this post." And it came back and said, "We think one we are 100% certain this post was written by AI."

7:39 >> 100%? >> 100%. It left no room for doubt. >> That's just being >> Great. >> And what that was my Raven. >> Like I was like, "Oh, this is just a shitty service." And I tagged the service and I was like, "This is Like, come read the post. You can tell it wasn't written by AI. There's too much wrong with it for it to have been written by AI." >> Yeah. >> >> Just write short sentences. Maybe you just wrote a long sentence.

8:00 >> I what It wasn't that long. And And the topic, I think, is actually like a really interesting one that we've been discussing a lot. It is will Well, well, it's basically what is the shape of the internet when the internet doesn't rely on Google search. Because basically, for as long as we most people remember. Now, there is a before time. There is a before time on the internet before Google search determined what websites looked like and what the websites wrote about. But, once Google search Yeah, it's it was It was chaos out there. It was It was the baboons doing war. That's what they were doing. That was >> Weren't you Weren't you in middle school or something?

8:40 >> Yeah. It was. >> You could start your You could start your internet session on Yahoo, AOL, or the other one. Microsoft >> Oh, I always hated AOL because the the dial-up service self-contained the browser. So, you would You would trigger AOL from the browser and the whole the whole login, the authentication flow was combined inside of it and I hated it. >> always for boomers. It was always for boomers. AOL was always for boomers. So, forget that. AOL was always for boomers.

9:12 >> Yeah. >> So, forget that. >> Yahoo. >> But, I remember I remember my 56K modem and I used Netscape first. My first computer, I used Netscape. and then >> would you use to find things? Yahoo or the Netscape homepage? Or would you just use Yahoo? >> Yeah. I wasn't doing a lot of searching, though. I was doing a lot of gaming back then. And then >> links. >> Yeah. >> So, anyway, if there's watching is too young, it was sort of like going to the library, except instead of the card catalog, there would be series of links and you'd click like category links like music and then there'd be types of music and you'd click and then there'd be types of bands and you could find content on bands, right? Or music. But it was like this clicking thing instead of a search thing.

9:54 >> Yeah. 100% and you would memorize domains, right? You would have domains that you would go to. Like I remember I would read Maddox and like I would religiously log in to see is there a new Maddox post and read old ones. >> Did your browsers help you? Did they have like the ability to like remind you what domains you go to a lot? They didn't, right? >> No. No. You remember you had bookmarks and you remembered. And >> Yeah.

10:19 >> I don't I remember the innovation of being able to type a search term into the URL like bar. >> Yeah. >> Like that was like oh my god. I can type I can type the search into the URL bar. That's crazy. >> by by '99, most people were starting in Google. Most people who are using the internet are start we're starting with Google. By '99, Google was only a year old. >> It's so crazy. I remember when it emerged I and I started using it.

10:45 God, I remember the first Gmail accounts. But so the point of this post though was that because that happened so early, right? If you think about '99 in terms of the internet, for all intents and purposes, the internet we all know is has been built since 1999. Like what everyone uses. Like when we when we think of websites, it's been built after 1999. Yeah. It because Google being the primary distributor of web traffic means that when you are building a website or you are building a writing an article, you care about what Google thinks about it.

11:16 Yeah. >> Cuz that is how you were going to get traffic, and behind it is a topic index just like I was describing, just invisible. >> Totally. And Google uses some magic algorithm they won't tell anyone how it works. SEO is this whole thing and we and we can talk about this for like from the perspective of the difference between SEO and AEO, it's like very strange. but the thesis here is that will that change? What does the internet look like if people writing content and making websites don't actually care about whether or not Google indexes it? Does that change what we write? Does that change how we write it? I don't know.

11:53 maybe. It might. it could it could significantly alter like how information is published on the web. And the question then becomes, well, what does that transition look like? And so I've been talking a lot on Twitter about this transition. And one of the things that I think is a dark reality that I think is going to happen is with people that are like, "Oh, AI overviews are going to kill referral traffic from Google." And I actually don't think that's going to happen cuz I think Google is smarter than to do that. I think they will artificially send traffic to publishers what they determine is just enough for the publishers to keep caring about what Google thinks and to let Google crawlers come.

12:39 I think that they could kill it. They could just AI overview everything and have no link outs anymore, but I think Google is going to play the game. >> But can I I think if they did that they would have legal trouble. >> Well, they would legal trouble? I think the publishers would just block their crawlers. I think if they if they if it really went to no Google search traffic, the publishers would just be like, "All right, you're not crawling my website anymore. Peace out."

13:02 >> Right. Because and it's because the answers are becoming from websites. Like They they don't really have the option of taking all the information from all the websites and putting it in an answer engine that doesn't send any traffic to those websites. It's not really open to them. Because at some point it becomes obvious where they got the information. >> And I I think the only way that it works is if we if we do invent a new monetization strategy. It's becoming a new economic >> Yeah.

13:32 >> Yeah. But I'm not sure that math works cuz I was I was also thinking about this. I had I had another quick cryptic tweet that I put up. Oh, I can't talk about this. It's a spoiler. >> But but you know what works was was paying the paying a newspaper on a 100 million a year works. And that seems to pay out for the AI companies at least so far. So, even though there's no market and things can't be priced, doesn't mean the right to scrape the important site can't be bought.

13:58 >> Yeah, but that's barbaric, right? That's like Coca-Cola calling up Netscape and being like, "I'll pay you a thousand dollars a month for the little player on the website." It will get more efficient than that. The AI companies will realize what content is worth something to them. and how much they're willing to pay for that. And don't get me wrong, I think that's a hugely undervalued thing. >> So, what happens to content? So, what yeah, what happens to what what content becomes valuable in this market?

14:27 What becomes more valuable? What becomes less valuable? Something like the recipes for all the known cocktails in the world. The value of that goes to zero because you just use AI for that and your cocktail website is now worthless even no matter how much money it used to produce. >> So, I got a word for you on this one and this is this is my psychopath theory of the future of the internet. The word is orthogonal.

14:50 >> Oh, no. >> So, >> Don't let those on me. >> Oh, I'm I'm doing it and you're getting some you're getting some SF buzzwords. >> >> But I love the word orthogonal. So, orthogonal means that it diverges from the expected path in a way, right? So, it is it is going in a utterly different direction than and going and is uncorrelated to the original >> math. >> And the point of this is?

