← Marketecture: Get Smart. Fast.
Marketecture: Get Smart. Fast.
Episode 123: Olivia Kory from Haus tells us why YouTube is underrated. Plus saying goodbye to Xandr Invest
00:49:16
Loading summary
Transcript119 lines
- [00:01]
Sponsor Voice
A word from our sponsors. Headed to can. So are we. With the addition of Audigen, Experian is now your end to end solution for data driven advertising, helping you understand your customers, reach them across channels and measure what matters. The result? More relevant experiences for consumers, more revenue opportunities for publishers and better performance for brands. Book a meeting with our team at experian.com can that's experian.com/c a n N E S.
- [00:41]
Ari Paparo
This podcast is brought to you by Adelaide. Media verification and measurement are undergoing major disruption. Legacy players are pivoting to performance. Advertising AI is reshaping brand safety and attention is replacing viewability. Adelaide is leading the shift with au, a new way to assess media quality that scores placements based on their potential to drive attention and outcomes beyond. Before your ads run, think of it like a credit score for media. Finally, a clear view of quality. Before you buy, take the guesswork out of your investment strategy and try Adelaide AU on your next campaign. Welcome to the Market podcast. This is Ari Paparo. I'm here with Eric Franchi and our guest this week is Olivia Corey, who is the Chief Strategy officer for From House. They are a research company. They measure effectiveness of ads and they've been publishing some really interesting reports and other things on LinkedIn, et cetera. You probably have seen Olivia out there. Eric, have you been following the work of House?
- [01:49]
Eric Franchi
Yeah, absolutely. And I've been following the work of Olivia specifically. She pops up all the time with really sharp things to say, particularly about some of these topics around like Applovin and Incrementality and all of these like very of the moment topics that everybody's thinking about. So it's great to have an expert that we can talk to about these topics.
- [02:09]
Ari Paparo
Yeah, and I love measurement. Measurement's a fun topic, so we're just going to dive right in. We also have a fun little story about everyone's favorite startup, Quibi. So let's hear it. All right. Welcome Olivia Corey, the Chief Strategy Officer for Houzz.
- [02:25]
Olivia Corey
Thanks, Ari. Thanks, Eric. Good to be here.
- [02:27]
Ari Paparo
So before we started recording, you reminded me about the last time we met, which was a little of a funny story. Would you like tell the story or should I?
- [02:36]
Olivia Corey
Oh, I would love to. It's one of, it's a career highlight of mine that when I was at Quibi, you know, we had come over from Netflix and so the mobile app ecosystem was, was kind of brand new to us. And so we set up a bunch of meetings with vendors that we thought we could learn from, figure out like what our tech stack should look like. And at the last minute, meg Whitman, our CEO, of course famous from eBay and HP, asked to join. She said, I want to learn about this too. And so we brought Meg to New York and we came to the Beeswax office and you all taught us about what's new in DSP buying. It was an interesting altitude because we wanted to figure out our tech stack, but also we had Meg Whitman with us, so it was really hard to program those meetings. That was pretty challenging, but it was a blast.
- [03:25]
Ari Paparo
She was great. I'm one of the few people who could say, I got a demo of Quibi from Matt Wigman on her phone. So I kind of. I liked it when I saw it. It kind of was cool. You demoed like a narrative piece of fiction that changed based on the way you held your phone. Like, the orientation. It was some pretty innovative stuff. And I think there's a selfie of me and Meg somewhere, but I don't have it.
- [03:49]
Olivia Corey
Beeswax was so cool coming from Netflix. They were trying to bring their own optimization, like bidding algorithm to these DSPs, and I think that's what you all are famous for, right?
- [04:02]
Ari Paparo
Yeah, that was what we were famous for. So it didn't work out between us, but it was a good meeting and a good story. So I asked you on the show because you are dropping knowledge bombs regularly on LinkedIn and on Twitter and you're doing all sorts of interesting stuff. So let's just start with the basics. What is House and what's your job there?
