S8, E01: Articulating a bull case for the open web (with Gareth Glaser)

Let's [music] get real. Half the time you're not sure if your campaigns work. You just kind of hope they do. Attribution numbers look good, but what if those [music] conversions would have happened anyway? Incremental helps you find out for real. No tracking people, no [music] guessing games, just clear answers on which campaigns actually move the needle. If you want to stop [music] guessing and start knowing, go to incremental. Because good marketing isn't about getting clicks, [music] it's about getting results. Check them out at incremental.com. That's [music] ncrmn.com or click the link in the show description.
The problem is that the distinction needs to be drawn [music] between the competence of the economists and the correctness of their analysis. Hello and welcome to the Mobile Dev Memo podcast. I'm your host, Eric Seufort, and I'm joined today by Gareth Glazer. Gareth, good to see you.

Hey, Eric. Good to see you, too, buddy.
So, I'm excited to have this conversation. I've wanted to do this kind of episode for a long time. And just to kind of set the table, I'll set up the the context. So, I don't know, maybe a month ago, I kind of tweeted out. I said, "Hey, I would love to have someone on the podcast to steal man the open web to just just to provide the to provide the bullcase, to articulate the bullcase for the open web." And a number of people that I respect a lot recommended you. So, RA Paparo recommended you and Eric Franchie recommended you. I think a couple other people recommended you and so here you are and uh thanks for doing this and I'm really excited to hear your position today.
Well, it's my pleasure. I uh spend all day every day talking about the open web. So, I'm ready.
So, this just another day at the office.
Just another day at the office, man.
All right. So, before we dive in, could you please introduce yourself to the audience?
Yeah, of course. So, I'm Gareth Glazer. I am like a disgustingly pure adte person. So I have only ever worked in independent adtech. My first job out of college was at a company called App

Nexus which was a very early programmatic company. And ever since then I have been starting companies in the space. So I was at App Nexus for 2 years. Uh, I started my first company, popped right out when I was 24, and started a performance marketing company that we built on top of programmatic pipes that I then learned the hard truths about programmatic and that was aqua hired in about a year. And then I started my second company. I switched from the buy side, right? The buy side of the industry being the side of programmatic that places that on behalf of advertisers to the sell side of the industry. The sell side being the side that helps publishers to maximize the amount of money that they make from their websites. So that second company was called RTK. I sold RTK to the Rubicon project in 2019. So the Rubicon project is a a large one of the few publicly traded independent adtech ad exchanges. And at Rubicon Project, I serve as the head of product for all things header bidding related, which is the framework that web publishers use to monetize their websites. And I actually

served as the chair of pre-bidjs. So I ran the pre-bid project for a few years while I was there specifically for websites. So the pre-bid project is the open source framework that 99% of web publishers use to maximize the yield of their websites, right? To make as much money as possible from the advertising opportunities they have. And then I popped out of that I guess my gosh about four years ago now to start my most recent company which is a data company. And it's a data company that is currently very much preoccupied with the open web. We have JavaScript that sits on publisher pages. We have a web crawler that does a bunch of technographic stuff that's super cool and like super purpose-built for digital advertising right now. And yeah, that's all to say that my entire career I have been doing programmatic ads on websites in some way, shape, or form, some flavor of it. It's it's all I know, man.
Well, you're the right person for this discussion. I'm I'm super super happy to have you here.
Okay. I think I've gotten into these conversations in the past about like

what is the open web and I think you know you could be talking about any number of things right and I think it's like I think it's it's maybe it's important to have a broad-based like universal definition of it maybe it's not for this conversation though let's do that and I want to just kind of kick you I want I want to toss this this this kind of first big broad conceptual question to you is what is the open web belongs inside that definition Is it useful to conceive of the open web as as a monolith or should we understand it as a constellation of distinct networks and markets that just share certain characteristics? What doesn't belong in the definition? And just let's kick off the conversation with with your definition.
Yeah. Uh this is a good one. There's a nice thread about this just a few weeks ago. So for me, the way that I think about the open web comes down to how the ads can be bought and sold. And when I talk about independent programmatic, right, I talk about where I've spent my career. Independent programmatic consists of, to use your word, a constellation of companies that buy,

sell, optimize, enhance the value of, enrich, decorate advertising impressions from one another. Right? So you have demand side companies that are primarily in charge of bidding and valuation and you have sellside companies that are in charge historically of aggregating supply. though these these distinctions are fluid. In my world, what makes something open web is that I can use an independent piece of technology that is placing ads on my behalf. And what I mean by placing ads is some sort of ad description is sent to that technology. It correlates the parameters associated with that advertising opportunity to the outcome that I am seeking and then places a bid and I can actually use this tool to deliver ads to that piece of inventory. An easier way to say this is I can buy the ads with a DSP with a demand side platform with a bidding platform. If I can integrate my own DSP, if I can use the trade desk, if I can use Yahoo, if I can use Amazon, if I can use one of those tools to buy it or all of those tools to buy it, I think about

