Designated *sshole (ft. Hillary Slattery, IAB Tech Lab)

tight. [music] RTB dreams lighten up the night. Adam says pivot. Garrett says

night. Adam says pivot. Garrett says

scale. [music] Together they weave one flawless tale.

[music] That's such a good rep.

[cheering] >> Hello everyone. Hi and welcome to another episode of AdTech Ad Talk. We

are joined today by Hillary Slatterie from IAB Tech Lab, the person at the heart of the sisophian task of trying to

build things that ade companies use. But

before we go into this, I want to tell you more about Pontiac Intelligence, also known as Pontiac DSP. The flexible

Oh, Sasha, help me.

>> Flexible, lowcost, transparent DSP.

>> Low cost and transparent DSP. That's

right. And doing amazing things.

Actually, I saw they had a press release recently about being the first CTV native DSP. So, we can tack that right

native DSP. So, we can tack that right on there where it was built for CTV. And

you're welcome, Keith. Have a wonderful day, buddy. Um,

day, buddy. Um, >> hey, hey, Gareth. I'm sorry. Before we

get started, we have a very special announcement today.

>> Uh, let's bring in Marley in >> and Sasha here. It is.

You say it's Garrett's birthday.

It's Marley's birthday, too. Yeah. You

say it's Garrett's birthday.

Sasha's going to have a good time.

That's me playing guitar show. [music]

And happy birthday as well to Marley.

>> Thank you guys. It's really [cheering] exciting for me to turn 30 years old.

I'm really happy about it. [laughter]

It's uh it's been a long journey to get here, but I made it. Um

>> so Hillary, >> yes.

>> Thank you so much for joining us on the show. Um, for those who don't know,

show. Um, for those who don't know, Hillary is basically in charge of product for the IB tech lab. And I tell you what, instead of having me like stumble through explaining what the IAB

tech lab does, do you want to explain what the I what is the IAB for I start there. What is the IAB? What is the tech

there. What is the IAB? What is the tech lab? And what what products do you make?

lab? And what what products do you make?

>> Thank you. Um yeah and great question and thank you actually for getting it right because most so many people conflate the IAB with tech lab. So IAB

you can think of as like the mothership.

So they do policy research um trends those kinds of things. Their

constituency is typically like sales strategy senior leadership things like that. Techlab

that. Techlab Google uh tech lab is the like the product and engineering folks. We're

we're a global organization. So we're

the product and engineering folks that basically do like the signals underneath that like gird those policies that come out of IB. Um, so yeah, I I as much as I

wish I did run product at Techlab, I don't. I am the steward of OpenRTB and a

don't. I am the steward of OpenRTB and a lot of our foundational specifications, but Techlab actually has quite a few pillars um that I don't that I don't touch, including so like VAST isn't my

specification. Uh, OMDK isn't my

specification. Uh, OMDK isn't my specification. Um, my I call her my

specification. Um, my I call her my bosy, Jill Whitop. Uh, she is our VP of product and she sits over there. And

then we have Shaylee Singh and Tony Katzer at the tippy top.

>> Oh, amazing. Um, I didn't know Shaye was at the IB tech lab. That's awesome.

Yeah.

>> Um I read her stuff. It's great. Um

>> so super interesting. So you're the steward of open RTB.

>> That has to be like >> and before that. So going through your history before I before I start throwing, you know, stones. Um

>> Sure. Sure.

>> You worked in product at a bunch of agencies, right? A bunch of agency

agencies, right? A bunch of agency holding companies. That was that was a

holding companies. That was that was a thing. Um what was what was your life

thing. Um what was what was your life prior to the IAB Tech pod? Uh, I

actually got my start in adops on the cell side. So like my the the like

cell side. So like my the the like startup that I was at got bought by Scripts Networks Interactive and so I like I have like a running joke overeducated ass hats who've never

trafficked an ad just like I'm allergic to them. So uh when I went over to like

to them. So uh when I went over to like agency side I was I was within the WPP umbrella for 5 years. I started at triad for a year and then I moved up to like group M global investment like the

proper hold co within not at one of the aos. I got

aos. I got that view and and like actual product discipline within an agency is actually pretty

rare. Like a shocking amount of that

rare. Like a shocking amount of that stuff in my estimation is just like Excel macros and prayers. So, but like building actual like tech stacks for

like like investment leads to like do burndown reporting stuff like that like true product I got I got into it and and like also you know soothing the poor FTE who was like

sobbing because at 10:00 at night because they accidentally added you know WH to the keyword exclusion list when COVID hit and the client said get rid of what I don't want to be next to COVID

content or World Health Organization content. stuff like that. So yeah, like

content. stuff like that. So yeah, like kind of therapist.

>> Yeah, I mean I can only imagine. Um

>> yeah, >> it's it's a strange thing, right? The notion

of agency is product company. Um but

there was like an era where it was a real thing, right? Like I remember when Zaxis was building stuff. I remember

like >> when all of the holding companies were spinning up basically ad networks like that's that because that became a thing for a little while. It went out of vogue, but I think it's coming back in

now. like I think it's back with

now. like I think it's back with curation. We could talk about that. Um

curation. We could talk about that. Um

but so wonderful. So you worked at a publisher and then you worked at agencies.

>> Um and now >> you work on Open RTV.

>> Yeah. Um that is a very different um >> the most different thing in the world.

>> It's like the plumbing, the pure plumbing. For those who want Oh, let's

plumbing. For those who want Oh, let's explain what Open RTB is. I'll bet there are lots of people watching who don't know what open RTB is. Um

>> Oh wow.

>> It's like really significant. Yeah.

>> Honestly, I feel like you would give as good or better of an answer than I would uh just as a reader of Gareth. Thanks to

Adtech, but OpenRTB is basically the um the the shared language that governs like how how inventory is described in a bid request and then how bids get

returned back to those publishers programmatically.

>> Yeah. So, I think that uh the way to explain this too because so many people are talking [ __ ] about RTB and it's awful and because the whole ADC CP thing and I know that you have feelings about

that and we'll get into them. Um, but

the the birth of RTB was really just this one company has their code on lots of websites and wants to sell ad space on lots of websites, right? And this

company knew that this other company over here had relationships with lots of people who wanted to buy ad space on lots of websites. And originally they

were like, "Okay, let's do a technology integration with one another and let's like make our two systems talk to each other." And the first times that people

other." And the first times that people were doing that, they did it like peacemeal, like straight up, >> yeah, >> custom. We're going to make up our own

>> custom. We're going to make up our own way to get these things back and forth to each other. like we'll use this field, we'll use that field. And OpenRTB

was born in that ecosystem where all of these companies banded together and were like, wait, we should have a standard way of doing this. Um, but I know when I started my first company, there were

still exchanges and DSPs that were not Open RTB. Uh, that was a thing. Like I

Open RTB. Uh, that was a thing. Like I

would go to an integration discussion and they would be like, "Oh yeah, we don't use OpenRTB or we don't like support these parts of OpenRTB." and we would just be like, "This is this is horrible." [laughter]

horrible." [laughter] >> I I did a I did a talk at Programmatic IO like this fall. I did a talk about programmatic IO. It had to do with CTV,

programmatic IO. It had to do with CTV, the CTV specific stuff I'm working on.

And someone came up after and also I don't know if this is an ad tech thing or just like a human being thing, but people have hot confidently incorrect takes. Someone came up to me after my

takes. Someone came up to me after my talk and went, "Yeah, but CTV doesn't programmatic." And I was like,

programmatic." And I was like, Oh, because it's like dealsbased. And I

was like, what what rails do you think that train is writing? Like it is problematic.