- [04:22]
Olivia Corey
Yeah, so House, I'll give a little bit of the backstory here and then I'll explain what we do, which is that, you know, a number of us were at Netflix, and as most of you all know, Netflix was running experiments all over the world. They wanted the product to really work or the marketing to really work. Like the product where it's kind of personalized, algorithmic. No two people see the same Netflix. They wanted to take those principles to marketing. And so we had a huge team of data scientists and causal economists and product managers dedicated to helping us run these very kind of rigorous experiments. We were able to bring our CAC down significantly in the time I was there through kind of continuous experimentation. And the insight was really like, okay, this is valuable, but it's super expensive. So what if we could enable access to that level of. In that quality of experimentation, but do it as SaaS for less than the cost of a headcount? And that is what we set out to build and so what the platform is really is the experimentation incrementality platform where any brand could log in and they could configure an experiment, a GEO experiment across any channel, online, offline, meta, Google, usual suspects, but also TV out of home radio. And we do all the data science work to say how much should you spend, how long should you run the test for? How large does your holdout group need to be in order to kind of sufficiently power the test from a statistics standpoint? And then because we're plugged into the data, we can report on the analysis in the data. And so we've just kind of like attached a jetpack to these, you know, the experimentation practice of these brands who want to do more incrementality testing, but they were never really able to get it going.
- [06:00]
Ari Paparo
Yeah. So how, how is that possible in today's fragmented media world where we don't have identity? You can't measure the same thing on YouTube that you can on open web or on, you know, wherever. What's, what are the tenets of how your system enables that?
- [06:16]
Olivia Corey
The product is GeoLift. When we were kind of like born from the ashes of iOS 14, the company was founded in 2021. And so when folks were figuring out alternate identifiers and ways to continue to stitch together these user graphs, we skated in the opposite direction and said, we're going to ignore all this and we're just going to build on Geo. This will persist through the future. And so that's the way in which it's apples to apples across every channel is we say turn this marketing on in these regions, turn it off in these regions, and we're going to measure the sales lift to your business based on the data coming from your back end warehouse.
- [06:51]
Ari Paparo
Yeah, GEO is lowest common denominator. Eric, you might remember a conversation you and I had like three years ago before we started the podcast about starting the Geo DSP with the exact same idea. And I think I have bliss coming on this podcast in about two weeks. So we'll hear what they did with geo. But I'm a big GEO fan, so let's go to the fun stuff. Some of what you found over the last series of investigations and studies. Let's talk about YouTube. We don't talk about YouTube enough on this podcast. Overrated or underrated?
- [07:22]
Olivia Corey
Oh, it's way underrated.
- [07:24]
Ari Paparo
Underrated. Okay.
- [07:25]
Olivia Corey
Way underrated. We, and this is obviously a prior I came in with, coming from Netflix, where we were spending a lot of money in YouTube. It worked really well for us but when I started working with these, you know, we work with a lot of E commerce brands who are heavy in meta, very concentrated in meta in YouTube is, you know, maybe not even on their, their plan or small share. And so I came in with a prior that brands were underinvested. And so we looked back at 190 incrementality tests that we'd run over the past year so and YouTube was underreported by, I don't have the stat top of mind. I think it was like 3.4x. So when you're looking in Google Ads at the effect in your CPA is it's underreported by 3.4.2x. There's a lot more there than brands think when they're looking at the clickbase model. And it's just so fascinating because all of these, these brands who are looking at MTA just think that YouTube can't work. It's like no, no, no, just like move off of attribution. Like people are not clicking on this media. It's mostly happening on TV screens in terms of the exposure. So it's a classic video story of if you're measuring this on clicks, you're not going to see anything. Doesn't mean it's not working.
- [08:36]
Ari Paparo
Yeah, people don't click on the videos generally or on the ads. So when you say underreported just to dive into that, you're talking about Google's own reporting is underreporting its most valuable media property.
- [08:46]
Olivia Corey
Isn't that so ironic.