15:22 >> not orthogonal. means a right angle. It means at a right angle. >> Yeah, yeah, yeah. It's at a different angle, but it's it's more than just a right angle. It is orthogonal has to do with like >> It's a different direction, right? It's a It's a different It's not along the same continuum. That's what they mean by >> And and not and not correlated. that's and that's the key is that the content that will add value is the content that is the least correlated to pre-existing content.

15:54 >> Yeah. >> and that's the simple way to think of >> right? >> Novel's good. >> Novel's good. >> Yeah, but but something theoretically could be novel. Well, is it novel if it correlates? And this is it's a gray area. I'm not sure. But the idea is so the the psychopath theory, right, is that in >> That's true. If it's just another drink recipe, if it's like, "Here's my special old-fashioned." >> Yeah, I'm not sure novelty is enough.

16:20 it needs to be weird. Like it needs to be something that the system has never seen before and still have value. >> Right. Here's a cocktail that I drink in, you know, the country of Georgia and it's delicious. Like that. >> Yeah. And Mar- Mar I'd say creativity. And I think there is a there is a deep creativity component to this, right? Like how do you create something that is really just outside of the lines of what came before it even though that's probably not super possible, but like >> Well, Mar I'd say creative and that's one body, but there's another one that's just new.

16:52 >> Yeah. >> Like something, you know, news. Something happened that I've never seen before. >> That and that I think that is the second type of content that's important. The second type of content that's important is simply aggregating real-world things that have happened. I think the news guys have a lot of pull here. They just have to survive. That's the thing. They've got to live. >> And they've got to change their model. >> Yeah, that's that's my concern is do they make it over the hump, cuz I think if they do, they have real value in the new ecosystem.

17:21 >> Do they even like shoot digital? Like I see whenever I see someone shooting in the city, the apparatus seems old-fashioned. Like why aren't they using lighter cameras and Yeah, I It looks like they got secondary stuff. >> I'm sure they spent a bajillion dollars on that 70s stuff and they're still amortizing it and they don't they don't want to spend money on anything. Especially because most Yeah. Most local news is being replaced by Yeah. >> If If there's a fire in the city, like why don't they just use like a phone?

17:52 >> >> Like what Why do you need anything better than a phone? It'd be so much faster and cheaper. >> Yeah, I I I mean, I wouldn't be surprised if there are some lean startup ones that do this. I do know that the private equity model of roll-up is you buy a local news station, you fire the top 25% of the employees by pay, right? So legacy anchors, like no matter how good they are, no matter how many people how much people like them, they're out.

18:22 and so they're really just running these operations >> I know but they just try to replace it with something people like even better. >> I I would hope so, but it doesn't seem to be the case. They seem to be like and I work with a lot of local news. Yeah, they're under so much pressure, man. >> York City newspaper. That was my first job out of college. I did 10 years at a a city weekly.

18:43 >> Oh, really? >> Yeah, still think about this all the time. Like covering the city. Like I would I would just have like a few young young reporters in every neighborhood like on call all the time. And anytime, you know, something's happened, like go cover it. Go get your phone and go cover it. Then you get video, a personality, and compile that. Like Like it could be so cheap. >> It should be so cheap. >> >> We we be getting our news with >> guys. If you see when you see a TV crew in New York, it's a van. For first, why are they driving?

19:18 >> >> Why don't they have people near there? And they have all this antiquated equipment. And and like that and our host who's like all made up and stiff. It's like that's not the way people consume content. >> You know, why there haven't been more startups in this space still eludes me. >> Yeah, cuz you can get a lot of local advertising, right? If you have if you're covering a city, you get the car dealers and the exterminators and the insurance >> Totally. They're not hard companies to call on. Now, there is not particularly good infrastructure for you to give to show ads to them.

19:50 like we've done a bad job of that. So, like the products that I sell to local news companies fill a lot of gaps. Like to put it in perspective, Google used to have something called spotlight tags for publishers, where publishers could measure conversions, right? Because most local advertisers have only ever advertised in meta or in Google. That's all they've done, right? They don't have some big sophisticated advertising strategy. They're like, " I ran some Facebook ads for my business.

20:17 I spent $5,000 a month." >> Yeah, well, they often do local TV, that's which is a big investment. >> And they do local TV. So, the sales people are calling on them. But to extend local TV to digital, they used to use spotlight tags to be like, "Okay, well well, here's your sales. Here's your sales that you drove when you ran this advertising campaign with us." And Google got rid of that. So, that's actually a big feature that we sell is conversion attribution, and that my primary consumers of it are local news companies. Because they're like, "Our sales people who are out there selling local TV want to bundle in digital, but the second they say digital, the advertisers go, "Okay, well, you got to tell me what the ROI was when I spent this money in digital."

20:53 >> Wait, how does it work? You put it on Can you do an offline conversion? >> No. >> You just do a >> That's purely a pixel that goes on a conversion page. So, like >> How do you do an auto dealer? >> User Yeah, exactly. So, user goes and fills out a lead form for an auto dealership and click submit. They fire the pixel. We say did this user see an ad on the local news website? And that's >> Take some credit.

21:17 >> And then we exactly >> Good use of last touch. >> Yeah, I mean it's here's the thing, it's not perfect, but the auto dealer doesn't want perfection. They just want something, right? Like just just tell me like did I show ads to people who ended up becoming leads for my business? Like And but it but Google took that away. So, they didn't have this. So, all of the local news teams when they're selling digital had no way to show ROAS for the past 5 years.

21:43 >> Why did Google pull that away? >> They said it was for privacy reasons. They didn't want to do conversion attribution anymore. I think it was so that they kept people inside of AdSense and AdWords because AdSense and AdWords one sec. I'm getting a a bunch of phone calls right now. I'm might be my kids' camp. I got to >> Okay. >> I'll be right back. >> I've been having a bunch of defensibility meetings like in the being the a non-technical CEO of a technical company, you can't like get too deep into what people are doing.

22:18 there's not a lot of angles to go in and talk to the tech people about what they're doing. And the best angle to go talk to them I found is defensibility. So, you don't question like the value of what they're building or the way they're building it. I'm going to with all my tech people and and giving them hypotheticals like let's say you were at a startup that has $100 million and the mission is copy chalice.

