← Marketecture: Get Smart. Fast.
Marketecture: Get Smart. Fast.
Ari Paparo on the Era of Outcomes in Digital Advertising at Marketecture Live
00:24:40
Loading summary
Transcript6 lines
- [00:00]
A
I'll be there.
- [00:01]
B
Will you? This is Ari and I'm talking about.
- [00:03]
A
Our upcoming Market Live conference in New York on March 10th and 11th. We've got an amazing two day agenda filled with top tier speakers and insights you can use right away. Some of the speakers we've announced already include Sophia Koluchi, the CMO at Molson Coors, Lance Armstrong, the General Partner at Next Ventures, Neil Vogel, the CEO of People, Joanna o', Connell, the Chief Intelligence Officer at Omnicom Media Group, Jeremiah Oweng, a General Partner at Blitzscaling Ventures. Of course we'll have the Startup showcase like we always do. I'll be giving my keynote and we'll have special guests for our live podcast. You should definitely look at buying a ticket. It's@marketecturelive.com and we have special arrangements for qualified brands, agencies and publishers.
- [00:44]
B
So go to marketlive.com.
- [00:50]
A
This podcast is brought to you by Audiohook, the leading independent audio DSP. Audiohook has direct publisher integrations into all major podcast and streaming radio platforms, providing 40% more inventory than what could be accessed in omnichannel DSPs. What's more, audiobook has full transcripts on more than 90% of all podcast inventory, enabling advanced contextual targeting and brand suitability. Audiohook is so confident that in addition to CPM buys, they offer the industry's only pay for performance option, where brands can scale audio and podcasting with peace of mind, knowing they are only paying for outcomes. Visit audiohook.com to learn more. That's audiohook.com.
- [01:41]
B
All right, so I'm going to get us started talking about outcomes. Outcomes is. It's kind of funny because I think we've all been in the outcomes business since the start of digital media. Obviously, if you spend money, you want outcomes. But it's become kind of fashionable to talk about it. So I'm going to talk about it, right? And I'm going to talk a little bit about the buy side and the sell side of how this is sort of changing, how AI fits in, how the open web is dying, how that fits in, all these sort of kind of, you know, try to thread the needle through all these trends and put it in the context of outcomes, because that's the theme of today. So this all started because Joe Zapp. Is Joe here? Joe? Anyone? Joe? No? Okay. Joe's not here yet. He should be here today. He coined this phrase, phrase, the outcomes era. And then I decided to call our conference that. And then I was like, oh shoot, I should give Joe credit because he came up with the phrase, but then I changed the name to the era of outcomes and his response was blaming me for changing the name. This is an exact quote from his email to me. So thank you, Joe. Why do outcomes matter? I think this slide sort of motivates it more than any other. The continued decline in CMO tenure since CMOs are being asked to produce financial results by investing and getting outcomes and they're not apparently doing that great a job or at least in the perception of their boards, they're not, because basically it's an ROI based business. And there's even talk, I mean obviously in the west coast very often growth doesn't report to the cmo. A lot of tech companies now put growth in the product org. You have some people saying, well, in B2B maybe growth should be in the sales org, right? So it's a pretty challenging time if you're trying to create outcomes as a marketer, a different approach. This is one of those unfair comparison slides. Applovin is the most valuable company in advertising outside of Google, et cetera. And they're in what I would call a very unsexy part of the business. App installs. It's a business that I don't really care for. I haven't been that involved in yet. They have a giant $200 billion market cap because they were able to just harness that little the FUS reactor at the heart of marketing, which is they could get installs regularly and repeatably. And that's where the value came. So starting with the buy side and some of this is a bit rudimentary, but it's worth talking about. What we're trying to do is say, well, you have media exposures and you have outcomes, which is the things you want. And now we have to create a structure by which we're going to measure and deliver this. And it always has this problem of correlation or causation. In the traditional digital conversion funnel world, there's two kind of categories of data in between these things. So the first category is proxy metrics, which everyone should be familiar with something like a click or viewability measurement or attention, which Mark from Adelaide is at a breakout and he'll be arguing about this. But fundamentally we're saying, well, there's something about this ad exposure that is measurable and therefore should be part of our understanding of how it generated outcomes. But then there's another layer which is identity, which is who was the person? Because ultimately until agentic happens, it's still human. Beings who are buying these things and downloading these things. And so the question is, who was the human being? Is a second layer of abstraction where you're saying, okay, I have a outcome like a website conversion and now I have to look up who this person was or my best guess about who this person was and then I have to correlate that to what happened in the media and try to make a story out of that. And I think we're all at this point fairly well versed in the problems with this. I'm not saying it's terrible, I'm not saying it's great. I'm just kind of, there's, you know, like Attribution 101. There's really two approaches to solving this problem very writ large. The first one is experimentation. And this is the gold standard where you run science on the problem and try to figure out whether the outcome really happened. And there's a lot of variation on this. I think we have Olivia up on stage from House who's going to go through a lot of experimentation results that she's worked through. I'm really excited about that conversation. For a while there was this a little bit of a