it as being part of the open web with some weird caveats in there, right? Like DV3 buying YouTube doesn't really count because the only way to buy YouTube is DV3. DV3 being Google's DSP. YouTube not integrating with any other DSPs. It's the only one you can buy it from. But my fil definition is precisely that. The open web is things that I can log into the trade desk and buy or things that I can log into Yahoo and buy. The closed web, right, the walled gardens are things that they say, "Oh, if you want to buy my supply, you have to use my tech." And Facebook is a perfect example of this. The only way to buy Facebook is through Facebook's buying interface. That is the only way to get it. It used to be different. They had something called FBX for a hot minute and that was a pretty much a failed project, but they're a nice clean walled garden in that if I want to buy Facebook, I got to log into Facebook to do it
yet. And anybody uh interested in in the history of FBX, uh I I always love to recommend Chaos Monkeys, the book. I have you read that?
Yeah, I haven't read it, but I've certainly gone back and forth with its

author on Twitter.
Well, it's it's a fantastic book. Yeah, I enjoyed it anyway. Okay, thank you for that definition. I think that's useful to uh to set the stage here. Sure. But just maybe just one thing. So So Netflix then counts as open web. You can buy Netflix through Trade Desk.
Yeah, I think so. So Netflix integrates with a lot of different DSPs. You can use your own tech to buy it. People would argue that they're in a weird in between zone because Netflix doesn't really give you any data about what you're buying. But in my estimation, if I can use my centralized buying tool to buy it without logging into a Netflix UI, still open web.
Okay. Fair. And I'm not I'm not pushing back, by the way. I'm just clarifying.
Yeah. I know that's a good one because it's a weird one.
All right, so the definition is a bit clarified. That begs the question. What is your bull case for the open web in the medium term? Let's call that 3 to 5 years. And then what drives the growth in this bull thesis?
Yeah. So I normally answer this by starting out with well why are you bearish on it? But I'll I'll answer that question for myself. The reason that I find that most people are bearish on it is that they think that websites are going to go away in some shape or form.

uh they and I am not that extreme of a believer like I work with people who are believers in that. I work I talk to people all the time who are like oh websites are these outdated things. I don't think it's the case and the reason I'm bullish on it is twofold. So one is I think people underestimate how big the open web is. Uh and the reason for this is the buying tools have been crummy, right? And that's the whole thing in part reason number two. But even if open web traffic declines by some astonishing percentage, right? Let's say it goes down by 80%. There's still a ton of open web out there. And I have publishers that I work with that use my tooling that have basically already been hit by AI. A good example of this are weather publishers. Google has been surfacing the weather in their UI as a search result, right? Their consolidated search results. I forget the name for them. Uh it was even it was prior to Gemini, right? It was

prior to their like AI responses. They had like, oh, here's the weather. Those traffic hits have already happened. They happened years ago. And weather websites continue to survive and thrive uh primarily through direct traffic. And I think that there's a huge portion of the internet that is this. I think at least 20% is this. And we have a nice useful statistic for how big the open web is because Google just had to talk about how big GAM is in their antitrust case. It is 8 million queries per second dduplicated which is an astonishing amount of traffic. That's a crazy number to have a to have 8 million dduplicated ad impressions per day. It is like let me let me add some commas in here. I did this math so I was coming in. But either way it is at a $2 CPM. It's $140 million a day in revenue. At a 10 million at a $10 CPM, it's five times that. And so I think that part one of my bare case is for all of the, oh, the traffic is disappearing people, right? There was an Axio thing that came out today. It's

like, oh, traffic's down 30% on news pubs. I'm like, yeah, but there's an astounding amount of traffic in the open web. It should be an advertising juggernaut. So part one of my answer is the open web is huge. And part one of my answer being the foundation for part two, which is we've done a really crappy job of running ads on it. And the bullcase is entirely rooted in the notion that we are making way less money from this traffic than we should be. So we should be driving more advertiser outcomes. It should be performing better. But I'll get into that in part two of my answer. But does part one of my answer make sense? There the open web is much bigger than your average person expects it to be because their anecdotal usage and the conversations they have with their friends are like I don't go to websites anymore. Tons of people do. People are weird. They go to websites for the strangest things that you can imagine with me so far.
Yeah. Let me just see if I can play that back. So it sounds like what you're saying is we can invalidate a a very

high-profile bear case which is just that the you know the web as we know it is disintegrating kind of in real time because a lot of that's just already baked in which you know I agree I think a lot of that is true and so invalidate the bare case the web is still very large and it sounds like where you're heading and I'll you know maybe just open the door for you.
Yeah. is that well the so the growth case the bull case then is that we've just done a really poor job of monetizing and maybe these same mechanisms that are uh making people really excited about advertising in very specific placements like in chat GBT or like on Facebook with you know this sort of AI enablement are also tailwinds for the open web. I don't want to put words in your mouth but it sounds like that's where we're that's the door you've opened.
Yeah to an extent. So the example I always use for this is average CPMs. Your average CBM on Facebook, and you could look this up globally, right, is the conservative number is $6. The less conservative number is $10 to $15. If

you were to take that number and apply it to the open web, right, let's take $10. The open web is now a $250 billion market, whereas today it's 30. So the ads on the, and my explanation for this, by the way, is that the ads on the open web don't work. So, I'm going to give you guys a little extra information as part of my second answer, but I think it's really important because I think it's structural. And the open web advertising, programmatic advertising, independent advertising doesn't work or doesn't have the scale that it should because it was never really built for performance. Uh there are a few examples of this, but the first one is that one, it's primarily a branding medium, and you can ask any affiliate this. Very, very few affiliates run campaigns in the open web because they don't convert. And that's because the pipes themselves were built for companies like huge brands that had giant lists of user IDs that wanted to run their campaigns at scale across the internet, but they're not like looking for postclick conversions.