- [08:49]
Eric Franchi
Because of the click issue?
- [08:52]
Olivia Corey
Yeah, I mean Google Analytics and all their measurement systems were developed to favor the click based conversion. And so in doing so and focusing on clicks, you're actually, you know, way under reporting.
- [09:06]
Ari Paparo
YouTube and Meta is happy about this because Meta's ads are much more clickable. Like Instagram ads are sort of meant to be clicked, is that right?
- [09:14]
Olivia Corey
Yeah, meta's, you know I would say they over report but most brands are not really looking at platform data these days anyway. They're looking at an MTA or some sort of third party like objective source of truth.
- [09:27]
Ari Paparo
So do you think video in general is under reported, under credited or is it a YouTube specific phenomenon?
- [09:35]
Olivia Corey
I don't have the data on other video. The OTT space is so fragmented that it's kind of hard to see okay, this specific OTT media is doing this in terms of performance. So we haven't done the analysis on. Okay, what are we seeing in the OTT world which publishers are doing better than others but we tend to see anecdotally that video is underreported.
- [09:57]
Ari Paparo
Right. Any other insights on YouTube about what marketers should be doing there besides measuring it? Obviously measuring gives you a good basis, but any other thoughts about what they should be thinking in terms of targeting and outcomes?
- [10:12]
Olivia Corey
We were all, I think, skeptical when they started transitioning over from video action to demand gen because it felt like PMAX all over again where it's like, give me some assets and a budget and I'm going to decide where to spend your budget. But we're actually seeing some good things here. Demand gen has been outperforming video action, so I think that's just a good, it's a good sign for marketers that like the, the way in which Google is moving us is actually probably going to benefit them in terms of outcomes.
- [10:40]
Ari Paparo
Would you mind explaining those two terms for the folks on this who are listening who are more like ad tech geeks and not social media geeks?
- [10:48]
Olivia Corey
Yes. So demand gen is like basically think of it as analogous to pmax. It's like this new, I think they call it demand gen because it's a little more like mid funnel than lower funnel and it's delivering across all of Google's O and O, but primarily video and not so much search. So they started migrating over to this of just like, give us your budget, give us some assets, we're going to decide where in the Google ecosystem this gets spent to best deliver on performance. And brands were so mad, they're like, oh, this is PMAX all over again where I am losing control. I don't have the ability to segment out where I want these to deliver. And so they made this change at some point this year where they deprecated video action, which is YouTube. It's like, I want to go in and buy YouTube for performance. Instead you have to buy Google demand gen, which runs across this whole swath of inventory. And advertisers were pretty mad about it, but it's actually working well, which is a good thing.
- [11:48]
Ari Paparo
And also it seems to me like given that the sort of studies you're doing take time and, and aren't always on for many advertisers they, they may, and tell me if I'm wrong about that, but they may actually be doing themselves more harm than good because if they're using, let's say DV360 to buy YouTube and they're optimizing on the wrong thing like clicks, they may be harming their performance in, in real world outcomes. Does that make sense?
- [12:15]
Olivia Corey
Yes, I agree with that. I think this probably moving in the right direction.
- [12:20]
Ari Paparo
What would you recommend they actually do?
- [12:22]
Olivia Corey
You know, I think that these platforms are like really good at doing what you tell it to do. You know, give it a signal and I'll go find it. And so in Meta is notorious for this of just like whatever event you go tell it to find, it will go find. And so that's my advice for these marketers is these systems are designed to go find whatever you want, but you have to validate that it's truly incremental and they're not just like knocking off or picking off low hanging fruit that was coming in anyway. So you have to be very thoughtful around the event that you feed it.
- [12:53]
Ari Paparo
All right, let's move on from that to everyone's favorite topic, Applovin. So you did some studies on Applovin and Applovin, as everyone probably knows, has been the most extreme success story, I think in digital advertising over the past six months or a year. Their stock skyrocketed, the results are skyrocketing and a lot of skeptics have come out with different degrees of aggressiveness on their skepticism. So you did a study on Applovin and it caused some controversy. But let's just start with what did the study find.