fad called ghost bidding where you would buy some media and then you would simulate buying other media that you didn't actually buy and then compare the results. So you had a testing control. Ghost bidding in my opinion, has really lost its luster because it was so dependent on cookies to be able to identify the users. So the user part became a real problem with that. Geotests, I think Bliss will be talking about this in their breakout, is very popular right now because it is kind of the lowest common denominator. If you can recognize where someone's geographically based, you can hold them out. A lot of other options, time based blackouts that, et cetera. The other general approach other than experimentation is AI or really it's a better correlation. So if we accept that we're just doing correlation anyway, you can do better correlation. You can use more inputs, you can have data from the E commerce stores, from the so called C API conversion APIs that, that are proliferating with the major social networks. You can model this to death and get a better sense of what's working, even if it is effectively correlation, which allows me to show my favorite comic ever, which I'll just leave here for a moment. So switching from the buy side to the sell side, we have media and media I think is not really, really well prepared for this outcome shift. And I want to talk about why and how that's changing. So on the media side, obviously we have the tech giants, meta, Google, you know, there. And they have these inherent advantages in producing outcomes which are scale. You know, having more scale obviously is a big impact. They know the user, the identity of the user in much better granularity than other folks. They have this AI investment that dwarfs what any media company can do. And they also have this closed loop. And I think this might be the most important thing is that when you're using Facebook or meta, I always call it Facebook, but meta, you're going from creative to targeting to outcome measurement in one system. You're not trying to stitch together multiple systems. And in contrast, everyone else has the opposite. These are weaknesses for your average publisher or individual app provider or even ad network to some extent. And so what I'm hoping maybe is an optimistic presentation is that some of this is changing and there's a mindset shift underway partially because of AI, but partially just because of changes in our ecosystem where agentic advertising has agentic developments. Let's say that has an opportunity to bring scale where it doesn't exist previously. Because the problem with scale is you're, you're fragmented across many, many publishers. If you have a AI agent who's doing all that buying and negotiating, that scale problem goes away, right? So optimistic, but possible that the scale problem could be going away with technology. You know, the identity I had to bring up the, the other thing that changed since last March at your life that we don't have, we don't have the looming sort of Damocles over our identity AI. Another thing that's changed since the last architecture is that, you know, there was this sense in AI that it was going to be, you're going to have to invest $10 billion to have a model. And now it turns out that many models might be tiny and might be commoditized and that may not be a competitive advantage to have the best model. And probably the biggest change that's going on is that we may have abandoned this buy and sell side separation that's been dominant in our industry and our conversations. And that may be a good thing because if you have the closed loop, you can do better results. Let's talk about some of these things now. Apologies to those of the pre bid summit. I'm reusing some slides. I'm sorry, I don't have a job. Even I don't, I can't just create slides all the time. So the typical, like when we talk about the web and the web is like, emblematic of all those problems, which is you have journalism, you have technology, you have, you have user discovery, traffic, then you have advertising, and it feeds the journalism. Side note, I gave this slide at a talk at Cornell Business School last week, and none of the people there knew who those journalists were. Zero. And no one knew who the Mad Men were either. Old. I'm old, you're old, we're all old. Okay, so this is the typical. And this is obviously just absolutely getting destroyed right now. So AI is radically changing the way consumers surf the web, find information. AI is changing ads in a way that we still don't really understand. It's like very lumpy. Is it creative? Is it media planning? Is optimization? Is it agentic? The content side is in the phase where we're not really sure what's going on here. We're in this phase where, like, if you suggest that AI is creating journalism, someone will murder you. But that doesn't mean it's not going to happen. I mean, you have to say, well, you know, Sora too, people were watching videos created by AI and they seem to like it. I think we have to get out of our mindset and saying there's like, you know, good and bad. It's actually, I think it's inevitable that some portion of the content people read and watch will be AI generated. And obviously it's going to radically change the way that content is presented and pushed to users. Right. So everything's changing and are these things good or bad? And what's the outcome is a big question. So starting with the traffic, there's a lot of examples. I just pulled two. So on the right side, similar web says traffic to websites, down 30% on CNN, 40% HuffPo, that's bad year over year. And that's on top. That's this year based on search changing AI. But that's on top of about two years where the referrals from social went into the toilet because of changes in the policies at Meta and at Twitter. So it's compounding less and less traffic going to websites. The left side here, Pew Research shows that when someone was presented with an AI summary, you can see where it says 15 on the. It says 15% of users who saw a search result without an AI summary clicked on the link goes down to 8 when there's an AI summary. So it's halving the traffic from search. Basically, that's kind of from the consumer side. So that first thing, the discovery