They are just saying, "Oh, I have a giant mm that I'm running as part of my marketing department and open web programmatic is 10% of my spend. I just need to make sure I hit these incremental users, which is like 1970s marketing course stuff." and the open web, right? DSPs, SSPs, all of the things involved in the open web were built to serve that market. Uh, and so the way that inventory is labeled, the way that SSPs and DSPs communicate with one another, the whole set of massive technology that we've built was built with that in mind. And the fact of the matter is that market is just only so big. The the brands who want to hit a list of user IDs is only so big. And where Meta sees all of their growth, right, is SMBs. SMBs doing postclick conversion based marketing. It's performance marketing. Same thing in mobile apps. Mobile apps work well because the app ecosystem is built on performance. I run this budget, I get more installs, I do my LTV cow

calculation, I spend more money. That doesn't exist on the open web. The flywheel is not there because all these tools were built for brands to run these giant budgets. So it actually means that there's a tremendous value case to be made if you can properly categorize the inventory on the open web and build infrastructure geared towards performance. And I think that's where my bullc case really comes in is the open web today. The reason we make $30 billion a year and not 150 or 200 billion is because our tools don't work for performance marketers. And if we can build tools that work for performance marketers and isolate the reasons why these legacy tools don't work for them, you can actually unlock a tremendous amount of economic activity. Uh, and this is in my investor deck. This is like my reason for being this is my bullcase is I have a few very specific like in the weeds thesis on why this developed the way that it did. But in aggregate performance marketing doesn't

exist on the open web. like it's just it's a tiny fraction of what it could be. And I think if we fix that, the open web grows like crazy. Even if the amount of inventory declines by 50%, the unlock of making performance marketing work has is such an tremendous upside that I'm not worried about traffic declining. So that's part two of my answer is it doesn't work for performance and if it did it would grow like crazy.
I'd love to hear the background like why did the tools develop in the way they did? like why are we why is the open web kind of the remit of like this you know I think that's a great way to describe it this 1970s marketing course uh sensibility why didn't it adapt or why did it evolve to what it is today
yeah it's a really good question uh there are three or four answers that I would posit for this and there's no way I can be exhaustive but if you go back to the beginning so like I said my first job was at App Nexus I work in programmatic advertising which are the the bidding ecosystems right so all of the different tools that exist to allocate campaigns to allocate budgets.

If you go back to like 2007, there was some performance marketing in the open web when ad network ex existed. So if you look back at like value click CPX, right? CPX was a performance network. They were known as one of the only guys who could make banners convert and that was a thing in the early 2000s. I think what happened was as the programmatic companies were growing, they realized, oh, it's a lot easier if you can get an IO from an agency, right, or from a big brand where they're just like, I need to spend $2 million and I need to hit these users and I don't really care about performance than it is to get into the knife fight of making CPA campaigns convert. And I think that programmatic in general, the first structural thing is they got addicted to that. They got addicted to the agency budgets that once you got them it was very very easy to deliver upon them and make it look like it performed. Uh and the first DSPs

really I mean the trade desk right the darling of the programmatic industry well ex I mean they're still a darling I I still love the trade desk but they were the best agency company that existed. Every major agency loved them. They spoke agency. they did the reports the right way. They weren't a performance marketing company. They were a agency services company to a large extent. Uh and that's where programmatic companies simply saw success. So people were like why would I do the hard thing when I can do the easy thing. So I do think that's why it evolved in this way. And then they can also give all of these very enterprise agency brand specific pitches, right? Like oh you can activate your first party data with us in this neutral way. we're not going to share your data with other people. There are all of these things that they can layer into the enterprise that that sound really good. Uh and as we've seen with LiveRamp and a few of these other identity companies, those pitches resonated like they they were very successful tech companies. So I think

that's my primary explanation is that programmatic got addicted to the easy agency dollars and once those were what made money move the product managers in charge of things did not develop things for performance marketers because why would they? That's fascinating. Thank you. Was there a geographic component to that? the agencies and the a lot of the companies you're describing. I think if I'm wrong here, please correct me, but it sounds it's like you're talking about these these network companies or like you know the companies that evolved to service the programmatic marketplace for the open web. Sounds like they were all colllocated in New York City and then you had the kind of walled garden ecosystem mostly evolved in on the west coast. Was there a geographic component to this? Do you think that played a part?
It's a good question. Uh certainly the early programmatic companies were more often than not Silicon Alley companies. That being said, the trade desk has always been in California. Not in San Francisco, but they've always been I think it's Santa Barbara. My California geography is very bad. [gasps]