>> I really think this is a thing. So,

OpenRTB at its core then is a way for one system to describe an advertising opportunity to another. And I think that the the really foundational part of that is the atomic advertising request,

right? The atomic advertising

right? The atomic advertising impression. um as CP as proposed if not

impression. um as CP as proposed if not using OpenRTB which some people say that it won't somehow or that it will somehow

replace OpenRTB would mean that instead of this atomic trading of ads different things happen and I think we've talked about on the

show before but before I go on my spiel tell me about your feelings about adp Hillary it Like

I I think Brian O Kelly is a brilliant person right?

I think though what ADCP kind of fundamentally misses is that building this stuff takes takes like

engineering product resources, right?

Open direct exists. The object model exists. Everybody has already agreed and

exists. Everybody has already agreed and understand or not like impleers right who do this for a living have already done it. So it would be relatively

done it. So it would be relatively trivial to slap ADCP on top of open direct to do the kinds of negotiations

that that ADCP is like aimed to do.

the expectation or like the the the expectation that people will instead like do a net new thing from scratch seems

wait let's pause. So what's open direct?

>> Uh what's open direct? Uh so it's it's very similar to open RTB. It's basically

like a it's a it's a way to like create direct deals with other participants in the s in the in the ecosystem, >> right? So like I I'm guessing it's like

>> right? So like I I'm guessing it's like when you say open direct like it's like programmatic guaranteed, right? Um where

PG. So the notion there would be that instead of abandoning open RTB, you have extra fields inside of open RTB requests

that allow you to delineate something as being part of a direct negotiated a direct sold thing.

>> Yes.

>> Um and I think where the departure is right and where a lot of people and and Brian walked this back by the way. I

don't know if you saw this subtly happen, but we we ripped into him for a little while about it. And uh and he and he walked back and he was like, I actually I don't think it's going to go

away. Um but programmatic guaranteed is

away. Um but programmatic guaranteed is just simply not used that widely. Like

people use it, >> but it's not common. Like it's not >> it's like >> it's primarily for digital out of home right now as I understand it.

>> Oh, interesting. I mean that makes sense because digital out of home sell things like crazy people. Um, but we had a comment, right, that I [laughter] >> here are here are a thousand ad requests

because I think a thousand people walked by this thing [laughter] like what that's crazy that's a wild thing to do.

Um, my so one of the comment was my take on ADCP is it's a workflow automation tool and I think that's that's true. So

I think that the the place for something like ADCP where it belongs is replacing stack, >> right? Yeah, totally

>> right? Yeah, totally >> did this. The MS's MS's did this. Hey,

you we're going to integrate directly with the agency's RFP system. They're

going to build it. It's going to hit us.

We're going to have these packages.

We'll automatically reply with them. And

then when somebody wants to traffic the deal, like we're automatically going to set it up in their ad server. Um

>> yeah, >> and I'm like I'm pretty sure Media Ocean plays in this space somehow. I don't

know. I still don't know what Medi Ocean does, which is pretty shameful because I [laughter] like working for a really long time.

>> Medi Ocean is the financial system. So,

they hold all the like they hold all the like transaction money like pieces.

>> Interesting.

>> Yeah, >> I'm have to dig deeper on that. I'll

Google it later. Um, [laughter]

but so I think that there's something here where the interoperability of these systems is a thing, but the granular nature of open RGB, I just don't think

we'll ever lose to anything. I think the idea of per impression decisioning is such an important and powerful innovation for making campaigns work that the notion of going back in time

and and selling bundles and then just being like, did this bundle work? Um,

that seems bad to me. Like it seems like we're gonna make something that's losing performance headtoheads now worse. I

don't think it works better.

>> Where I think Agent is a really good idea, it seems like a very straightforward use case is the media planning use case. That's a phenomenal

use case for agentic especially especially where like it just makes things faster and it doesn't necessarily replace the FTEEs that are sitting there today but it does help them like sort of

turbocharge right?

>> That makes a ton of sense. the

the notion of like we're going to blow up what's already like super entrenched and instead replace it with this net new

thing that is robot driven that will someday just not have any human in the loop seems very long term if ever right >> and I have a fun exercise for this one

so there's something called functional programming >> where instead of looking at the the composition of a system, you just look at the inputs and outputs and then you build another program that just creates

the inputs and outputs without knowing why. And if I were to look at ADC CP and

why. And if I were to look at ADC CP and look at the outputs and not care about how those outputs were arrived at, all I would see is an automated something that

magically sets up direct sold line items in my ad server, right? if I'm a publisher >> and if I'm a buyer, all I see is something that magically sends emails my creatives [laughter] out to a bunch of

publishers and [gasps] and like that's just that's you you've just automated a sales house. That's all

that it is. It's a sales house that like has some automated workflow in it. And

uh I did just I'm sure it's going to work great for that sales house.

And also like especially if you as you get into like high value CTV inventory, creative approval is a thing.

>> Yeah, >> it's it's a pain in the ass.

>> Is that why they keep showing me the same ones?

>> Oh, that's a different act. I mean, you and I both know that's actually a different problem. The like frequency

different problem. The like frequency manage management piece of it, especially in live.

>> I will say though, there's a spec for that. Our ad creative ID framework as

that. Our ad creative ID framework as >> Yeah. that we released two years ago

>> Yeah. that we released two years ago that has yet to be widely adopted. I

don't understand it. sol like goes a very long way towards solving that frequency problem because if you have like in the US the the ad registry is

ad- ID it's market based right but the so the the brand will register their media creative that media creative then

gets sent and like approved by the sellers and this is how it this is 100% of linear TV is done this way like you have to have it

I don't understand why digital hasn't hasn't taken that lesson from linear TV because there's just a lot of stuff that needs to happen in order for the ad to

get actually like rendered on the 12 foot glass screen and ASV helps a like a bunch of places in that process including frequency management.

>> Oh my gosh. So

in that registry is that maintained by the IAB where there's a list of all the creatives and like this is a whole thing. So in techlab form, no. So in the

thing. So in techlab form, no. So in the the registries are like ad ID in the US.

Uh I think it's like clear channel in the UK. There's a few like country level

the UK. There's a few like country level registries. So what techlab does though

registries. So what techlab does though is and just like almost all of our specs, right? It sits on top of that and

specs, right? It sits on top of that and it it creates a way to like p like pass you can pass the the like registered ad ID in the um in the vast which I know

you love vast so much. And then

um so it's kind of like the the like macro version of of all of the different it's it's like the macro version of all the different registries and you can pass information from those registries.

That's so cool because my $10 startup idea for many, many years has been there should be a central repo of all banner

creatives and what brands they apply to, what categories those brands are in >> because it actually is a huge yield management topic where different every

single exchange categorizes creatives differently right now, right? So, the

same brand in four or five different ad exchanges will be labeled differently.

So, you if you're if you're a buyer listening to this show, >> when you are running a campaign, each exchange is actually categorizing you different and they're comparing you to a

list of prohibitions that each publisher has set up. That means that depending on how the exchange has categorized you and how they've structured their prohibitions, you will run on some

publishers and not others. And that will be inconsistent exchange by exchange.

And this is one of many. Yeah, please

go.

>> No. And fun fact, so tech Techlab has a spec that is specifically designed to describe products and services being advertised. It's called ad product

advertised. It's called ad product taxonomy 2.0.

today like today the the way that the that the ad texts in between right are categorizing those

ads are using content taxonomy. So not

designed to do the thing. The canonical

example of why that's a shitty way of doing it is that if a website wants to if a website wants to block crypto advertising they have to block all of

finance. So that means that like MX

finance. So that means that like MX can't show up on their website.