22:40 How long does it take you? How many people do you need? How much does it cost? Like scope that out. and these meetings tend to be really productive, right? It's this is the best Defensibility is the best place for the business and tech to meet. And I think if all you know, it's probably only at tech companies that things like this happen but not tech startups I think we think about defensibility a lot. But when we're talking about news and I think agencies, like they don't have to think about defensibility that much because there's not that much startup action in their in their domain, right?

23:10 They don't always have new companies, especially funded new companies, nipping at their heels. Like no one's funding a lot of no local news companies. So they don't have this problem. But it's a good problem to have, right? Cuz it forces you to think about what makes your company truly valuable and hard to copy and allows you to to, you know, charge money for what you do over over the long period of time. So these meetings have been really interesting, I think, for both for me and the tech people. It's a It's a hard question. It's a hard hypothetical to think about, right? Is money the barrier to what we've built?

23:42 Is time the barrier to what we built? Is the number of people a barrier to what we built? Is it knowledge that we have that they can get? And how would we protect that knowledge if someone quits? Like these are These are really good questions to try to justify a tech investment. And it's been really interesting work. >> I mean, it's something that most startup founders stress out about all the time cuz you see cuz there's there's this interesting double double-edged sword, right? Where if you have no competitors, that's not a good thing.

24:14 like like that's a sign that you might be doing something where your TAM is off or it's uninteresting or something like that. But if you have competitors, it's stressful because you're like, " How are we differentiated? Are they going to like take our clients? Like all that stuff?" And I And what what be what makes your business defensible has all sorts of other things wrapped up in it, right? Cuz also build versus buy decisions? Like do I build this? Do I buy this from someone else? Like And if it doesn't lend to your defensibility, building is not always the right thing to do. Like there are lots of things where it's like we should license this from someone.

24:52 Like we we have a bunch of things we're doing right now where it's like we should be building this. No, we should we should license this. >> Yeah. Yeah, I mean that's a really hard question, is right? We've made the boundaries of our business certain ways where we partner at the edge. And like what if somebody built it differently where they included that part in their core? You know, would that be harder for us to compete with? Would they be able to charge less for the combination? It's really hard to conceive.

25:14 >> Oh, it's it's it's so fluid, man. I I take solace in the fact that I don't think a lot of people actually have this much intention thinking about it. Like I when I think about a lot of the people who are running companies and I know a lot of people who have founded companies in ad tech, they do not think this deeply about things. quite often they just find some little niche and they're like, "Oh, this makes money. Like I'm going to sell this a bunch of times."

25:38 >> Yeah. >> There's a There's a Don't get me wrong, there's like a short list of companies that I do think I think think deeply about it, but I think a lot of them don't. I think a lot of them just fly by the seat of their pants. >> Sure. I mean I mean there's one way of starting a business where you see people online for something and you're like, "What are they online for?" Huh, oh, I'll make that, right?

25:57 But there's another kind where you try to have a lot of foresight of what people will be online for 5 years from now and that's brings up a whole bunch of other questions, right? And it's a bigger risk and bigger rewards and all right, that's how you That's how you get the big money. >> Oh, yeah. I mean I was having a conversation with one of my sales leads yesterday. And they were talking about, "Oh, wouldn't it be so cool if we did this, this, and this?"

26:20 And I'm like, "Yeah, it sounds cool. Why don't you go talk to 10 of our clients and ask them if they would pay money for it?" >> >> Yeah. >> Cuz there are a lot of things that sound cool that people are not going to pay any money for. >> Yeah. >> And that's that's There's a form of discipline there that is like really important when you're thinking when you're ideating new products. It's like, "Okay, there needs to be a combination of vision and like this would be cool with oh, this actually has utility."

26:51 So wait, do you have any revelations you're comfortable sharing about your defensibility discussion, or is it keep it on the hood? Cuz I love the concept of defensibility. I think it's really interesting. >> Yeah. Well, I mean, I think a lot of my insight is is the is, you know, you learn more about the minds that you work with and the way that they solve problems. Like, engineers are very like, they solve for X, right? I once heard the phrase of mind castle to describe an engineer that they they they think in terms of structure. They're always thinking about how to build structure layer by layer, and they think less about is it the best thing to build, right? I found it's it's been challenging for them to step out of solving for X. Or the reason I phrase it as a hypothetical where you are solving for X is because I find that's the way their minds work, right?

27:39 If you say if you give them a scenario, like, you're hired by a a startup with a hundred million dollars, and this is the assignment they give you, then they could solve it. Or just like you could ask them how many golf balls go in a bus, and they'll go about solving it. So, so being able to translate what I want to know into the the mind castle, the frame of other people, has been really helpful.

28:00 also, I've I have to learn a lot about statistics, which is a tough topic for me. I'm like, I'm I'm good with the concepts, terrible with the calculation, but I have to learn a lot about it to try to understand where the defensibility is in the in modeling. Mm. Let me promote my new piece. I I wrote a the last piece on my blog, and I got promoted today on Tech Sheeting, which I really appreciate. John Ebbert's the best. he's he's mentioned every piece in my two architecture series on Adam Loves Ad Tech. The fifth and final one in the series launched yesterday, and it's paired with a case study with Bayer that we released, got great endorsement from Bayer's head of head of media.

28:39 and then behind the case study was a much deeper article on the modeling, and some of the some of the statistical modes that we've built are detailed in this. The idea of modeling on the joint and why it's better than regular look-alikes. It's a long read. I I hope people read it and comment and challenge it and all. But it's Right, if you want to answer that question of what makes a modeling company defensive, you have to you have to get into some of this stuff and like what makes things hard. What can the What can AI do and and not do? And how fast does it change?

29:13 And what amount of What amount of craft do you want to go into the modeling and how much should be automated? Everything in advertising turns out to be an edge case and that's that's where the defensibility comes from, frankly. It's like something I learned right in the beginning is like all the data science problems in advertising are edge cases. They're they're weird for lack of balance in the data, very few conversions, high cardinality in the data, whole bunches of text, you know, weird variations of things aren't always the same and completely different in different verticals and different campaigns. So, it's a you know, that all those problems are what makes it so that you know, you can't build something like Chalice with with Claude or an enterprise co-pilot. but it in in those details of like what can you do How do you build this thing that Claude can't do?