So I think that the the easier explanation for me for that one is if you've ever worked with an agency holding company, once you succeed in one part of it, penetrating the rest of it is is gets easier and easier where everybody knows everybody. They share data, they share recommendations, they share these things. And I think once they really saw success there, they it snowballed. And so the only geographic component that I would think of is if you look at where independent programmatic is very very successful, it is the EU and the US. And those happen to be the two places where like four of the five major agency holding companies are headquartered. The programmatic industry in Japan is fledgling, right? Dansu's over there. And then if you look at the other markets right you look at like India it is totally dominated by the world gardens completely and utterly. Uh so I think there is certainly a a relationship between where the agency holding companies are and where programmatic advertising took cold.
And when you say penetrating the the holding companies what do you mean? You

mean then becoming uh a customer of these ancillary services beyond just media buying? So no no no. So when uh when a programmatic company is trying to work with an agency holding company holding company has a dozens of agencies, hundreds of agencies with brand teams within each agency. If you succeed for one brand team inside of one agency, you can now wriggle your tentacles through the rest of that agency and then through the rest of the holding company and that becomes your primary sales mechanism.
I see. So not not a client to the holding company's subsidiaries but becoming a vendor to the whole I I see. Okay, got it.
You are a vendor. It's vendor sales. There's another nasty component to this that I didn't want to mention but I will is there are some unscrupulous histories here which are rebates right and things like that where programmatic companies became a way for a lot of agencies and agency holding companies to recapture margin. So the story that goes on here

is it's another knife fight for agencies when they're pitching for brands and there's been tremendous downward pressure on what agencies can charge brands for their services. Tremendous downward pressure to the point where they're often pitching below where they're operating profitably. and they can run a programmatic budget and the programmatic vendor says, "Hey, you know, if you spend $10 million with me, I'll pay you $1 million of it back." And that $1 million payback now goes into a slush fund for the company. It's coming out of media spend, but it's a slush fund. This is illegal, but it happens, right? There was just a huge rebate thing in China for one of the big holding companies, and they were like, "The slush fund was huge. It was hundreds of millions of dollars." So there that component also exists, right? where these programmatic companies became a way for agencies to reclaim margin.
And is is there a consumer adoption or behavioral change proposition in the in the bull thesis?
I don't think so quite honestly. I think that current usage is don't get me

wrong. I don't it's not going to get to Facebook CPMs. Like Facebook CPMs are a beautiful thing, but it could certainly get halfway there. And my my reasoning for this is at the end of the day, we're talking about ads on a website. If you look at Facebook's banner ad revenue, it is meaningful. It is nine figures. And a banner ad is a banner ad at the end of the day. Like the demographic targeting, I there's tons of demog amazing demographic targeting vendors in open programmatic. Uh I really think the ad experiences and how users engage with them today. I don't think there's something so unique about meta. Uh, I think that the websites have those users, the same users, and it's just a matter of getting the right ads in front of the right people and modeling it properly, which we are woefully bad at.
When you say Meta's banner ad revenue, what do you mean? You mean the static image revenue from their feed? Okay.
Yeah. Yeah. Exactly. Well, what they categorize as banners? I've always assumed that there was static images

from the feed and like the little sidebars, but they're not super specific in their filings.
Got it. Are there categories of advertisers that are underinvested in the open web just on the basis of a potential return right now? So like if they were investing more and they were pursuing this with kind of you know this direct response postclick conversion analysis mindset that they that they would be seeing better returns.
Yeah. I really don't think we've built the tooling for them properly. I think there's like there are two companies out there that have done it that are focused on this. It's Tibula and Teds, right? or outbrain and tabula is seeing growth. So Tabula is like not a terrible business and I I like the way that they run things other than the fact that they report on their financials with tech included which drives me crazy. But I I do think there's a thesis there that hey, some performance marketers can make this work. But Tiboula isn't a big SMB company, right? They don't have localized sales teams going out the way that Google does, right? Where Google is

like, hey, are you advertising on Google? You should be buying on Google search for your veterinary clinic. Why aren't you and incentivizing all the localized agencies and doing all that stuff? That motion hasn't been executed. But the technology needs to be built to service that first. And it's just it doesn't exist today. Like there's part of my thesis when I was raising money is we need to do some remodeling of the house before we can start inviting guests over.
I see. So there's not like like some you know app developer or whatever like they or some ecom brand or DDC brand they couldn't just flip the switch on programmatic web right now and you know see compelling rorowaz. You don't think
I don't think the platforms out there offer it. I'm sure there are people who would argue with me about this. Uh like I'm sure Adam from Tibula would argue with me about this, but I don't think an open programmatic it's a thing.
Yeah, I mean I've bought on Tiboula before. It's a mixed bag particularly for apps. I mean just because of the the platform misalignment.
Yeah,
I don't know about ecom. Maybe it's also, you know, the placements are not