>> You shouldn't do that.

>> The SSPs have created like their own internal rules to do that kind of mapping.

>> Yeah. And then like custom taxonomies.

Yeah. It's so dumb. They Well, they make their own taxonomies. Every time an edge case like this comes up, they go they call up their team that of like, you know, 300 people probably not in the US

and they're like, "Hey, 300 people who we make look at ads all day and tell us what brand they're in." Um,

>> now when you see crypto, you label it crypto. You don't label it finance. And

crypto. You don't label it finance. And

we're gonna make a crypto tab in our UI.

And this whole thing just seems dumb.

Like exch this shouldn't be like something exchanges do. There should be a centralized authority that says this ad is this. And like in the DSP, you

you're going to register your creative.

We're going to hit it here. We're going

to make sure that's what's actually delivering. And now we have consistent

delivering. And now we have consistent blocking. And I actually promised this

blocking. And I actually promised this would make publishers more money. I

really isolate publishers fall victim to this right now because so I'm mad to get blocks that publishers don't want blocked and 100%. It's it's an inefficiency. Um but no one's no one's

inefficiency. Um but no one's no one's ever started [applause] this company. No

one's ever done it.

>> You want to go?

>> Oh no. This is the problem with this company is it won't make any money.

>> Uh fair enough. I mean Spencer just retired, right? Let's see if he wants to

retired, right? Let's see if he wants to get back in the game.

>> I've yelled at him about this before.

This is this is one of those inefficiencies that like this should live in pre-bid or this should live in in like IAB Tech Lab would be the reasonable place actually because it ports directly. I I didn't know about

ports directly. I I didn't know about all that infrastructure for for TV ads.

That actually makes perfect sense because this is a real problem in TV ad.

People went like, "Hey, we can solve this because we just need to create a central authority of all the ads." And

uh >> yep. Like my my mental model is like if

>> yep. Like my my mental model is like if something is annoying enough to enough smart people, they're going to find a way to fix it.

>> Like I think each individual piece of that though, like every time some rule hits, right? Like crypto, finance, like

hits, right? Like crypto, finance, like that kind of thing because the like one-off nature of those things is there. There's not really a priority. But if you look at it

priority. But if you look at it holistically of like SSPs having to like you know all of the time that all of those rules like have have come up have

had needed to be made then it's like no just upgrade to like ask your DSP or like DSPs upgrade to ad product 2.0 when someone's inputting a creative in the DSP make that a required field using AD

product 2.0 Bob's your [ __ ] uncle like it just it boggles my mind.

>> So Hillary I used this word at the beginning of the show. I'm going to bring it back up. Maybe Tech Lab's life is Oh, Pontiac Intelligence. The

transparent, flexible, and lowcost DSP that also is CTV native if you're interested in CTV budgets because they do really cool stuff. And they actually work with

stuff. And they actually work with Chalice, too. I'll give Chalice Chalice

Chalice, too. I'll give Chalice Chalice a little love here as well. Um, but

Hillary, so the reason I referred to a tech lab as sisophian is because I know that I've seen that tech lab has built cool [ __ ] and people just don't use it or they ignore it and they build their

own parts of it. Why does that happen?

Why >> hubris?

Like I don't like I I I think I get put in a position a lot es especially the more contentious conversations that I've led like PLCMT

like buyer UID uh so so like for context for the uninitiated when you buy digital video online what the way that that video inventory is described in the

bidstream there is a lot of incentive to describe that inventory in a way that follows the letter of the law but not the spirit of

it right so your expectation is when you buy instream video as a buyer is one thing.

The the old specification was left wiggle room, we'll say. So, I

had to update that stuff. That was a really contentious conversation.

>> Wait, contentious with who?

>> Um, so this is actually one of my I I learned through that conversation. That

was my first really hard one. My goal is to make every single party walk away like shedding having shed one single solitary tier because if either the sell

side or the buy side is like openly weeping, something's in not balanced there. So

there. So >> like >> a a good part of my role generally is is being like the designated [ __ ] here,

>> but it's like for everybody cuz it's the the the feedback I get from a lot of the buy side constituents is like why are you shilling for the sell side? The

feedback I get from the sell side constituents is why do you only care about what buyers say? And I'm like governance matters. Everybody's in the

governance matters. Everybody's in the room. Like programmatic supply chain is

room. Like programmatic supply chain is once a month. I get between 75 and 150 people every month like show up. It's an

hour a month and like just see what we're working through. And I I actually had to put rules in place around what actually does get into the specification because it's important to note techlab

doesn't have enforcement power. So like

that so like if someone's messing around, right? Like you still need to

around, right? Like you still need to validate what's being what you're being told in that bid request. But like

that's between you and your DSP or like the publisher and the >> person sending the data.

>> Totally. Totally. But like by having it in the spec, it gives people like it gives people who want to have that conversation basically like a beating stick to be like, "No, do it the way that Techlab did because that's what

everybody agreed to."

>> Yeah. And here I'll I'll get real because I talk to agencies all the time.

Uh agencies hate buying the [ __ ] little floating things in the bottom right of screens.

>> And I don't blame them. They hate them.

They think that. And to be fair, the thing that you're talking about following the letter of the law, not the spirit of the law, was the original definition of instream, which when I hear instream, if I'm an agency, I think

YouTube, right? I think this is someone

YouTube, right? I think this is someone watching a big old video on YouTube or CNN. Um the letter of the law was well

CNN. Um the letter of the law was well if there's content after the video and the aspect ratio is this and it's in the

screen for a period of time at its full size then it counts as instream. And so

what all of these companies did, all these little floaty like [ __ ] things, what they did was they were like, "Okay, we're going to meet all of those to the

letter." But the reality is five four

letter." But the reality is five four minutes and 58 seconds of the five minute session are going to be this little video that's in the bottom right of the screen that's just cranking

through and showing as many instream ads as it possibly can, which could not be further from a YouTube experience. And

that is the PL placement discussion. And

and so when you said it was contentious, the reason I said who is I was like, "Oh, but the the people in the room who are making millions of dollars from the little floaty things like yeah, I'm sure that they were in there shouting at you

not to [laughter] >> Yeah.

>> not not to like properly label what it is." Um

is." Um >> um the buy side also though was like like I got an angry I got a few angry phone calls from the buy side too where it was like why aren't you like standing up for the tenants in your spec and I

was like bro like is that really where we're going to start like really >> the the the tenants in your spec. So

basically he was like the the feedback was like you know people are lying about this you know publishers are lying about this or SSPs are lying about this why aren't you saying that what they're doing isn't to spec and I was like well

cuz if you two things what you want me to say is is a policy thing for your DSP. It is not a

DSP. It is not a >> Yeah, if someone's lying, it's on the person lying. And because the spec, it's

person lying. And because the spec, it's actually I had a lot of respect for this one because it is very tricky to write a spec that properly covers every possible

manipulation of it. Um, like what makes something >> Oh, girl, I let that dream die a long time ago. Right.

time ago. Right.

>> This is not a thing you can do, right?

Like people are going to just stretch it as far as they possibly can. And it's

it's I you brought up two of probably the most important issues that I think that people who would like high level people don't even care about this, but people who do programmatic buying should

know about both of these. They should

know about placement because the video placement changes and I want to hear what the actual changes were because I like am not as red on this as I I'd like to be. And buyer ID, I talk about buyer

to be. And buyer ID, I talk about buyer ID and or buyer UID on the internet all the time. This is a hot button one for

the time. This is a hot button one for me because >> I love your takes on this like >> Well, this is a real one. People lie.