30:06 are all the defensibility questions. >> It's so cool. I mean, it is it is delivering on the original promise of ad tech, right? So, when I think about what you're talking about, 10 years ago, doing this level of data science and actually differentiating yourself on data science was a fool's errand because somebody who just figured out how to deliver the media more directly and more efficiently would win. Yeah, they would beat you.

30:43 but now that's not the case anymore. >> right? When you first you work on a model for like 2 months and when you first put it in and it performed better than anything. Anything you've ever seen, but you could never you couldn't keep it up because it took so long to build the models. And the deployments and the deployment endpoints were were thin pipes that would take a lot of the value out. >> totally. I and and still like the thing that I yell at DSPs about. I'm like, DSPs, you should all adopt ARF.

31:09 Why have you not adopted ARF? Just use ARF. Let your all of your vendors integrate with you via ARF. You can do this as a DSP the same way an SSP can do it. There's no >> you hearing? What do they say? I haven't said I haven't said that to any DSPs. >> Oh, they're they're like, "Oh, yeah, we're keeping an eye on it." that's normally what I get and I'm like, you guys are all nuts. You're going to you you have to do this.

31:30 >> Yeah, PubMatic and PubMatic just went live and Magnite just announced. So now the two public SSPs are adopting and this is like, you know, the standard's a success. We have two more in the in the works. >> Yeah, I just I mean it's the right way to do things. and so a few things I want to talk about, but but before we leave defensibility, one thing that I Yeah, I mean I it should should always come up.

31:57 One thing I think about a lot, too, is the market is also big enough for multiple players. And as long as you service properly, which I think a lot of people fail at, it really makes a big difference where the tech needs to be strong enough, but then there is so much to be said in execution. And companies have this weird tendency to forget this that we're not all Google, right? Google does not have to give a about execution. Google just has it. and they can treat their clients terribly.

32:34 >> Google never would have gotten that enterprise money if it hadn't done tons of service. They poured money into service. Big clients had like Google teams of like 20 people. >> So, it's total myth that you can you can serve enterprise as as pure SaaS. And it's crazy that VCs want you to be pure SaaS. What? >> Yeah, they never did it for pubs. >> >> They didn't do it right. They didn't do it for pubs.

32:57 >> Oh my god, the largest publishers on earth get the most junior account managers. You'll have some super senior person at a $10 billion a year in revenue to publisher emailing like a support alias and getting a reply from someone who's been there for 2 years with a link to like WikiDocs. it's it's wild what they do for pubs. >> Yeah. >> >> Yeah, when we when we're pitching Chalice, the mindset was everything has to be SaaS or it's a consultancy. And I think AI my idea was AI is going to be something a third thing. It maybe maybe exists between them or perhaps is orthogonal >> >> to that entire model.

33:34 >> There we go. Might be. Might not Might be orthogonal. >> but there is it's not the same every time, so it's not SaaS. And you don't just pay for a seat, and there's always service with it because you have to truly be involved with the brand to understand what they're trying to do. So, it's you know, it's not SaaS, it's not a consultancy. Like we're building scaled software and infrastructure and recommender models and feature stores and containerized pipelines. Like it's all it's all going to scale. We're going to be able to serve thousands of clients at a time by next year, but it is not SaaS and it's not going to be.

34:07 >> Yeah. I agree. I get really fed up with the amount of lying that people do about being SaaS companies, and we all do it. Everyone wants a SaaS valuation, but the reality is for this software for enterprise software, like there has to be a component of servicing your clients and customizing for your clients. >> Yeah. >> >> It's a big idea of of the It's one of of big ideas in my two architecture series, that I'll promote one more time, is that if you want to do the same thing every time, like you pull the data, then all your all your results are based on averages. Like everyone is is within a margin of error from the average result.

34:50 And if you think about it, like this has to happen cuz you have competitors in there. You can't do much better for one than the other, right? It's Coke and Pepsi are both using it. They both have to get average results. Like you can't help Coke you know, conquest Pepsi. But when you're doing it per advertiser, if you're a per advertiser architecture, that's all you do. And you could work for both of them trying to conquest the other. And like it's but they'll have completely different models.

35:15 >> That does >> Completely different data. >> That does also take discipline, though. Because in theory, right? You could use one of the models to inform the other. Like >> You could take money from one. >> I mean that's part of the infra thing. We expect if something like that happened, someone would inspect and make sure we're not mingling data or using it. And of course we need different service teams. But anyway, we're not up to having two autos or two candy companies. Maybe they'll have >> Yet.

35:43 >> Yet. >> >> But yeah, I I love defensibility. I I think it's It's something I stress out a lot about, too. And I It's It's so funny I have two minds of it where I'm like, "Oh my god, we're doing something no one else can do." And it's so great. And then, you know, there'll be a big press release from someone else and I'll be like, "Eh, you know what? We're just going to duke it out."

36:04 Like let's just duke it out. it's it's fine. >> Yeah. You know, also it it's like investors don't want to hear this, but you're like you're you're almost like preparing for a price war. Like they Right? It's like you have to be able to win a price war. Like if there is some giant comes into the space and makes your product free, then you have to show value at some minimum price. It's a much different mindset than the SaaS mindset, which is a margin mindset, of how cheap can I make my delivery so that the fees are mostly margin, right? That's what they always do and and they make add-ons that add add margin, you know, they completely always trying to build that ratio of fee to cost, right? Margin. Trying always try to increase margin. That's the SAS mindset. That's why it gets big valuations.

36:48 And and defensibility there is kind of a nasty game of like you just you pour so much money into this thing like Slack and you get so many network effects from people using it that you know, there's lock-in. But like me and you I think and some others that we know have a mindset that there's going to be any lock-in, it's going to be total competition all the time. It can only end in price war and I'm I'm ready for that.

37:14 >> >> I I feel that way all the time. like >> Another way of saying that is that value will accrue to the end user, but the value's going to accrue to the people who do use it, right? Because everyone's going to be competing and driving the cost down, driving margin down, trying to make better and better products and that the the winners will be people who use a lot of custom AI or or custom publisher data sets.