really optimal. It's at the bottom.
That's the key, man. The the best programmatic placements are just not in these native platforms. Uh and so the fact and in fact I actually think it is a remarkable contributor to the bullcase that those content well things work for anything like [laughter]
like I think it's amazing that what many people in programmatic affectionately refer to as the chum buckets convert for anything. That's crazy to me. Uh like there are way better ad placements on publisher websites and that's what Tabula's realized product is trying to get into. But man, they they weren't born in programmatic. So they don't like it's a bane thing. They're they've come to the darkness. They were they didn't grow up in it. They don't know what they're getting into.
What about like a content component? So, you know, there is I mean it's happening now. I mean, there's this dramatic growth in output, right? And we see this in the app store certainly,
you know, which has been overwhelmed by new app submissions. I mean, the app store has seen the review weight times balloon, right? just as a function of

these vibecoded apps that you know are being submitted.
I mean that also applies to the web, right? And potentially disproportionately so because publishing to the web is fairly straightforward. Publishing to an app store requires like some you know integrations with platform tools. So is there a content component that well you're just going to see more content being created for the web and so that potentially just gives more surface area for these ads to be applied to?
It's certainly possible. It really depends on consumption behavior from consumers. there. There's also like a little bit of a backlash against AI generated content on the web. Even though, you know, AI assisted versus AI generated, it's it's a discussion. I don't harp on it too much because we're already in a supply glut. Like that's that's the one of my core tenets of my bull case is that we have more supply already than we know what to do with by a significant margin. So, I I think that certainly it AI publishing helps publishers run leaner. It helps them be more efficient, which is good for them as businesses because one bare case

could be that well the publishing companies go under before the open web figures out how to monetize. They just croak. But I I don't know. I don't think about it too much as because I don't worry about surface area. I think we already have more than we know what to do with.
When you say the publishing companies croak, what are you talking about? Like the New York Times or who are you talking about?
Yeah, man. It's hard to be well New York Times is a rare exception because they have a burgeoning gaming and subscription business. But a lot of the independent publishers, news publishers, people like that who don't have big subscription businesses and are having trouble building them, they they could have real problems, right? They've been in financial stress for a couple years now because they actually a little bit of unnecessary history, but they were hurt by programmatic itself. Because if you think about the thing that I was talking about earlier where you have big brands who just want to go buy lists of users, that means that they don't care if they get that user on the New York Times or on a mommy blog. And that sucks for the New York Times' sales teams. And so there's a the publisher

world actually is relatively disillusioned with open web programmatic because the CPMs never came back to where they were. while at the same time, right, we see what we're seeing now with declining traffic. So, it's kind of like a double whammy for for poor publishers.
Where does a company like Substack fit into this map of the ecosystem? Are they open web? I mean, you So, you I don't believe you can. So, first of all, they don't really have ads per se, but like let's imagine that they would at some point. Do you think they would open up to DSPs or would they just do kind of self-s serve?
Yeah. And like Beehive has ads. I spend time thinking about this. So Beehive for me I I'm pretty fairly certain is like a closed ecosystem right now. If you want to buy Beehive ads, you got to go into Beehive. There are one or two ESPs who allow you to insert ad tags, but ads inside of emails suck because there's no JavaScript. So like the way that you can execute the advertising experience and track things is super duper limited. And Google understandably keeps the crazy lid on what emails are allowed to do

from a code perspective. So I certainly think that these guys have flirted with the idea because Beehive invested in building this site builder that you know Tyler Denk is constantly tweeting about uh or posting on LinkedIn about. But I think there's a next step for them to bring this inventory to the open web which is hey major publisher migrate your business to my ESP. I will help you get new readers and then when you drive users to your website you can monetize them with the open web. Uh because monetizing with the open web generally scales magnificently. There are tens of billions of dollars flowing through the open web. That will be better than a you know tier four walled garden can pull off. Apple's also very restrictive with what you can do in emails.
Apple mail. Yeah.
Not just in the mail app, but also in the registration flow on iOS with uh hide my email.
Totally. Now, I actually haven't heard too much about that, which is really interesting because when that first came out, that was when the programmatic ecosystem was rolling out hashed emails as persistent user identifiers. And

everyone was like, "Oh my god, this is the apocalypse. Like, it's over. User identification's gone." And I don't know, a few of those really nasty Apple features, I just I I'm not sure how much adoption they've seen, but I'm sure it's there. I'm sure it's a percentage and then percentages up. AI is changing everything, but it's only as good as the signals behind it. Branch connects customer [music] interactions across paid, organic, offline, email, web, and app touch points and turns them into the trusted context you need with links and attribution that capture the full user journey. Learn more at branch.io. And while you're there, [music] check out Branch's AI search and discovery report covering insights from more than 300 enterprise marketing, [music] growth, and digital leaders to understand how the industry is responding to the rise of AI [music] search. That's branch.io. Talk to me about the independent publisher model. How is that evolving? We talked about kind of New York Times, but are there any like interesting cases