People lie on this one, right? People

actually I I actually think ID bridging is lying. I think if you and but Jud and

is lying. I think if you and but Jud and Tony got into this on X. Uh that was that was that was one of my conversations was Jud and Tony arguing about the IAB spec of what buyer idea

is.

>> Um >> what do you think buyer idea is Hillary?

So the >> let's go >> I'm like translating it for for audiences that are not like media buyers. So there is a very there are two

buyers. So there is a very there are two fields in there there are few fields like so open TV is 107 pages I about 85 of which >> yes

>> it's crazy it's crazy >> totally about 85 85 of those pages are dedicated to describing the inventory.

There are there are a handful of those things like those atomic pieces that are worth an awful lot more money than all the rest of the stuff in there.

>> One of the So the first one is is like what kind of video this is? That's a

really big one. Another one is what user is this? and DSPs have a very specific

is this? and DSPs have a very specific expectation for how those how that specific field needs to be filled out

and if it is not done in that very specific way they don't want it in that field by and large right like this was the argument >> real big feelings on that but that was

like the TLDDR of that >> what is happening now so so like that so buyer UID got updated like weeping and nashing of teeth.

Jesus, that was a gnarly one. That one

got updated, I want to say like last fall. What's happening now though that

fall. What's happening now though that we're seeing and there's really good um J research on this. So if you're a John's client like go it's he Chris Kane like really does a beautiful job

explaining what's happening but it's the someone either the publisher or the SSB someone with access to the stream is now saying hey instead of so I know if I if

like I send it using this path then it technically violates the law but it but there's like an end run around it where if it happens before like just one step

before I don't actually have to declare myself or and we there's still like whether or not it's reasonable is up to implementers, right? But there's still

implementers, right? But there's still some way that they can like make some argument with some plausible deniability that like that that the ID that was create that was not created in the way that the DSP expects should still be in

that field.

>> Well, let's say this is let's not speak in abstraction here. the when you say the way that the DSP expects the way

that user IDs work is DSPs have cookies >> and buyer UID is supposed to be >> the DSP knows this is their cookie in

this user that is what it's supposed to be for >> and any other way that you would get a user ID and there are lots so if you're a media buyer listening to the show know that there are lots of companies out

there that do probableist IC identity resolution where they say hey we think this user is this person who's done these things we think that this is the

case and this stands in direct opposition to deterministic identity resolution which is a cookie which is I know this user is someone who came to my

homepage and then went somewhere else because I have a cookie and people were putting probabilistic ones and still are putting probabilistic ones

into this field. That was supposed to be for a deterministic one where when it's being sold, you might be think you think you're certain you're buying someone

where the person is certain who it is and they're not actually certain. But so

okay, >> I got let's let's go.

>> I'm I'm kind of allergic to the terms probabilistic and deterministic. Like

there is no single data set to ever exist like whole data set that is fully deterministic. So if anybody says this

deterministic. So if anybody says this is 100% deterministic even if it's like an email log it right they're lying to you. Some most even might be like truly

you. Some most even might be like truly deterministic but there's going to be some in there that like that match isn't perfect.

>> Wait wait wait wait.

How could how could it in Okay. I have a list of user IDs that I have seen on my homepage.

>> The Okay, what you described fair enough there is no wiggle room there. Fair

enough. The DSP dropped a cookie in the user's browser. Fine, fair enough. Like

user's browser. Fine, fair enough. Like

that makes a ton of sense. But then I think like if you're a brand right then you can only run your web advertising in

Chrome which like what's the so the market share of Chrome the the ad spend the share of ad spend to the market share of Chrome is very

outsized because of that reliance on cookies. Lots and lots of people use

cookies. Lots and lots of people use cookieless browsers. So it's you are

cookieless browsers. So it's you are like it is not a good media strategy to only rely on cookie based >> things. Also

>> things. Also >> cookie um >> also DSPs though it's very easy like so DSP you get this user ID right you drop the cookie then you do a redirect after

in the ADM >> and if you don't see that cookie >> you can find it out but like hey that's I digress. No, no, no. But there's Well,

I digress. No, no, no. But there's Well, and the original way that people got into this because I remember when I I first heard ID bridging proposed, um I was at a large exchange at the time.

>> Yeah.

>> And I was like, won't DSPs know >> like they're going to serve an ad to the user. They're going to see that if it's

user. They're going to see that if it's not who they thought it was, they'll know this, won't they?

>> And the answer to that is, well, cookies degrade over time.

>> Yeah. And all DSP exchange integrations have a percentage built in for mismatch.

>> Yep.

>> And that became a vector for gaming, right?

>> When it actually came out, >> when it actually came out, the DSPs lost their everliving [ __ ] Like it was >> Wait, when when what actually came out?

When >> Yeah. When it came to Do you Do you know

>> Yeah. When it came to Do you Do you know what happened? How it came to light? It

what happened? How it came to light? It

was when Chrome did the 1% cookie deprecation for sandbox testing. Yeah.

>> And so what happened is the DSPs started seeing mode B like the cookie list mode in Chrome, but then they were seeing UIDs. Yeah. And that's actually how it

UIDs. Yeah. And that's actually how it came out. So it wasn't it was completely

came out. So it wasn't it was completely by accident that the DSPs figured out that it was >> God I if you see something like this though, right, let's operate under the

assumption that DSPs have mostly been user ID buying machines for the past 10 years because that's what they've been.

They don't really do much else.

Um, how does Trade not just go scorched earth and just be like, I'm I'm not buying anything from any of you anymore?

[laughter] Like, how do they not I think they have actually like a fis fiscal responsibility to do what they're doing.

And I don't mean to be a trade desk show, but like this one just seems like such a naughty one to me. Like we're

they're you're we're being lied to about maybe the only thing we care about.

[laughter] I mean has trade desk has some pretty serious like MQ policies in place that I like I get it. I mean, yeah, like look

at look at like buyer UID, PLCMT, um, what OSA did with TIDS. Like it is it is well within their right to say that what

to and I think it's actually while I don't agree with the specific move of TID that Michael Sullivan did, right? the it is absolutely within the

right? the it is absolutely within the trade right to say we're not going to buy the inventory unless you unless you have a transaction ID on it and capitalism baby like let the market fall

where they are and like good on them for making their expectations >> widely known like fair enough. Oh yeah.

It's It's one of those things where like if I am trying to get deliver a successful product for my constituents and my buyers, I saying that I want to

be able to verify what I'm buying just seems eminently reasonable. Now wait, so where did the buyer UID thing land? So

like there was all this uproar. Um so

you updated the spec. What did you update it to? Um, I updated the actual buyer UID attribute name that said like the expected default behavior is this. I

also added a three-page appendex that that Ian Trader, who was then at Basis, now at human wrote to describe to remove like any amount of plausible deniability of what exactly is meant by a cookie sync.

>> And then everything else, it says >> the the spec now says if there's if there's explicit agreement between the buyer and the seller, have at it.

Everybody's a consenting adult. Have

your fun. Otherwise,

only this very specifically like this only this very specific thing can be put here. Put everything else in extended

here. Put everything else in extended identifiers.

>> That's amazing. That's exactly how it should be. Like [clears throat] I

should be. Like [clears throat] I actually think that there's a a I really like the eids because I think that there's a a world we need to move past which is there is a viable ID. Just get

rid of that. There there are 30 different ID vendors. let every one of them go on every bid request and buyers should buy vendor by vendor not from like this overarching principle ID

field. Uh that

field. Uh that >> I completely agree and like it's something that I see a lot in the in the in [applause] the buying community as well of just gross oversimplification.