37:39 >> Yeah, and it's good for publishers, it's good for advertisers. And this is something we talk about a lot, but it's the new architectures unlock a lot of us. We were talking about ARTF earlier. I want to give credit to Fabric. I had a cool call with the Fabric team the other day and Fabric's cool, man. Like I I don't get me wrong, I get scared of behemoths like Amazon or Google becoming the canvas that we all paint all of our businesses on, but they're forward-thinking in this. They're like, "Hey, we're going to create a tech-enabled integrations like marketplace where DSPs in our infra, modules in our infra, SSPs in our infra, everything's there and we they can just turn each other on and off with the touch of a button.

38:25 Like >> Sick. >> It's sick. It's great. It especially if everyone's in there, right? You have a new startup, like you're like, "Hey, I want to distribute this at scale to all of these big companies and start adding all of this value." like yeah, it's sick. It's really cool. >> Yeah, the only defensibility is going to be performance. >> Well, and and that's the thing, right? It could be in this architecture that's it's all on AWS. It'll be really fast.

38:52 So now it's going to be who's who's picking what vendors, who's testing what vendors. They even have testing frameworks built into it. Like what what vendors are you selecting cuz we're going to take the switching cost down to zero. And then you slap an agent on top of that to manage the vendor testing and that it gets real wild out there. >> And a measurement, right? If it if it >> Yeah, the feedback loop. >> could break measurement lock-in, like all bets are off.

39:15 There are a lot of discussions about breaking measurement lock-in with with Nielsen and DV, right? And my my view is like, it's hopeless. Like people want measurement lock-in. All the incumbents want measurement lock-in. It would take all the startups all the upstarts and startups working together in fabric to break measurement lock-in. >> And I'm not even sure what it would mean. So I I got into this with Rob Webster because So ExchangeWire posted a a really nice graphic of what the combined Nielsen Double Verify thing looks like.

39:47 >> Oh, yeah. A lot of boxes with arrows in between them. >> A lot of boxes with arrows. However, it makes sense to me. Like as in I I I replied. I was like, "As an ad tech nerd, having all of these business units in one place and having them interoperable with one another assuming that they can make that happen, that makes sense to me. You have targeting, you have measurement, you have anti-fraud, you have all of that.

40:11 And then you have an MMM and you have a custom algorithm. Like then just tie them all together. And when you go pitch a client, you say, "Can we use this for all of it?" And >> Yeah. >> And like I'm like, "Oh, yeah, technologically sound. Totally sound." And then most of the But most of the like pushback I see from people like Rob is "Yeah, but agencies don't like this." Well, yeah, like this is >> Yeah.

40:37 >> This is >> Nobody likes their lock-in, right? Like they're they're in that other world, like >> Like you could build all that If you built all that in AWS fabric, using modern technology. First of all, it'd only take like a year. >> Yeah, it wouldn't take that long. >> have to do some partners. It's like Yeah. And you could charge a tenth of what they charge. So, that's what That's what people are frustrated with. They want that to happen.

41:00 >> I have heard rumblings about grumpiness with the price of DV and IS. That's That's where I'm I When I hear grumpiness, people are like, "I'm paying 5% of my media budget to these companies." And that's steep. >> We'll see. It's up to It's up to Like I said, the the holdcos, the giant publishers, the giant advertisers are all happy with these, you know, they're they're satisfied enough with being locked in with their Nielsen and DV contracts that you know, there's a there's a chance There's a world where they don't blink and just keep paying and pay more for the combination as as they want as the whatever firm behind that wants them to. But there's another world where where they starts to break.

41:42 >> I think it might At the very least, some people will crack in. Hopefully, some people other than Augustine Fou. >> Is there Is Fou in in Fabric? No, that's I don't want to ask that question. My question is Is Is Fabric on any standards? >> I I'm not sure to be honest. I mean, I had to call them. We should ask Stef. I I would like to ask Stef about this because she's so involved in all the standards. Like she's so she's so tuned in.

42:12 I would like to push on like is this AR TF right I would assume there's a I got a TV on that I haven't heard lately I think it's still TBD but I do know every time I'm on a phone call with someone I'm like you should be using AR TF why are not just using AR they are right they're all over it Amazon's all over all the tech lab stuff that's why I was like >> >> wait don't build something else like just use the thing that everyone wants to use like every vendor is already migrated to this every vendor who takes one look at AR TF and goes oh my God my life is so much better so much better I get access to the whole bid request which is what I want like and I get sit a standardized like response where I can transform and repent lovely >> it would get a lot more coverage if ad CP hadn't jumped the shark first or whatever it's called when you come when you rush to the front of the line with with a huge amount of hype but that was going to take me into the next topic what was it oh yeah you Gareth Clayser becoming the main promoter of Google direct I get to say yeah so >> by a direct >> but Google's by a direct on my point about Google by a direct is that this is where the hyper bad CP just the hot air could just comes out of the balloon because it's Google by a directing is going to be cheaper and better and there there's no ad CP doesn't have a chance like it's a setup or like six at least three companies have to get paid on every implementation and all you get is a guarantee that's not even reserved in the ad server the way it is in Google by a direct your take is a little more is a little gentler and now you got you wrote it up you've been covered extensively and you're on ad exchanger talks this week giving your point of view which I I haven't heard yet so how does it feel to be the poster boy for Google Buyer Direct?

44:09 >> I feel underpaid. That's how it feels. I I feel like I have not been compensated for bringing this to everyone's attention. >> Have any ACP people mad at you? Cuz I'm getting phone calls. >> No, I haven't got any ACP people mad at me. In fact, one of the Scope 3 employees reposted it and gave a very generous interpretation of my take where I was like, "Now, Games' footprint is only really big in OLV and banners. So, you know, they don't really have much of a CTV footprint. So, >> Oh, yeah, they should run to CTV. Yeah.

44:41 >> And that that is exactly what she did. She was like, "As you can see, this is only banners and OLV. If you want to scale CTV agentically, you got to work with Scope 3." And I was like, "Yeah, man, maybe." If you look at my press releases, >> Why would the networks pay Scope 3? Like, they don't you know, there's there's not enough publishers, premium CTV publishers, to need a middle layer among them. >> Yeah, I mean, in in every premium CTV publisher will have some sort of seller agent And what I I still don't get here is the need for a the middle layer.

45:16 and you got a nice big argument on LinkedIn about this. And the only justification I saw for the middle layer in that argument was audit chains. >> Audit chains, yeah. >> Yeah, audit chains. To which my response is like, "What?" You don't look at those today. Like, people don't give a today about audit chains. >> But the argument I have, which was which is from, I think, the head of sales at Samba, >> Yeah. >> Yeah.