here similar maybe to New York Times? You know, you mentioned their gaming business, the subscription business. New York Times is growing, right? It's a healthy business. Are there any other sort of interesting cases there that are worth hovering on?
Well, I do think and this is really funny because so when I was first starting my company, the first thing that we built was LTV measurement for publishers because no web publisher on earth could tell you the advertising LTV of one of their users. The infrastructure that existed for publishers, so it was Google Analytics and Google Ad Manager could not measure it. I couldn't sell that to anyone four years ago.
[laughter]
Like I would pitch publishers on it and they would just be like we don't run our business this way. Like we don't know what you're talking about. Uh and I was like but you know you have to understand this is the lifeblood of the app ecosystem. The understanding how much money you make per user for your free tear users is what enables you to grow as a business. So we are seeing this now through AI through the decreases in traffic. I no longer get blank stairs

when I talk about this product to publishers. They're like, "Oh, yeah. We have an internal subscriptions team. They have a lot of power. They're thinking about this and we're trying to think through our strategy for this." So, I would say I don't have a single interesting one to talk to you about because they are all doing this now because they all know that they need to and the ones that won't will not survive because organic traffic is drying up. But, they can be very successful doing this. So I actually think if there's a there's an interesting class of companies that I would refer to as agencies for open web publishers who resemble install networks for apps at the beginning of the last cycle.
What were these publishers orienting their business around if it wasn't LTV? I mean what was the headline metric that they cared about?
Page views from organic.
I see.
Crazy, right?
This is pretty eye opening. So like if we think about the level of sophistication, it's like that gap is a pretty pronounced opportunity. It seems like that's the substance of the thesis. And you know, as you're explaining it to me, I agree. It does seem like there is

a a very substantial operating gap there that just in addressing that unlocks monetization and value.
Totally. Whenever I'm coming to these conversations, whenever I write about this, whenever I talk about this, I'm like the open web that we know never really got started. It never became a grown-up business. It was basically a hobby business for Google search. That's what the open web has been where every publisher, their entire worldview was oriented around Google search and maximizing Google search traffic. And it's just time for it to resemble the app ecosystem a lot more with one really lovely caveat and that's that the architecture of the open web is open. Like it can be accessed with a browser. There is no central authority through whom you need to move. when you're monetizing, when you're starting a website, there are all sorts of attractive things about the decentralization of the architecture of web browsing in general that I think makes it a really attractive medium if you really take a step back and think

like, oh, as a way to get the thing that I built in front of humans, do I prefer an app store? Do I prefer a browser? Uh, and it's it's fun. So I think adapting the operating model more towards the call the design sensibilities or the the the business model sensibilities of of the app ecosystem then does invite the question well then these publishers that are now tracking LTV who have been weaned off of this reliable stream of organic traffic do they start doing user acquisition campaigns for themselves if I'm operating a subscription business where I know the ads LTV of my users as they come in in aggregate or segmented or you know whichever the case I then know what I can go and bid to acquire one of these. Does that become part of the the the operating calculus?
I hope so. I talk to companies about doing it all the time. Uh there are one or two companies that I know that are in their fledgling state that are thinking about this as an operating model and I think it's coming. I think this is an

inevitability. What there was a really interesting thought here which is can I run ads in meta to drive subscriptions to my newsletter? Can I harvest users from walled gardens and bring them out of them? And the answer is yes. Like and we should be doing this. It should be a thing. So yeah, I'm I'm very bullish on that. I uh I hope people get it together. I built my initial tooling at my company to support those businesses. I haven't seen the adoption yet, but I certainly can name a few companies that I've talked to who are like, "We want to be that agency. We want to be the company that approaches publishers and says, "We're going to do user acquisition for you." Well, you know, I I run a a subscription and I've spent a lot of time thinking about what you just articulated and uh I I certainly believe there's an opportunity there, but time will tell. All right. Talk to me and I don't like I didn't want to make this the focal point of the the conversation because it feels like kind of a distraction from what I wanted to get from you, which is the bull case, but how do the outcome and the remedies in in Google's adtech antitrust trial contribute to the bull thesis? and and

it and maybe they don't and that's that's totally fine too but you know how does this change things how does this sort of like reorganize things
yeah so I'm real deep in that so that's like one of my things because I was a chair of brejs right so when actually Jedi blue was happening we were having conversations with Facebook I remember this so Tom Kershaw was the chairman at the time Facebook was engaging with us about taking over the mobile aspects of pre-bid and then they ghosted us just like out of nowhere ghosted us and that was Jedi blue they did the deal with Google to at preferred rates when they're buying through addicts. And a part of that deal was you do not work with pre-bid. Could you just stop there for a second and and just explain to people what Jedi Blue is because that was a really fascinating story. And I I do think most people are pretty ignorant of that that chapter uh in the in the history of of you know the well in the history of the Google antitrust uh chronology. But if you could just give a give a kind of a history of of the Jedi the Jedi Blue incident.
Yeah, of course. So, and I haven't read

these court docs in a while, so I I'm probably going to butcher it, but the extent of it was Google was very angry about pre-bid. So, I'll give a little history about what pre-bid does where traditionally when a publisher would work with third-party marketplaces to sell their ads, in order to sell to them, they had to integrate them with their ad server in a very inefficient way. What I mean by that is any publishers or anyone who's listening who runs a mobile app, you are very accustomed to mediation and that is how web publishers used to do things. You it would be like I'm going to send this opportunity to this ad network where this price floor. They'll take it or they won't take it. If they don't take it, they'll send it back to me. I'll send it to someone else. It's waterfalling. Pre-bid existed to undo that. And so what pre-bid would do is it would solicit bids from those ad networks, from those marketplaces, and instead of a waterfall, it would take all of those bids and jam them into Google's ad server uh using a hack of key value