- [13:23]
Olivia Corey
Yeah, we've been running Applovin tests pretty consistently for the past maybe nine months or so. And what I would say is that these brands that we work with are dying for a strong number three partner. I mentioned this up top. Like, they're very concentrated in meta. They want to diversify, they want to be less reliant. So they have Meta and Google and then they just have like a bunch of partners hanging out at number three with like very small shares of spend that they can't scale. And so, you know, I came in pretty skeptical. Just we all did in terms of like, all right, how is this differentiated from anything else you can access through a dsp? But they were, they, you know, they gave ad credits to these e comm businesses, credit to them that they got them testing and they actually got them to try it. And so we started running holdouts on it and we saw pretty clear, discernible lift. Like, the story is broadly pretty positive. Now the question is how much can it scale? Just like every other number three partner, it's like you can see very strong lift at these smaller, lower spend levels, but how much can you really push in? And that's where we're at right now, where these brands like what we saw in the Data was really strong efficiency, very clear lift at these lower spend levels. 5, 10, 20k a week. But when you start trying to punch up to like 50, 60, 70, 80k a week, efficiency is, you know, it's, it's falling and it's like, okay, well then does it make sense to keep moving dollars from Meta or from Google? It's like this next best dollar is not as efficient. Is my first, you know, my first 10k. So that's where we're at right now is like, I was just, you know, as somebody who wants to help brands grow, I was excited about the promise of a new channel for them and the ability to diversify. But the question and the open question is like, all right, how much can this really scale?
- [15:13]
Ari Paparo
So that seems like pretty milquetoast sort of research result. What happened next? You sort of turned into, you got QAnon more or less by these lunatics online that somehow thought that you were the devil incarnate for publishing this research.
- [15:28]
Olivia Corey
Yeah, they think that I'm like a Apple stock pumper. They call me a stock pumper, which is never, I've never owned any stock in Applovin. And again, I was skeptical to begin with because like I came from this world of dsps and I know how hard it is to really differentiate. So it's, it's hilarious in that the, my motivations are so, so transparent here, which is like, we want to help brands find places to diversify their spend. And so when we see something that has promise, we're of course going to talk about it because we, that's, that's a great story and it's a great opportunity for these companies to diversify. So it's been interesting in terms of the things I've been named on the Internet.
- [16:07]
Ari Paparo
Okay, let's talk about the future of measurement. Right, so geotesting we've talked about, is there a role for things like media mix modeling and multitouch moving forward or is the overall trend changing?
- [16:21]
Olivia Corey
We feel pretty strongly and kind of our North Star at house is that experiments are the best way to establish ground truth. So everything that we do will continue to be kind of anchored on experiments. And again, this is like the idea of causality. Like we, we have a holdout group, this group of markets or regions where you're not spending so that we can understand the counterfactual of what would have happened anyway. This is like the core of what we do and what we've built our product around in terms of. Mmm, it's really hard for A model to, to establish causality, to like disentangle. Because when you're spending up, that is often at the same time your business is going up based on seasonality. In Q4, you know, you're sonos, like it's holiday season, you're going to spend up on every channel and your business is going to go up. And it's really hard for a model to disentangle what's driving what, like what is cause and effect. And so we, and we're moving into the MMM market, like we think experiments are the way to of fix this problem, right, Is if you could feed experiments into the mouth and build an MMM on top of this experiment data. This is actually a very useful tool. I actually don't think it's like experiments versus mmm. I still think experiments need to be the foundation here. But an MMM makes it a lot easier to do things that are kind of manual right now in house, where it's like budget planning, like scenario planning, asking what if questions of what if I move this amount of money From Meta into YouTube, what would happen and what does that return curve look like right now it's just a little bit clunky to do without like a nice slick kind of ui. And so that's why we're entering MMM is to just make some of these questions like a little more turnkey to answer in like a clean kind of slick ui. But yeah, this is this. The root of everything we do is going to be continued to be founded in experiments.