45:46 I want to say Samba's one of the ACP with the one of the one they're in that core group, right? >> No, Samba's something else. Samba's some big fancy thing. I'm not sure what it is, but I I know it's >> Well, I don't know. But anyway, he was defending the CTV >> in ACP, but they have another business, I think. >> Oh, yeah, yeah, yeah. They're ACR data company. >> Yeah, yeah, yeah. >> So, I think they're trying to do they have a big sales business and I guess they're trying to run it through ACP.

46:11 >> Yeah. >> And this idea of the audit trail being important is is completely bogus. Like, you could get a log out of anything. You don't need a middle layer just for an audit trail. Anyway, that guy >> No, you really don't. >> He got salty. He was mad at me. >> He got salty. yeah, I think that the the audit layer is is a farce, right? They they're something that only matters when people want to exert some sort of leverage over whoever they're mad at.

46:37 but the cuz the audit the ability to audit is there. Like, it is it is a doable thing right now. >> Yeah. >> and you just there's just so little appetite for it. It's it's remarkable. but I think I certainly >> to news. This is going to be like a 2-hour show. >> oh my gosh, it's 12:48. yeah, let's let's go to news. Let's do news. >> Fans might like a 2-hour show. >> Yeah, that's true.

47:01 >> Think about poor Robin Steinberg who's trying to have him run a double marathon. >> She has to keep going. It's hot out today. >> All right, this was just a funny one. This guy posted some AI look at the share of search of the international marketing experts. Now, I don't think any of these people are American and not that many Americans read this group of people. Do you know these names, Gareth? If you're a buy-side, you've have to read some of these people.

47:30 cuz especially in global CPG, their theories and work get passed around. Ritson's an English one, Sharp is Australian. And he's English, kind of Nelson Field is Australian. Do you know about this? >> I have no idea who any of these people are. >> Yeah, I would think so. they're all personalities who write you know, they they publish work on what works in advertising. >> Mhm. >> Okay, Byron Sharp's famous for a book called How Brands Grow.

48:02 Mark Ritson has been in the business forever, but most recently wrote a bunch of pieces on share of search as a as a good KPI, good general KPI for brands to watch. I like both their work. and we could talk about I could I could tell you about all the others some other time as they come up. But what I thought was funny was Ritson's response where he notes that he came in second, but calls the leader Byron Sharp. Can turn the page, Melda?

48:32 No one dares usurp the dark lord of penetration. Fortunately, Ehrenberg-Bass doesn't believe in share of search as a valid metric, so that kind of makes me top. No. All right, so you know, to an outsider, I think this is totally impenetrable, but Byron Sharp is famous for writing about what we would call incrementality as household penetration growth cuz he's done pe- a lot of CPG research. I just thought it was funny to call someone the dark lord of penetration.

49:00 >> All right, it's in this con- I mean, I'm assuming there's context. It's I mean, it's certainly interesting. So, we What does share of search mean? Does that mean like how often someone searches for your brand? >> Yeah, versus your competitors. >> Versus your competitor? >> Google gives this out in Google Trends. >> Oh. >> I think it could definitely be I mean, I don't know if there's a way to groom it, but it could definitely be distorted by a brand name that's hard to spell and people will search it more than navigate to it. Like if your brand is four letters, Target or something, like everyone could spell it opposed to Victoria's Secret, which a lot of people think is called Victoria's Secrets.

49:37 Right, you have to use You end up searching cuz you didn't get it exact. >> Oh, that's so interesting. So, actually being complicated can make it seem like your share of search is higher because people need to search. They can't just type in, you know, like like pets.com. >> Yeah. Bookmarking too might screw it up. But anyway, I guess you baseline and check. But anyway, it's kind of interesting work. We've we've done the share of search optimization. Brands have been very happy with it. Work with the SEO.

50:08 >> Yeah. >> so using So you're running a campaign, you run like a digital campaign or an OLV or a CTV campaign, and track share of search as a metric? >> Yeah, locally. And do you kind of lift up lift up where it's low automatically all the time? With the get the data from EDO. I lift where it's low all the time. And brands who have established correlations between share of search and market share, as some have, are very happy with this result cuz they know if they're lifting search, they're lifting market share.

50:42 >> Oh, that's so cool. And is there a way for you to like demonstrate incremental share of search? Like does it have to be local holdouts? Or can you like segment users somehow? >> I mean, you could leave out Yeah, you can't you can't do user any kind of user tracking in here. >> Yeah, you can just do geos where it you are and aren't doing it. But it's a cool one because you know, like you're automatic like you don't know what your competitor is doing or why they get popular in a region. So.

51:12 >> Yeah, that's awesome. >> This doesn't matter. >> Marketing people, man. They're everywhere. >> All right, next. Sorrell is back. We're even going to cover everything Sorrell says cuz he's interesting. And he's the only one who talked about holding company competition. It's over 80 now. Still going. but he made the point a point I strongly agree with is that aggregation of knowledge is not doesn't make a agency defensible. Right? This is one of the one of the points of an agency. Like an agency is an aggregation-based business.

51:44 Like it is if you're the biggest buyer, that's pretty defensible. Like if Publicis became the biggest buyer, the only way Omnicom was could compete at becoming the biggest was to consume IPG. And they became Right. That's a hard thing to do. Like they So the idea of a holding company, agency holding companies, the bigger you are, the bigger buy you aggregate, the more pricing power you have. That's the whole business. Basically, bulk discounts. You get bulk discounts, buy wholesale, sell retail.

52:13 but they also did this with knowledge, agencies traditionally, that they could at least claim that they'd aggregated so much knowledge that they could make better decisions, right? And have better ROI through that knowledge. And he's saying that's not a way they can really sell anymore. And I would say it's a caution for anybody doing any kind of pulled stack optimization is like you don't really have a data advantage. You certainly don't have one over over Google and Meta. So it's, you know, there there are ways There are other ways to get any piece of data an agency might have.

52:45 So, aggregating data, hiding knowledge, building fiefdoms, none of that's going to work well. >> Yeah, that makes perfect sense. Yeah, you have judgment. >> You have judgment, yeah. >> And execution. >> One one one AI expert told me just like if you want to imagine what AI got where AI goes, I said this to you early in in our in our show, it's like just imagine everyone has every access to every document at all all the time.