pairs and line items. And Google didn't see this coming. So, and right because Google had a huge advantage. your your primary mediation platform that mediates all those integrations with the others almost always has their own ad exchange that's operating and it's in such an advantageous position because it can see what other people are taking or not taking. It can insert its bids wherever it wants to. Uh it's a like this was part of the GAM adex tie that upset people so much is Gam's Google's ad exchange had this huge advantage and pre-bid undid it. It made the battle occur in the ad server with live bids. publisher revenue increased by some astounding number like 70%. So my my company RTK was one of the first companies to sell this technology to people and it got too big too fast for Google to kill it because I'm sure Google would have loved to have killed pre-bid in the crib if they could have. Uh and there were internal emails that came out in discovery where they were like ah this is we hate this pre-bid

thing. This is annoying like we need to do something about this. So fast forward, pre-bid is talking to Facebook because the same notion, right, of an open-source auction where the code is open source, the mechanics are open source, everything can be viewed by anyone, it's maintained by the community could be a really nice thing for Facebook to throw their weight behind in the mobile app ecosystem because ultimately if pre-bid becomes the boss and Facebook is the primary driver of pre-bid, they could exert a lot of control in the mediation space. And I actually think if Facebook had done this, the history of mobile app monetization could be very different. Like the rise of apploven might not have been as meteoric as we saw it. Like it would have there would have been a neutral third party platform that everyone used that they ran their auctions through. Could have been a very interesting thing. But Google got wind that Facebook was talking to pre-bid and part of the stipulations of them getting a discounted deal, right? So they got a bunch of incentives to spend their

Facebook media dollars, right, their advertising dollars through Google. And one of the stipulations was you will not participate in prep. You will not work with this open source organization amongst a bunch of other things that you know Google got in a lot of trouble for doing with them like bidding behavior and auction dynamics and things like that. But this was my firsthand experience with Jedi Blue was that Google did not like the competition that this brought into the ecosystem and did not want further adoption of pre-bid elsewhere, especially adoption of pre-bid by someone like Meta. So the crux of the Google case was tying between Google Ad Exchange and Google Ad Manager. Google Ad Exchange being Google's marketplace, Google Ad Manager being the tool that Google built for publishers to manage ads on their websites. The distinction between the two being a marketplace is I send an opportunity out. They go and they get bids for me and then they pay me money. The ad server, the ad management tool does not have any demand in it. It is

literally a tool. I put a piece of code on my website and then I get my own demand and it helps me add that demand to my website. So a huge proportion of publishers the tying was that in order to get access to Google ad exchange in order to use their marketplace you as a publisher had to use Google ad manager. You had to do it. You did not have a choice. It was the only way to get access to adex which is not true for all the other exchanges. Pre-bid's purpose was that all of the ad marketplaces can compete in a consolidated auction with one another. Adex was of course the loan hold out for this. the only way to get adex was to install GAM. So the brunt of the outcome of the antitrust trial was that okay that's not allowed anymore. You can't force publishers to use GAM to get adex. Addex has to buy from pre-bid. The net effect of this I think is going to be that all of the small publishers who don't use GAM which is a crazy proportion like 90% of small publishers

don't use Google ad manager. They don't use the ad management platform because they have nothing to manage. They don't have sales teams. They don't have things like that. They'll stop using it. Ads will get faster on their websites. They'll make some more money, but that's about it. I don't think much else happens, man. So, I really, apart from making ad stacks on publishers a little bit more efficient and a few new ad tech companies popping up to like help with management there, I don't think it moves the needle on the thesis that we're talking about. Uh, I think that they're deeper than just, oh, publishers have to use GAM. they are the construction of the programmatic industry was not well suited to performance marketing. PubMatics CEO estimated that a 1 percentage point market share shift to it as a result of the remedies imposed which the remedies I think we started talking before the remedies judgment certainly before the judgment was released in its entirety maybe it was after they had given the synopsis I

don't actually remember but well one percentage point market share shift to pubmatic would could represent 50 to 75 million in incremental net revenues they said that after the remedies were decided but before they were released but they also said that in prepared testimony I believe in the trial. So how much of that opportunity how much how much of the and you just said there's not a whole lot of of the the the both thesis itself is not uh predicated on on the remedies but how much of like if you think about that share shift and that revenue to PubMatic how much opportunity is just spend transfer versus okay well now Pubmatic's going to invest more now they're going to create better tools now they're potentially going to provide more value to advertiser how much how much market expansion is inherent in that is there or or if any Yeah. So, Pubmatic is a perfect example of a traditional programmatic company that is addicted to agency style stuff. There are few companies that play that game as well as Pubmatic does. So, shifting