So deterministic good, probabilistic is all trash. Instream good, all other

all trash. Instream good, all other video execution is all trash. Like stop.

Like there's nuance here. Value your

inventory in the way that you seem appropriate. But like read and give it

appropriate. But like read and give it some thought.

>> Yeah, but Hillary, you've worked at an agency.

>> [laughter] >> I I sell things to agencies like and so getting on the onto the agency topic, let's talk about IPG Omnicom for a minute because I think it's directly

like it's apppropo to this conversation where the reason these definitions matter so much >> is historically and I talked about this with Jay Freriedman and now he has a new startup which I'm very excited about because it talks >> he has a new startup

>> uh he seems to be flirting with one coming out like I've seen like whispers of like this thing and >> okay >> it's in it sounds like it's going to be in Rob Webster space which is like I

don't know if you know Rob Webster with TA where they're automating that it is the proper use of agentic which is the automation of the create the media plan

process right how do you build the media plan and what Jay talked about when he came on and we were chatting was like these media plans are built at the top

but by the time they get to the bottom they are just buy this list of user IDs or buy videos that look like this as cheaply as possible.

>> And that's why and and there's not really a performance thing behind it.

There's some performance thesis at the top using some mm or using some incrementality thing or using some crazy ass thing that's going on that's running across all the media, but the person who's actually activating was just told

your job is to make this happen. And so

if they buy and it's not what they think it is, that is like the one thing that breaks that structure where, don't get me wrong, people buying campaigns and

not optimizing them to things is so horrifying to me in the world of programmatic. But still,

programmatic. But still, >> the reason this was such a big deal is this was the one bad thing that could happen in that scenario, which is, oh well, we're not actually buying the

thing we were supposed to buy. So now

we're we're actually in trouble.

[laughter] >> So um well, yes and no, right? Because

the the the bonus malice clauses that are as written in the contracts are the reason that these poor media buyers end up crying in a corner. And

>> what is that? What's bonus malice?

>> Oh, so it's like, hey, so agency contracts between especially large clients have what are called bonus malice. So, if you get me x amount of

malice. So, if you get me x amount of impressions at YCPM or less, I will pay you more. If not, I will like ding you

you more. If not, I will like ding you in some way.

>> Interesting.

>> Those are the reasons these poor kids at the end of the day are incentivized to buy like I'm not going to say shitty traffic. Like, I don't even think it's

traffic. Like, I don't even think it's all shitty. I think it's just those are

all shitty. I think it's just those are the because the whole goal here is find me the lowest possible CPM for to your point

this list of user IDs video placements that look like this.

>> Yeah.

>> You know, and so if it's but we we've seen like look at we've seen just just the existence and like scale

of say PMAX or um the Metaw Gardens, right? tells me that like buyers as long

right? tells me that like buyers as long as it's performant are actually okay without the level of transparency. But

then these four kids FTEES like the Adoffs kids at the at the agencies are like held to this incredibly high standard that is just like not the same

for the walled gardens and I don't entirely understand it.

>> Yeah. I mean I I feel like the walled gardens are safe, right? You buy what you know you're going to buy.

PMAX, like how many how many scandals has PMAX gone through?

>> Well, so PMAX is zone beast. Like I I'm thinking of Advantage Plus or whatever it is on Meta. Like Meta only has so many ad formats and they're going to tell you where they're delivering.

Pmax's major naughtinesses were extending you out to their network, right? Where you were you were you

right? Where you were you were you thought you were buying search, but you're actually buying search extension.

And search extension doesn't historically perform as well, right?

like cuz it's a lot of ARB stuff. Um

but I don't know I just feel like I feel so the this Omnicom right IPG merger >> this the the whole like discussion here

was Sir Martin Sor wrote this amazing article where he basically >> good >> he went hard um and it was very interesting >> and he basically came out and said that

one traditional media is where scale matters. So, and then like this was my

matters. So, and then like this was my big takeaway. I read I read that quote

big takeaway. I read I read that quote and I was like, "Holy [ __ ] this is crazy." He's like, "Traditional is where

crazy." He's like, "Traditional is where scale matters. What matters in digital

scale matters. What matters in digital is not scale. It's AI and your ability to activate assets and to do smart things with algorithms and stuff like that." And I was like, well, A, that's

that." And I was like, well, A, that's crazy. and B, this is like two companies

crazy. and B, this is like two companies that are basically dying because they're way overindexed on traditional. Um, just

trying to to shore up and survive because traditional is actively shrinking. Um, and they are leaning, he

shrinking. Um, and they are leaning, he called them two drunks leaning on each other next to a lamp post and which is not nice. No, it does. But then again,

not nice. No, it does. But then again, but then again, S4 Capital, I I checked the market cap. It's at about 110 billion, $110 million, which is a lot lower than it used to be.

>> Yeah. Dang.

>> It's Oh, yeah. No, it's it's it's been a rough go for S4 Capital. Um

>> Yeah.

>> But I think that it's it's this really interesting thing where what we were just talking about, the whole buy side is getting sideswiped by automation. um

where all of these agencies who used to be buying these super naive goals and Tom you say there's a lot of supply side bashing allow me to bash the buy side anyone who was going into DSPs and

setting up campaigns and then being surprised about where they ran they are adults who set up campaigns so it's uh their fault like like this is there's a

little bit of shame on people taking advantage of people on the buy side but responsibility at the end of the day lies with the buy side unless it was truly misrepresented to them which >> I could not agree with you more like I

agree with you that the that the agency like the hold model as it exists long term is not sustainable something absolutely has to give

true story however right like agencies media agencies hold sort of two really important pieces here they hold media strategy so if I'm a widget builder I

want to focus on building widgets not on like where the hell are my ads going to run, right? So really valuable service

run, right? So really valuable service there and also being agencies are the designated [ __ ] when something goes wrong. So if if like some ad shows up in

wrong. So if if like some ad shows up in some place that it shouldn't have, it's the the brand is less liable or culpable because it was oh my media buyer accidentally at the agency actually

accidentally did that. Right? I agree

with you. Especially in very high value environments like CTV, right?

Anytime I hear somebody like a brand or agency [ __ ] that about like fraud or like lack of trust in in CTV, I'm like, "Do you require OSDK to be implemented

on all of your buys?" And if the answer to that question is no, I'm like you should probably be asking your verification vendors very simple like certainly like side eye at the very

least and figure out like why because what OMDK Techlab standard I don't run it is a way for the it's it's a standardized way for the verification

vendors to say like was the TV on?

>> Yeah.

>> Like what was the last interaction with the TV? super powerful signals and it's

the TV? super powerful signals and it's standard across all of the apps and all of the OEMs. So, like it seems very straightforward to like do that and you don't have to know like the

technicalities and the nitty-gritty of it, but you damn sure should ask the question right?

>> Is it turned on? Yeah. I mean, well, and this guy's into deals too, right?

Because I know all the all the smart media buyers I know, and I know a few of them. Like, there are some media buyers

them. Like, there are some media buyers who I get on the phone with them and I'm always just like, "Ah, God, you just really know what you're doing, don't you?" Um, and

you?" Um, and >> they they set up PMPs. That's what they do. They were like, "The way I'm going

do. They were like, "The way I'm going to the way I'm going to cut through all of this noise is I'm going to go set up PMPs with people with with the sellers themselves, and then I'm going to know what I'm buying." Um,

>> yep.

>> And they tweet about this all the time.