- [18:10]
Ari Paparo
And there used to be a big thing around ghost bidding that was a really hot topic. My take was the ghost bidding, which was the idea that sort of fake bid on a holdout group that, that died when the cookie died. Is that your point of view too?
- [18:24]
Olivia Corey
Yep. Yeah. We used to love ghost bidding and conversion lift at Netflix though. This would have been, you know, pre2020 and iOS 14. But the privacy changes have like really challenged user level Lyft studies. Like again, we loved them. I think they're more, they're like much less noisy than geotests. You know, in a perfect world you're running these user level studies and they're a lot more accurate because they're just more precise. But the privacy changes have kind of made them like, you know, very, very difficult, very challenged. And as an example, Google's conversion lift, I think they have to model all of iOS like it just doesn't work with iOS. And so then you start to, you know, really question the, you know, how well they can model it and so, you know, I used to be a user of Conversion Lift, but it's tough.
- [19:12]
Ari Paparo
Yeah, I think it is. It is pretty challenging moving forward. All right, well, let's call it there. That was a really interesting conversation. We definitely lear some stuff about YouTube and different methodologies. We will take a break and we will come back with the refresh news of the week. This podcast is brought to you by Bert Intelligence. Take back your Mondays with Bert Intelligence, the publisher bi platform that turns chaos into clarity. And now with their new product, Bert leads publisher sales teams can uncover hidden gems, recapture lost opportunities and upsell deals with Intelligence. Stop guessing and start closing. Sign up during May for an extended trial at Bert AI ADTechGod. That's B U R T AI ADTechGod.
- [20:05]
Eric Franchi
All right, welcome back, everybody. And we are here with the refresh. So another week of a bunch of news. We'll start with Microsoft DSP being shut down that was announced yesterday. And then we've got takes and news out of the Luma DMS conference. An acquisition, an IPO valuation coming in adtech and a bunch of other stuff. We have time. So let's start it from the top. Yesterday there was announcement that The Microsoft Invest DSP is being sunsetted in end of February 2026. There were a few takes about this and to be honest, the post is quite confusing. We'll link to it in the newsletter.
- [20:48]
Ari Paparo
It is very confusing.
- [20:49]
Eric Franchi
The title of the post on the Microsoft blog is Empowering Businesses for a Future that is conversational, personal and agentic. And it's flowery and long and clearly written either by or with the augmentation of AI. But in the middle of it there is a line that says, as a result, we will no longer support media buying through our DSP Microsoft Invest, starting in February 28, 2026. Ari, you had, I think, a very sharp, nuanced take on this. Let me get yours first so then maybe we can talk about some other things that people are saying.
- [21:22]
Ari Paparo
Yeah, I'll just start by saying this. This blog press release was, was like, it was a Google level quality for obfuscating what they're really trying to say.
- [21:30]
Eric Franchi
It was a tough read.
- [21:31]
Ari Paparo
Usually Google gets to the point in their third paragraph and you had to go to like paragraph five before you got to the point of this. So, okay, so the quick, easy take is as follows, which is Microsoft's advertising business is killing it. We talk that in a previous episode. That's like $40 billion. It's being driven by LinkedIn and a lot of Other really great proprietary supply and they have demand coming in from effectively three sources. They have their own ad manager called the Microsoft AD platform, like AdWords, you log in and you buy stuff. They have the Xander DSP and then they have third party DSPs like the trade Desk and people like that. So of those, the Xander DSP is probably the smallest, growing the slowest. It probably has the lowest margin overall take rate when they do that because it's expensive to main and probably in order to make it compete, it would require a lot of investment and it's unlikely it would be that successful. So just pure business considerations, shutting it down probably has very little effect on the metric they care about, which is their margin on their O and O ad inventory. And so unlike Google, who's big enough to support Google Ad manager and DV360, there's not a really compelling reason why Microsoft needs to have a dsp. That's kind of the very short, easy answer. The longer answer, if you take what they actually said in this long blog post for real, is that maybe they're thinking that a closed system will be better in an AI era because the closed system can share data more freely, like data about the person's LinkedIn profile if you're running on LinkedIn. And that's the kind of data that you can't send over the wire to a programmatic source of demand. Now, they are still going to be available in some form to users of Yahoo DSP and tradest dsp. So they'll still be available to some extent, but they clearly seem to think that they'll get better results, better yield and better margins from their closed loop system, which is another piece of evidence that it's not a great time to be in the sort of open web, traditional programmatic area.