spends to them. I'm happy for Pubmatic. Like, don't get me wrong. Like, I I love it when independent programmatic companies succeed. I want things to go well. But does that move the needle for performance marketing? I do not think so. I think it will be will do more of the same. They will double down on the business lines that are making them money. They'll spend more money in CTV trying to get CTV budgets which really do resemble the old programmatic budgets because measuring outcomes in CTV is very difficult. And yeah, I I think that the Google stuff is just it's more the same. The the old programmatic guys, they're the king is dead. Long live the king, you What would it look like to build, you know, something from the ground up that was designed to serve the interests of SMBs? Would it be that dramatically different or is this more just like an operating style, you know, a business development quirk or or would the core products look different?
Little of both. So, I started my company to address some fundamental issues in the construction of the data used for

programmatic auctions. I'll get specific because I can explain in two seconds. I got my elevator pitch. uh it is the way that individual advertising opportunities are described in open web programmatic is incredibly unreliable. So an example of programmatic being built for user ID buyers the level of granularity that most programmatic buyers operate at is domain. They look at their ad performance by domain. If you have ever been to a website within a single domain, there are lots of different ad experiences like like the variability in performance is huge. And the way that ad experiences individually are communicated are labeled or described in open web programmatic is it has to be completely redone. So like a core piece of our IP is standardizing that and making that into a thing. But I only bring this up be to illustrate how systematically dysfunctional programmatic is for performance marketing. Cuz if you were to ask

someone at Meta, hey, when you're optimizing campaigns, do you take the advertising experience? Do you take the placement into account? I'm sure they would say, of course, are you insane? Like it's one of the most important things that we look at when we're optimizing a campaign. So there are some structural things that need to happen in programmatic if we're remodeling this. Uh and then there are some companies there's some companies popping up. I got some buddies who are running some some DSPs. They're doing cool stuff with like um procedurally generated creatives from the LLMs and contextualization has become a non-ifferiator, right? LLM have made it so that anyone can figure out what a web page is about to create conformity between the advertisement and the web page itself. So there's like this is a long way of me saying there are people out there trying to fix the problems and to start a new company that does this once those problems are solved I think will be actually relatively straightforward. It will be easier now certainly than it was 10 years ago. 10 years ago to start a new programmatic

company and one the technology itself was wildly expensive right like things that handle high QPS infrastructure that handles high QPS we've come leaps and bounds in how to build that and two is pre-bid the emergence of pre-bid as an open- source centralized auction means that a small company can get an integration with a lot of media relatively easily so is it a herculean undertaking 10 years ago Yes, herculean. Really, really hard. Now, eh, 50%, it's hard. It's not impossible, but it's hard. Give it two or three years, I think it'll be you'll see some upstarts.
What about identity? So, does that play a role here? Like behavioral targeting or I mean, is this all is the sort of like crux of the opportunity still on page context for targeting?
Yeah, I really think the crux of the opportunity is not just context, but the inventory itself. There are lots of ID companies out there and don't get me wrong, this is still a big asset that Meta has, right? Their demographic data is very interesting, but Amazon's is

even scarier. Amazon's ecom data is wild. It's an insane data asset. The tricky part becomes communicating those identities across the open web. And where we are going to land with this, it is very simple. It is a cat-and- mouse game of hashing, right? Where it's how can I probabilistically predict who this user is? And we're seeing it now. Apple keeps rolling out things to prevent people from fingerprinting who users are. People keep figuring out new ways to identify various behaviors in browsers and computers and things like that to try and figure out who they are. And third party cookies are still around on Chrome. So I think the ID solutions are there. They are going to exist for anyone who advertises in the open web. But I really, really, really think that the core of the opportunity is fixing all of the parts that aren't identity. Uh because identity has been what programmatic has been obsessed with for the past 15 years.
The publishers know a lot, presumably

some do anyway, about their users. Is there a way to communicate that at the moment of the bid request that doesn't involve, you know, the programmatic operators keeping a dossier on these users? Uh, all right. I'm I'll get myself in some trouble with this answer. There have been a laundry list of companies telling publishers you have valuable data about your users for like a decade and getting publishers to pay them for it. And nothing has ever come of it because it doesn't work any better than the data that's available in DSPs. The things that web publishers know about their users are so limited in scale unless you're an absolutely gigantic publisher and so abstracted from buying behavior that I have yet to see someone really create first-party data assets that are super duper compelling. And there are companies that have raised nine figures who say that they do this and I remain a

skeptic. So, I'm always wary of that pitch because I've seen it delivered to web publishers for such a long time and they've spent so much money on it and nothing has come of it. Garrett, thank you so much. This was really fascinating, really enlightening. I really appreciated all the historical context you brought to the conversation. How can people engage with you? How can they uh learn more about what you do and and your company?
Yeah, I'm super active on X. Ping me there. I have my own podcast that I run every week where we talk about super ad tech nerd stuff and I'm like serially online. Like that's a problem how much I'm chatting with people. So don't be a stranger. Feel free to add me on LinkedIn. Feel free to ping me on X. Whatever you like. I'm around man. And my company I didn't actually say the name at any point. I run a company called Gamera. And like I said we're a data company
and we'll link to that in the show notes. Thank you so much. Appreciate it.
Cool. Thank you Eric.