One of my boys, Eric Tilbury, tweets all the time. If you're buying open market

the time. If you're buying open market CTV, it's on you. like [laughter] like like there you can you could yell about fraud at Open Market CTV all you want,

but everyone knows who has ever run any sort of report that it it's madness in there. It's just madness.

there. It's just madness.

>> I mean, okay, >> and it's not [laughter] hard to set up PMPs to buy actual things. And

>> it's the DSP's fault, too. Don't get me wrong, but like but just don't do it.

Everyone knows that.

>> Couple things. The I agree. Yes. And the

if you're buying CTV, so I just went to public comment for a deal API.

>> Yeah, >> that took me a year. Shockingly big

feelings. What that deal API does, and it's short, right? It's it literally just outlines the basic tenants of the deal. It goes from origin to receiving,

deal. It goes from origin to receiving, typically SSP to DSP, and it's like, is the inventory on this deal going to going to change as it's in flight? Um,

who packaged it? Who sold it? It

actually like names the parties involved so you can pretty easily see like where things are going sideways. The um the

the deal piece though especially in CTV if you accept a deal ID and like there are some companies that are doing some cool stuff around this like Mobian's doing some cool stuff. Chalice is doing some cool stuff like lots of AI based

companies about like whether or not a deal applies.

I if you are accepting only a deal ID and not an app bundle ID and not absolutely requiring an app bundle ID, you could think that you're buying from

a perfectly legitimate seller and it's full fraud because all they have to do is get that app bundle ID. Like it's

really easy to game the system if you're not requiring a genuine app bundle ID.

>> I agree with you. That's a nice double check for CTV specifically because even a lot of the big trustworthy CTV players run networks where they bundle garbage in with their good stuff. Uh because

they can arb the [ __ ] out of it and they can sell the $2 CTV at their $20 direct sold price. Um, but

sold price. Um, but >> I mean I'm not gonna I'm not gonna weigh in on the value of the inventory, but I will say if I was in the publisher's position,

like if you if I hear a lot of buyers say they like want show title on every single request, that is not reasonable.

Like that is a completely unreasonable ask because if you've bought >> if if you're paying a lower CPM for sports ball,

>> fine, great. You're not gonna get this is game seven of the World Series or this is football or this is whatever.

You're gonna get the level of transparency that says sports ball.

>> I I think there's a a distinction there which is normally when buyers are complaining for showle transparency, it's because they cannot get it at all.

Um because right now the ways to get show transparency are really janky and like hacky.

>> Fair enough. However, like cap capitalism is a thing, right? If the

buyers continue to buy without that show level transparency, >> I know they're not going to get it because because the sellers have no incentive to give it because then they lo then they lose the ability to bundle.

But even though Iris TV solved this very well. Um, however, I do want to point

well. Um, however, I do want to point out >> and I introduced an I introduced extended content identifiers in OpenRTV last year, early this year.

>> That's what I was going to say. the at

the end of the day >> all of the things that people do with deals right outside of the chalice implementation because the chalice implementation is unique it's the imple it's moving buyside logic into the sell

side because I always thought about this when like when we were working on we like when our my company built our first deal ID implementation where we had to support deals for our buyers >> and when the next exchange I worked at

we were working on the deal engine >> they're targeting systems right that's that's like a set of the targeting stuff in a DSP. That's what a deal ID tool is.

It is like a component of what the DSP does. It is therefore redundant. Um, and

does. It is therefore redundant. Um, and

it just is like you you can't do anything with a deal ID tool that you can't do with DSP like by definition.

until index rolled out their sellside decisioning thing where they were like, "Oh, wait. We can open up the

"Oh, wait. We can open up the infrastructure to super flexible algorithm companies in ways that DSPs

seem to be refusing to do so." And this is where deals being a mechanism for implementing that is so strange because it still is the first thing kind of,

right? It's still like a subset of

right? It's still like a subset of targeting like it's still a targeting tool. However, it's no longer a like

tool. However, it's no longer a like it's it's no longer within the scope of what a DSP can do. Is now past it. And

in being past it, it evolves what the exchanges like the exchange now can has capabilities for targeting that the DSP does not.

>> Uh that's a weird is that still I don't know. I don't know if that's an exchange

know. I don't know if that's an exchange anymore. I think it might be something

anymore. I think it might be something else. I think we might

else. I think we might >> I think it's I mean I think if you look at and this is a perfect use case for ATCP right like it's still going to sit on open RTB or programmatic rails tech lab rails

>> still open RTB guts so Sony guts >> totally totally the the deal API that I just that I just released is also literally just naming who sold it who

packaged it and who the curator was just having those three names I think will go a long way toward because there's also as we talked about before a lot of incentive to like pull whatever shannery

you can get away with, right? So like

let's at least require the parties involved who had their finger in the pie to be named, right? First first step the the I I imagine future iterations of the

deal API is going to have a lot more like validation discoverability stuff like that, right? This one though, it took me I've been working on this deal API. This is my third running start

API. This is my third running start since being at Techlab. It took me a full year to get this iteration of the deal API and like if you go look at it it's it's simple like there is not much

to it because it's meant to sit on top of those existing systems like it's to make sure that like it doesn't conflict with uh you know bid switches deal API deal API we kept it really really simple

and really really top level >> I I so I have this weird aversion to trying to standardize deals let me tell you why I think deals are a band-aid And

I think that deals >> they were always a hack for sure.

>> And but and to take that logic out further. Yeah. Yeah. Yeah. Okay. We'll

further. Yeah. Yeah. Yeah. Okay. We'll

do postmortem after I finish this thought. Um to take the logic out

thought. Um to take the logic out further, the end of deals will come when DSPs can accept the raw data that's used

to compose the deals.

And the reason deals are implemented now is because DSPs are inflexible in the fields that they accept.

And as they become more flexible, people might actually start to use your specs, Hillary.

>> And I think that's broken dream that's the end game that so I think facilit So deals really are used widely today because people do not use OpenRTB properly. I really think that's a thing

properly. I really think that's a thing and I think there's a stringed tug on there where deals could go away if buyers start to go, "Oh, we're going to

really lean into Open RTB and we're going to like use it to its full extent and we're going to listen to EXT fields and we're going to like like build verification for the contents of fields,

which is a thing that's coming like Clean Tap does this, Shilin's company does this. They're just like, "Hey,

does this. They're just like, "Hey, let's just look at the contents of the OpenRTB fields and just verify them."

Um, >> I I love that. Is that hope? I feel

>> that's hope. I think it's I It's the most hopeful time ever. Let's do

postpart. I'm sorry, guys. I've been

since 12:55, but you know, we're This is This is going to be a Nerd Out episode.

I knew it was these joys I get to talk about.

>> President Barack Obama and Kahindi Wy to unveil the portrait.

All right, postmortem. What have people been posting on social media? People say

the silliest things.

>> Oh, Adam liked this one. This one

tweaked Adam. This one just triggered PTSD in me. Um, so the post is there's no such thing as serverside header bidding. It's not header bidding. If

bidding. It's not header bidding. If

there's no bidding in the bloody header, it's definitely a bread. Uh, serverside

header bidding is just the industry being too lazy to admit this is the same thing. we were all building 14 years

thing. we were all building 14 years ago. An ad server [ __ ] out to a

ago. An ad server [ __ ] out to a bunch of DS SSPS picking a winner and handing it back. Um just 2011 with a fresh coat of adtech BS paint. The only

reason this term exists is because Gamo refuses to behave like a modern ad server. So we've ended up with this

server. So we've ended up with this Frankenstein setup where pre-bit does the auction and Gam pretends it's still in charge. Call it call it whatever you

in charge. Call it call it whatever you want, but it's not header anything.