53:10 >> Yeah. >> So there's no private documents really. >> That's >> Yeah. >> And one way to think about it. >> it's it's really cool. >> But something I didn't see coming, a Digiday wrote about, was agencies now aggregating compute the same way that they aggregated knowledge and buying power. Like compute is another version of buying power. As they're saying they could buy such big contracts for compute from AWS or and Anthropic or whoever that they can sell it to clients at a lower rate than clients could get it themselves. So another way to buy wholesale and sell retail and some agency supposedly making deals for for compute which I guess could be a new business for them.

53:50 I think it'd be temporary, but >> I mean, it's it's really interesting because we always think about the the ad tech aggregation of the ad tech implementation layer, right? So, the ad tech infrastructure layer. And that's like where you implement Chalice and an SSP and a DSP and data and blah blah blah. In theory agencies could create a bigger one. Right? A bigger infrastructure layer. Where that's this is one This is one organelle in the greater cell, right?

54:21 Like the the digital organelle, but like there are other ones, right? Like there like there are there are other things. And if agencies could reimagine their businesses that way, that's just such a such a transformation though. Like such a transformation. >> But Fabric starts Fabric is where it is one place where you see like it could actually grow up. Like >> Yeah. >> I don't know. Amazon would have to convince Meta to run an Amazon or or Google for that matter.

54:50 To make it one infra. >> Yeah, make it one thing. >> It's not impossible. Microsoft may >> Microsoft, the dark horse. >> >> I love I love Microsoft. >> an agency. Yeah. >> They should. >> All right, next. We're getting up to Yes, Trade Desk. So much on Trade Desk this week. It's a festival of kicking Trade Desk when they're down. They had another not great earnings, stock down further than ever. A lot of a lot of mean comments and joking in the socials.

55:23 I don't know. Maybe it's it's different for for us. We know Jeff personally. you know, I find that it's the the it's I don't know. It's a little it's a little unseemly. People who might have kissed Jeff's ass 2 years ago now making fun of him. So, >> I completely agree. >> Yeah. >> And you know what? That's why we're covering this, cuz this is cool. This is how it should be. And I like when I think about the Trade Desk strategy going forward, I'm like, all right, the pieces are there. The pieces are set up to execute. There was some bad press along the way. It happens. Sometimes you know, rollouts don't go as smoothly as you would hope. But, now just throw the agentic thing on top of it and let it just work better than everything else.

56:04 Like let's let's just get to that point. I think this is a really nice step in that direction, cuz I've been talking to a lot of people about like the agency layer that that you were skeptical of and and my CTO was skeptical of, too. When I explained like the people were building media plan and execution layers, like to sit between agencies and advertisers and activation pools, like DSPs or social or search, he was like, "Claude's just going to do that, bro." Like He was like, like Claude Claude is just going to manage that. I was like, "Yeah, yeah, yeah, I don't know."

56:37 but this seems like a lovely step in that direction, where I was like, "Okay, we're going to start to lean into some of the agentic things for what agentic things are good at, which is humans type words, and the agent does things. >> Yeah. >> I just like that. That >> Yeah, the demo is cool. It's like, "Build me an It's like, this is what I want to reach. Build me an audience." You know, Trade Desk always won on UI. cuz it it It makes the buyer feel in control and like they're that they're driving a complicated piece of equipment that they have fine-tuned to their goals. And doing that kind of stuff through prompt is a natural, as you said, next step. And with, you know, real proof point for Trade Desk embracing large language models in the in the current wave of AI.

57:28 >> Yeah. It seems great. Like have this have an MCP, like just be the activation channel that just works. And I certainly think there's a narrative like for The Trade Desk in saying, "Hey, you know what? It's been a weird couple years, but now we're going to live a life of just making your life as easy as possible, agency. We're just going to make your lives easy as easy as we possibly can. where we're just going to hit your goals.

57:58 We're going to we're going to perform better than anything else you could run cuz we've made all of these investments, and now it's just now we're back to easy town. We don't need to talk about how the sausage is made anymore. We don't need to think about anything other than, oh, just use us. It's going to work better. And we'll fit into whatever strategy you have. Here we go. >> Yeah. >> >> I always listen to carefully to Jeff. I read the earnings call transcripts.

58:23 I've known I've known Jeff, I don't know. I met him like 15 years ago, and he seems exactly the same to me. He's someone who doesn't think what you think. He doesn't think what anyone you know thinks. He thinks something completely different than everyone you know and what you know, and he's entirely convicted on it. And, you know, I think when whether his company was 80 billion or 10 billion, you know, other people see that see that characteristic completely differently. But I actually sympathize like it's it's all the same to him. He's he's always convicted in what he thinks that's different than what other people think. It doesn't change cuz the value of his company changed. And I think that's it's something unique and commendable, and it makes you makes you want to listen to him and see what's coming next. So this is my favorite part of the last earnings call.

59:11 they asked someone asked him about the longevity of DSP, something we talk about a a lot. he says he says that DSPs will not going to be disrupted by a not AI because they are AI, right? The The one to focus on is objectivity. And he says, "You have the biggest brands in the world to trust you with your data and reassure them that you're going to preserve their data, so their insights from buying are put to use for them and exclusively for them. Most of the biggest platforms in the world are not doing that. This is really key to the the two architectures argument. And I would say we used to talk about this with advertisers, and I found they didn't really care. But what I found since then is that the results are so much better when you do it this way that it's like another reason to care. It's like not only should you withhold your data from platforms that share data for privacy reasons, it actually will perform better than than what you get when they pull it and use everyone else's data. Let me read the other part.

60:08 So, they're asking the biggest brands in the world to give them their data, they use that for their clients and especially their competitors. We view the injection of AI as not a question of disruption, but in fact the very essence of what it means to be a DSP. That said, I just want to I just want to do that. Say that I think Index is one of the biggest opportunities advertising will ever see. A huge opportunity for us we're already seeing blah blah. So, I think this this is the the closest thing to a fingerpost of of where Jeff is going, and I love it.