of content AI is just whacking it. But on the Other hand. And I picked these almost entirely at random. I was just like googling various data. There are positive signs everywhere about consumers engagement with digital advert, digital content. Spotify hitting records. Pinterest hitting records. New York time hitting records. Podcasting hitting records. Consumers aren't paying attention to what I just said on the last slide. They are just enjoying digital content because that's what they do. And many of these things, not New York Times subscribers, many of them are ad supported. And if you look at the bottom left graph, I think this is mass and wall data. Advertising growth, digital advertising growth year over year. It's always positive. It doesn't go below that zero line. It's growing. And the middle one, time spent on digital is growing. So what does all this mean? It means that consumers are, I'm going to skip ahead. Consumers are moving to kind of what I call multipolar world where they're seeing their content from all different sources of different types and it's monetized by advertising. And that is actually kind of a pretty big opportunity because we're going from the old world, which was where you had one gatekeeper, effectively Google, to a world where there's a lot of different things going on and the marketer, the advertiser, needs to look for their outcomes in all these places. The ad tech vendors need to try to enable those different experiences to produce advertiser outcomes. And publishers need to figure out how to get their audience in different ways. It's a lot that's just changing, which is kind of a theme. I picked this, this image from the destruction of the Berlin Wall because I think it's kind of interesting, you know, you'd say, well, you know, who would have thought at the time that like Poland would be the fastest growing economy in Europe, which it is for the last several years. But also you have Vladimir Putin. So, okay, so I talked a little bit about scale, scale buying. I actually had this slide from like at least 10 years ago and it's still true, which is why do ad networks exist? I often get asked this question, usually on Twitter by someone who doesn't know anything. But fundamentally, from a buyer's perspective, the more sites, the more media I buy, more types of media, the more expensive it is. There's a cost to buying a given piece of media in overhead. So the overhead goes up as I increase the number of sites on my plan. I know that's old fashioned terminology, but it works for performance too. And this is a direct quote I got when I talked to some big media buyer When I was doing research on this subject many years ago, our job is to buy big sites for big brands and that's not conducive to the fragmented environment. When your competitor is, is, you know, applovin or your competitor is meta. And the publisher side has the same problem, which is, you know, I only have so many sellers and they're very expensive. Right. I can only reach, you know, a big publisher may only reach 300 advertisers at the most. And the rest of them, the rest million, they just don't talk to. That's why ad networks exist, to solve this problem. And they're not going to go away. Right? Because this problem's not solvable unless AI solves them. So, so ad CP got a lot of hype about a week or two ago when they launched the idea. It's a lot of ideas. But one of the big ideas is that you're going to have agentic agents that will allow a buyer to say what they want and negotiate with many publishers and potentially book ads in many publishers. And they say right on their website, it's built for outcomes. So we have to believe them. Right. Says right there. So this is maybe this is the glimmer. It's not reality. I'm not vouching for this at all. I'm not that involved in it. But this is the glimmer of a future that may overcome some of those hurdles I talked about with publishers in this kind of fragmented, difficult to create outcomes environment. Right. There's things happening that are interesting that give you a glimpse of the future. All right, let's talk specifically about programmatics. This is another unfair example. So I took the Applovin market cap and put it next to the trade desks and that doesn't look so good. So. When we talk about the history of ad tech, which is my favorite subject, you have this kind of funny situation. And we go back in time to the pre ad exchange world. And the pre ad exchange world was really dominated by ad networks Blue, Lithium and Interclick and all those names the many of us old timers know. And what they were doing was effectively in the context of this presentation, they were creating outcomes for advertisers. That's what they were doing. And then they would keep as much margin as possible for themselves and give publishers effectively $0.01 more than the publisher demanded to keep them on their website. And it was really a great business for them and it was pretty good for advertisers. And publishers always get screwed. So it didn't matter. And then the ad exchange showed up in 2005, 6, 7, 8, depending on how you look at it. And you ended up with a sort of split and in my perception you had the yield managers who work for publishers, they turned into exchanges eventually and they still were in the business of optimizing for the most part just like giving publishers $0.01 more than they demanded. And then DSPs and old fashioned term ATDs agency trading desks emerged and they effectively their motivation was entirely just to capture that margin the ad networks had previously. So they're basically ad networks without any inventory was, was what the early DSPs and ATDs were. They weren't really software companies at the time, but that was, that was an improvement already because they could access more inventory and because they were no longer in the waterfall, everything became a lot more liquid, etc. Then header bidding was invented and deployed very widely and it had some really weird interesting effects I don't think we talk about enough. The obvious effect was that anyone could get access to publisher inventory in a sort of competitive way versus Google which had previously dominated it. And it had this secondary effect that everything moved