This person is technically correct, but the this this is what we were saying in 2015. So like this is a very late

2015. So like this is a very late reflection on this.

um header bidding.

>> Don't we see that? Like don't we see that so much in adtech where like like nobody has an original idea but somebody like it'll occur to somebody >> just wait long enough.

>> Yeah. And they're like >> bring it back.

>> Why didn't we do this? I'm like we [ __ ] did like here you go. Like

you're welcome.

>> Like I'm like literally this guy is saying what if we made jeans that were really wide at the bottom. [laughter]

Like what if we did that?

And um Huh?

>> So basically [laughter] he is technically correct though the reason we call it header bidding is because the initial implementations the

code had to sit in the header which header was we're talking about the HTML head of a web page >> and it needed to sit there because it

needed to load before the ad server did >> and the original header binning implementations were actually blocking JavaScript which was [ __ ] incred insane. Um, but like literally the

insane. Um, but like literally the header bidding code would run in the header. That's why they called it header

header. That's why they called it header bidding because it sat in the head of the page and it bid. And this guy is saying, "Oh, well, when it's server side, there is no head." And that's true. But the header bidding moniker of

true. But the header bidding moniker of an auction system that sits outside of the ad server and injects bids into the ad server just stuck like that. It just

became header bidding is any auctioning system that sits outside of the ad server and shoves bids into it. Um, and

then he talks about GAM and it's like, well, yeah, yeah, we had to build header bidding because of GAM. It exists

literally because of GAM. Um, and it was a beautiful hack as displayed in the anti- like the monopoly suit that we won

against Google like [laughter] like this is this is decided. We have

done this. This is a war that we've won.

So the the reflection on GAM is also late because holy [ __ ] we beat them.

Like they there were there were internal memos inside of Google saying, "Man, it would have been sure would have been great if we killed this thing in the crib, wouldn't it?" Like,

>> yeah, >> like literally they wrote this in emails. They were like, "This we we

emails. They were like, "This we we should have stomped this out before it became such a problem." [laughter]

So good takes Sparky.

>> Yeah, that was a good one. It's just

like What else we have?

This one's a two-parter, so I'll scroll down when you read this part.

>> Ah, all right, Hillary, you want to read it?

>> I can't. It's like too small. I'm old.

>> As Blank points out, the situation is ex is insane. The problem here is that

is insane. The problem here is that nobody has an incentive to really do anything for the platforms. The clue is in the first sentence. Meta now earns more selling scam ads. There's no

penalty apart from mild social displeasure, but there is money for the advertisers. The juice is not worth the

advertisers. The juice is not worth the squeeze. The figures are huge in total,

squeeze. The figures are huge in total, but when you split amongst individual advertisers, likely to be relatively small as they are not the drivers of the advert ad platform's ad spend. Uh the

biggest losers here are roughly 80% of metas and Google's search ad spend comes frommemes, including Chinese e-commerce spenders. They have

little time, resources, power, any power to remedy this. Three implications.

There's no use trying to persuade the platforms to financially benefit from them. No signs will change. There's no

them. No signs will change. There's no

point in targeting large advertisers.

There's little economic on the in an individual level. Uh probably the only

individual level. Uh probably the only remedy is via government action the grounds the ads are funding criminal activity. Um okay so the point of this

activity. Um okay so the point of this post is discussing the $16 billion a year that Facebook makes from scam

advertisers.

Um, I guess the theory is thatmemes are losing because it pushes up pricing.

Like that's >> I don't know if I agree with that police work, >> but like >> who who like what how how are being hurt here?

>> Yeah. Is that is that like I'm I'm I'm I'm not like putting two and two together on like >> I >> why advertisers would care about this

other than my prices are higher cuz there are scam ads in Google >> but I don't even think that's necessarily true be and like I I would need to see some pretty ser that

quantified in a pretty big way to like take that and then also like I think going back to you so AI for as much of an existential threat to like open web publishing as it is, right? I think

we'll do a really good job of this here too. So, it's like, hey, I'm anme and

too. So, it's like, hey, I'm anme and like to your point earlier about the agency model being turned on its head.

It's I'm anme. I can say I want to I want to buy an advertising campaign.

Here's my budget. Here's what I care about. Like, get me in this kind of

about. Like, get me in this kind of inventory. And you're already starting

inventory. And you're already starting to see these kind of natural language prompts happening. Yeah,

prompts happening. Yeah, >> I like I can't wait for the DSPs to actually like get on this train as well, right? And so like

right? And so like and and I think that actually already has that functionality in their right.

So I don't I don't understand it don't make no sense.

>> Yeah. I mean well and maybe the ultimate reflection is how do we get rid of 16 billion dollars in scam ads being run every year? And the answer might just be

every year? And the answer might just be that we kind of like don't unless we hold Facebook culpable for criminal damages as a result of their scam ads.

Like if your advertising platform serves a scam ad to someone like you are >> you are as you are as in trouble as the

scam advertiser themsself.

>> But that would have to start with the user who was harmed, right? And like and then you'd have to trace it back from there.

Yeah.

>> Yeah, it does. It does sound quite difficult, >> right? [laughter]

>> right? [laughter] [gasps] >> Yeah, it sounds complicated. Um,

>> yeah. I mean, this is a really hard one.

This is a really big topic. This one

blew my mind because it's more money than flows through OpenRTB. So, that's

>> like like there's there's more money in scammy payday loans than there is in, you know, banner Open RTB. It's kind of it's kind of a dark reflection of >> Thanks. We're gonna end on a high note,

>> Thanks. We're gonna end on a high note, man. Wait, do we have one more? Because

man. Wait, do we have one more? Because

now >> we have more. We're not ending on this.

No, don't worry. I actually I actually think Open RTB is gonna like five to 10x in the next like year or two. Um

I I I think it's I think it's really coming.

>> Um and we'll do the news.

>> All right.

>> First, we just have that that Stan in um thing that we thought was funny. Oh

yeah.

Oh, we didn't want to do this one, but >> when you're sending out the the [laughter] the two funny parts about this the two funny parts about this trade desk creative one is you got to

you got to preview your creatives, man.

>> You got to you just got to do it.

Talking about buyers being responsible for what they're buying.

>> Yeah.

>> Um preview your creatives.

Troubleshooting.

Um, also if this came from a native platform, the native platform needs to fix how they assembled their native ads because this to to be fair, this might not be Trade Desk's fault. Like maybe

the preview didn't work or something or maybe it was a native ad. The other one is my favorite part about this ad is the button.

>> Um, >> does it not work?

>> No, the button's a spend more money button. Like

button. Like >> this [laughter] is a troubleshooting your DSP.

>> Hold on.

It says improve your ability to scale >> and like >> Yeah, I want to build a spend more money button if I'm a DSP. That's that's the

first button I put in my UI is way spend as much money as possible. Go.

>> Yeah. But I also think this whole thing has been like kind of an exercise and like not bothering to listen to people who are hands- on keyboard in your system. Like I remember at like great it

system. Like I remember at like great it it this is great for them. It should be easier to troubleshoot. Also, if I were if I were doing an AI startup today, I would create an agent to troubleshoot

campaigns or like create tear sheets or, you know, figure out which point in the chain this is going to go haywire so that I can fix it in front, right?

>> Oh, I mean, my first agent would be a an agent that automatically builds decks that says the campaign performed great no matter how the campaign performed.

>> That's a good one. Here's the audience you had.