60:41 >> I mean, it's your talk track. >> >> You don't have to say that. >> I'll take it. No, I'll take it. >> Like I said, he doesn't think like me, he definitely never been influenced by me as hard as I've tried, so I'm sure it's a little different than what I'm thinking in ways that I can't predict. >> I've met with him, too, he's been very focused on the smart brands, as he puts it. Like, there are some brands who see this, they see that the value is there.

61:05 And I I love it. I a part of this, too, is especially like investors want to talk about the DSP model being disrupted. And I found this in a lot of old hat ad tech thinkers, right? They're really caught up on DSP, SSP, exchange, ad server. They're really caught up on on this nomenclature for ad tech companies. And I think one thing that Adgentic is going to do is it's going to do away with all of that. Like you will have execution companies and you will have data companies and that is basically what will be there. And Adgentic is the thing that will make that happen. And in a sense, it means that the only ad tech companies that are kind of left are DSP-esque companies, right?

61:51 Companies that interface with agents to make strategies happen. to make media buying >> to put it is like there's going to be two kinds of agents, people and tech. >> >> Like I think the agency and the tech just collapse. It's like you you trust a person and technology with your money and the person and the technology spend your money. So like you don't need all these modules. There's just the human side and the tech side and a bunch of trust built in and everything else is inside that.

62:24 >> Yeah. I I think I'm I'm an optimist for DSPs. I think that they got caught with their pants down with the first like iteration, sell-side decisioning. But I by no stretch of the imagination do I think that their position is so compromised like at all. I think that they still have the pole position. >> Yeah, but the context has to be Google dominance, right? Like there was no ad server competition, so they were always disintermediated on the ad server border, right? Sell-side decisioning starts to break that and the conversation we had about ad CP and CTV is a good example of where there's starting to be a a hole in the ad server piece.

63:05 Right? And if that if that could be widened and maybe a little help from DOJ, then right, the DSP can do the entire job. It could do the end-to-end job and not have to be paired with a dominant ad server on both sides. >> Oh my goodness. It's always been the dream. This was always the dream of Prebid. >> Yeah. >> Tom at Prebid used to joke about buying an ad server and open sourcing and making it part of Prebid and just saying, "Okay, the whole interface with the publisher is Prebid, the whole thing."

63:32 >> Yeah, I've heard a lot of jokes to that. I don't know if anyone's done it. It's not a compute, right? >> He tried to get Rubicon to put together the money to buy DFP and the exact team always laughed about it and Tom was like, "No, I'm serious." >> What did it What did Zeta do with Sizmek? Do they Do they offer an ad server? >> buy-side ad server. So, that's like Yeah, it's just buy-side. So, it's just like creative management and rich media creative.

63:58 >> the business the day that Sizmek was showing everybody's accounts? >> No. >> >> Right before Sizmek went out of business. Not coincidentally, there was one day where if you logged into Sizmek, every advertiser in Sizmek was available. Everyone had admin access. >> >> It was like having like a key Suddenly having a key to every locked door in the city. You could Every Every We spent the day just looking at advertiser accounts. >> Poor Mark, man.

64:28 So, I was I was trying to sell my company to Sizmek during that period of time and I So, I got to know Mark. He was a super nice guy. Mark Rieder. Really nice dude. >> Rieder. He's the guy? Isn't he Is that the PayPal guy? >> the Uber guy and the PayPal guy. Like He's like He is the guy. Yeah, he was also one of the founders of Axciom. >> Oh, yeah. >> Yeah, Mark Rieder's the man.

64:52 >> Have you Have you Have you invited him on the show? >> No, I could. I I would want to hear what he's doing. I actually So, here's the thing, I don't know like the retail media space as well as I'd like. and just I would love to hear the story of what he built at Uber. Okay, I'll have to talk to Mark. cuz Uber is like such crazy unicorn success story for ad people. Like it is >> Yeah.

65:18 >> Like Uber brings in ad guy, becomes profitable. >> >> Yeah, that's a good story. >> It's a great story. That's why I said there was a LinkedIn post the other day like why are why are all these Uber ads people? Like we were we were talking about it. Why are all these Uber ads people going to all these other companies? It's like well, cuz they can go to these other companies and be in the C-suite cuz they can tell them about the Uber story. Like we started the ads product at Uber and now Uber was profitable and look at the stock price.

65:47 >> What tech do you know what tech Uber runs in there? >> I I'm I'm assuming they built it. >> Yeah, could be. >> Like that it's such a custom experience as part of their app. I think that they built it. I mean they they they developed the ad product and they did the smart thing. They brought in an ad guy when they were building their ad product and he was like here was here was what we can sell. He brought in a >> Zax is guy.

66:10 >> Zax is guy and he was like this is what buyers will buy. And lo and behold you got an ad person at the helm of trying to add ads to your business goes well. makes sense. What a surprise. Don't bring in a a Facebook person who >> Glad that seismic thing didn't count against him. >> Oh yeah, that really stunk man. It was so stressful and it was such a hard business. It was like had a bunch of interesting parts. I think that I think seismic was in it. I think Peer 39 was in it or one of those one of the contextual companies was was part of it at the time and that spun it out. and there was there was a hodgepodge of things.

66:50 And there was like a trading desk business. >> Fuel in there yet? Or is Zeta but Rocket Fuel separately? I don't remember. >> Well, yeah they they they they broke it up into all these different parts when they got rid of it. >> Yeah. >> But yeah, it's a bunch of it went to Amazon, right? Some some Mark went to Amazon for a bit. And yeah, I It was a by side of server. God, what was the old name before Sizmek?

67:13 >> Mediamind. >> Mediamind. >> Really? >> I believe so. >> Oh, I remember Mediamind. First time seeing a Mediamind tag. >> I can't believe that came to me. Can't even always remember my kids' names. >> Yeah, I remember that was a stressful time. Like there's it's really hard when like that's happening to your business that you're in charge of. It's like this is the most >> Everybody read Adam loves ad tech. Read Gareth hates ad tech. Listen to Gareth on the AdExchanger podcast.

67:44 Look at the Chalice coverage of the Bayer case study. Ali Manning on the AdExchanger podcast also highly recommended for the Chalice story. Anything else to promote? >> I think that's a good list. Thanks for joining everybody. >> All right. See you next week. >> >> See you next week. >> >> Darn, that's deep. >> Yo, the crowd is back. The lights are dim. Front row's buzzing. It's about to