from second price auctions to first price auctions. I don't think people really predicted that was going to happen. And the reason was because just if you're not that familiar is like that there would be no reason to bid into a header on a second price basis because you wanted to win the headers, you wanted your bid to be as high as possible. So the end result was because it was moving to first price and because these companies were now kind of arm's length with each other, we got into the situation where both sides got this demand for transparency. Publishers said to their SSPs, you can't have hidden fees. And back then people were removing fees. It was big news. When you know, Magnite, I think at the time, they're still called Rubicon, removed their buy side fees and then DSPS were being hammered by everyone saying oh you have to be more transparent, you might be more transparent. And both sides felt like they were in this world where they were becoming true software companies. They were going public, they were talking about that. And that sort of has now where we are now sort of collapsed on itself where it turns out that on the sell side there's really no reason to be transparent because once again you want to win the header auction. Who cares how much money you're taking? Who cares what your fee is if you're not winning? Right? If you take too large a fee and lose who cares, right? The only thing that matters is that you win the auction and you Give the publisher $0.01 more than they wanted. And the way you do that is by manipulating the auction. That's what Google taught us. Basically, if you want to win, you put in more data or you get more advertisers or you do direct deals or do whatever to win and the publisher makes more money. They're your customers. Why wouldn't you do that? And on the buy side, the counter reaction is the buy side stopped trusting the sell side, so they started getting their own supply. And now every leading DSP has its own supply to different degrees. And what ends up happening, I think is we end up back where we started, which is, you know, basically the buy side has a choice. I want outcomes. I could go to one of these 10 companies. Some of them are full transparent companies, some are not, some are, you know, applovin style performance companies. Who cares? Do what you want, it's your budget. And on the sell side, I just want $0.01 more than the other guys. So I'm just going to take demand from any of these partners and whoever gives me the best demand with quality wins. And that's another way of saying we're back to outcomes. And this is good, this is a good thing. You could produce better outcomes if you have the buy and the sell side in one system. That's what Google taught us also, right? That AdWords to AdEx, that combination buy, the sell side gets better results. That's why. For a lot of reasons. But one of the reasons is because they have enough data to make it work. The open world, like if you go up to the third row here where we had this dream of SaaS, products that were totally transparent on the buys. On the sell side, it turns out that's a good idea. But it is less competitive. Effective effective, competitively effective at producing outcomes for advertisers, which is what we're here for. All right, last thing, I just have to talk about antitrust because it's the only thing I talk about. My wife won't listen to me anymore when I talk about it. So you folks need to listen. This is Judge Brinkama. The remedy trial is over. Well, they have, they have closing arguments in two weeks. Effectively they're debating what to do about adx, what to do about Google. I just want to kind of give you some bullet points about how this relates to outcomes and then I will hand over the mic. So this is really complicated. There's a lot of moving parts. But here's the bottom line, the first one is the ad server ad exchange combo that Google has built is going to break down in some way. I mean, I think everyone agrees on this. They have to let publishers free to some extent. This doesn't really have any effect on advertisers. It's good for publishers. They need the ability to choose vendors more freely. Whether it'll be a spin out or not, we don't know. This is probably the most important point. ADX is not a real product. ADWords is the product. ADX is just a little shim that gets the ADWords dollars to publishers and forces them to use the publisher ad server. 70% of ADX is actually just AdWords. So even if AdEx was shut down, that AdWords demand is the real thing we're looking for and that is entirely an outcome machine. It is taking advertiser dollars and spreading them across open web and YouTube and all that other stuff that's going to exist no matter what. There's no chance the judge turns that off. She might move it around, it might say, oh, you have to bid on magnite with that AdWords demand. But it still exists. Google remains a leader in the outcomes world. If adx is dismantled or spun out or reduced, they have a 20% take rate which is very high. It will probably go down and that will probably be good for everybody. You know, ROAS will go up, publisher outcomes will go up. That's all good. And lastly though, and this is a kind of cautionary statement is some of these effects could hurt Google's outcomes business. Or another way to say that is ROAS might go down because they've built in all these vertical integrated advantages to their stack and chopping those up will just reduce efficiency, which may reduce their effectiveness for advertisers and likely will reduce payouts to publishers, which is a shame and sort of the opposite of what we wanted to accomplish here. All right, I went through a lot, but what I'll say and I love, I love my AI generated images, they're just so sweet. You know, measurement is a big factor here. We have a lot of speakers today about measurement as it relates to outcome and it's changing. Media is changing a lot. And I'm, I have, I'm optimistic. I think there's a lot of opportunities in media right now. Probably not, you know, journalism on a website, probably other things, but there's a lot of interesting opportunities and outcomes are king. Keeping that as your North Star gives you the opportunity to produce great results.