And then news. Oh, I loved this one. We

can end with this one. Uh,

>> okay.

>> So, this was Meta defeats the antitrust case.

>> And the reason the reason I was so insistent on this one being in and people are going to challenge this because apparently the judge has some sort of Mishagos, but um the reason I loved this one is the foundation of the

antitrust case against Meta, right? So

that it had it had Meta and Instagram, they were integrated together was that people said that you can the ad density has been increasing and increasing and increasing on both of these platforms

over time. And the only way that that

over time. And the only way that that would be possible is if users had nothing else to use. So this is obviously a monopoly because they're making the user experience so bad that

users can't go anywhere else.

And the reason I loved this so much was because Meta's defense was so good and they were like, you know, we uh we've done we've we've done this experiment in Europe.

>> When we serve zero ads to users, zero ads, they spend a grand total of 7% more time in the app. So, in our estimation, users don't really care about getting

lots of ads, and it's not indicative of anything other than we can keep adding them, and users don't give a [ __ ] >> And this gets me going on the rant I was

on last week, which is people love talking about how bad the user experience is on the web, and I just don't think it's that bad. I really

don't. I I've seen enough real human beings consume massive quantities of MFA websites that have ad densities that will make anyone blush and they were

real human beings reading them that I know that uh what triggered Spongebob. Oh, I don't know. Um was it MFA? Um

know. Um was it MFA? Um

>> it was me typing exclamation point chalice in chat. So chat commands are working folks. Oh, I I love this one.

working folks. Oh, I I love this one.

But this is to say that uh yeah, the the web being in shitified is really just not such a bad thing. It's just the way that publishers make money.

>> Two things. That rant from it was last week, right? Chef Kiss I I giggled my

week, right? Chef Kiss I I giggled my way through that. People were staring at me. Fantastic. Second thing, people

me. Fantastic. Second thing, people don't actually hate ads. People hate bad ads. People hate ads that like are

ads. People hate ads that like are annoying or like very very loud or you know stuff like that. Fine. But as long as it's not disruptive, as long as the

ad experience is relatively anodine andor like like neat enough to grab the attention, right, you're you're good.

Oh, and the great irony of all of this is the more that users hate the ads, the more it means they probably work because they really are getting noticed. And

like I I really think this is like the the most damning thing about banners is that users find them so unintrusive because the users that means they can

kind of ignore them. Um, like I I'll just never forget Bounce Exchange when Bounce Exchange came out as a company and became this big thing and everyone was like, "Oh my gosh, this is performing so well. It like works for

performance marketing." I was like,

performance marketing." I was like, "Yeah, it's an exit pop."

>> Like, anyone who's ever worked in like old school digital marketing knows Pops and knows that pops work. Like,

[laughter] >> yeah. And the internet basically did

>> yeah. And the internet basically did away with pops because that was like the first thing where I think users really came out with pitchforks and they were like, "Someone please do something about all of these pop-ups and the browsers

all built in pop-up blockers." Um, but pops worked, man. Those were effective ads. They just screen takeover like

ads. They just screen takeover like [laughter] >> Yeah. You could always tell too in the

>> Yeah. You could always tell too in the off this happened when I was like starting my career and you'd hear some somebody at some point in the office go, "God damn it." And you're like, "That's [laughter] what happened. Those were

pops and they converted. They really

did. They were like linear ads.

>> And like now Netflix bought um um Warner like as it's official as of like an hour a couple hours ago. So, you know how the how how things have changed, how we've all grown up.

>> We're just going to let all the advertising companies buy up all the content companies and then they're just going to keep slapping more and more ads on there >> and the economy is going to keep growing. the economy is gonna keep

growing. the economy is gonna keep growing and we'll just make sure that they all happen via Open RTB. That's the

goal. As long as they happen in via Open RTB, >> it'll survive.

>> There's a respect for that. Let's do it.

>> But thank you for joining everybody.

Thank you, Hillary, so much for coming on. I could talk to you for hours about

on. I could talk to you for hours about Open RTB stuff. And it is actually >> We definitely need to get together soon, dude. Like this was fantastic.

dude. Like this was fantastic.

>> Yeah.

>> Oh, we have one more story.

>> Oh, one more. Oh, so we kind of covered this. We talked about this. We talked

this. We talked about this. We talked

about this. Um, but I Omnicom IPG, two drunks leaning on a lamp post. I know

you just had to put that up there. Sir

Martin Sorl went really hard at the Omnicom IPG merger. Just went really hard.

>> And uh, >> you were so mean.

>> You know, he's he's kind of a grumpy dude. Um, and but I think there's a lot

dude. Um, and but I think there's a lot of truth to what he said in that these are legacy business models, right?

like the like he had a very explicit line in there that I mentioned earlier which was scale helped in traditional media scale doesn't really help you in

digital and if scale doesn't help you I'm not sure I understand why holding companies work in digital um like why do you need a holding company if you don't

need scale uh if you can achieve tremendous results if you don't achieve better and better results the more and more people you have then why do you need a holding Yeah, but like but also let's keep in

mind like the human eyeballs like the open web as we know it is in for a a major shakeup. So human eyeballs are

major shakeup. So human eyeballs are going to be like largely I think on the apps like t the social apps and on TV

screens and like places where we're entertained, right? So and there you

entertained, right? So and there you absolutely still do need scale, right?

like streaming streaming TV you still need scale like >> well but that's that's that's scale of distribution though >> digital so I think his his scale comment

was >> you can you can bulk negotiate rates right that's really what it was for was when you're an agency holding company I can go to the cable network and be like

or I can go to ABC and be like ABC I represent this much budget so you're going to knock 50% off my prices or I'm not going to buy from you.

>> Yeah, but the upfronts have been getting smaller and smaller every year. So like

>> and you know and and no matter how many brands you represent, [laughter] Facebook makes $16 billion from scam advertisers per year. So So

you can march right into Betta's office and be like, "Hey, I want a discount."

And they'll be like, "Well, great. How

much money are you going to spend [laughter] >> if it's not $16 billion?"

>> Be like >> like I don't really care. Oh, good.

>> You can log into the UI and buy with everyone else or you can not do it. It's

really up to you. Um, but good luck explaining that to your brands when you say, "Hey, we're not going to have like meta in our media plans for this year because they wouldn't bend over

backwards on rates." Exactly. You can't

do that. They have no leverage.

>> Um, >> yeah, >> we'll get the open web there. I That's

my my my goal is to get the open web there. That's what I want. I I think it

there. That's what I want. I I think it should be there because it's it's still ads.

>> Yes. Let's do this.

>> Yeah, guys. Anything else? Any other

news?

>> That's it.

>> All right.

Only an hour and 12 minutes. I knew this was going to go over. Like I was when I went when I was when I was chatting with Hillary before this, guys, I was like, "Uh oh, this is we're going to have to keep this under two or three hours cuz

>> I know I >> open RTB is important." I'm really curious to see the comments that happen on this on this video, too, because I'm like because we we got we got we got in in the weeds here.

>> Well, it was in the weeds. I knew it was going to be in the weeds, but this is an important one in the weeds [applause] for people who buy media. Like, these

are things that matter if you buy media.

>> Yep.

>> Um even if you're a brand with stuff in house, these are things that matter.

These are things you should know about.

>> Um >> totally. But yeah, guys, thank you so

>> totally. But yeah, guys, thank you so much for coming on, Hillary. It was

wonderful.

>> Thanks for having me. [music]

>> Yeah.

[music] >> Yo, the crowd is back. The lights are dim. Front rows buzzing.

dim. Front rows buzzing.