- [23:43]
Eric Franchi
When I read it, it's 90% about agentic AI and the future of AI and a little line about this and what it means. This also comes during a week where they announced they were laying off 7,000 people. And reading between the lines, a lot of this was about AI. So my take was this was aligned with what seems to be happening at Microsoft and frankly other companies as well, which is like aggressively moving towards an AI future, which is probably connected to, you know, what you were saying. But I read this, I saw AI. I did not see this as kind of sunsetting something that had low margins.
- [24:23]
Ari Paparo
Yeah, we also have talked about this, I think on the pod, which is these tech first companies are now AI first and they're light years ahead of Ordinary businesses, even knowledge working businesses in terms of the way they've been able to maintain their cost control while revenues go up. And I think that the more traditional businesses are just starting to wake up to this fact that they could have massive improvements in productivity if they bring AI to every employee or at least every knowledge worker.
- [24:53]
Eric Franchi
Yeah, that makes sense. Bolivia. Does Microsoft crush it?
- [24:56]
Olivia Corey
We do a lot of Bing tests. We have some LinkedIn tests. I have to look at that. Maybe that's a new meta analysis for us. But what's really interesting about Xander, it's called Xander, right?
- [25:08]
Eric Franchi
Microsoft Invest.
- [25:09]
Ari Paparo
Yeah.
- [25:09]
Olivia Corey
Okay. Well, I hadn't heard much about them for a very long time and then Netflix, if you remember, decided to build their ads business on Xander or what is now known as Microsoft Invest, like three, four years ago. And I just remember being so confused. One because Netflix tends to build versus buy. And I just like. Xander was just, it was just like to your point, I think number four, the number five. So that was always a very confusing decision. They don't use them anymore. But I remember being surprised.
- [25:41]
Ari Paparo
I think the scuttlebutt there was that Microsoft guaranteed them a minimum CPM that was extremely high to get the deal. And because the ad operations were sort of rushed out, Ted Sarandos really wanted to juice revenue in the first couple quarters. So they took advantage of Microsoft's desperation.
- [25:59]
Eric Franchi
The legendary $65 CPM.
- [26:02]
Ari Paparo
$65 globally. So you had to get that in like Korea and Vietnam and places like that.
- [26:07]
Eric Franchi
The CPM heard round the world. I'm sure this is going to have some further implications. So this is not going to be the last time we talk about Microsoft giving their shifts to the ad tech business. Let's move on. So Tuesday of this week, the DMS by Luma conference happened in New York City. Both Ari and I attended. Shout out to Terry for the invite and putting together what was like a really good conference.
- [26:34]
Ari Paparo
Great event.
- [26:34]
Eric Franchi
Great event, Terry. It was really, really good. Yeah, Terry, great job. Let's talk about what we saw, what we heard there. Why don't you go for us, Art?
- [26:43]
Ari Paparo
You know, it was good to see so many senior people. Jeff Green spoke at the end of the day. I thought he was pretty forthcoming. Krishan Bhatia, who's now the head of video at Amazon, it was good to see him. I'm trying to get him on the pod. So if you're listening, Krishan, or waiting for you, it's just great to see executives talking and talking about like what they see moving forward. I don't have anything like meaty to say here. It's just, you know, a good event.
- [27:09]
Eric Franchi
I got some meaty stuff. Let's talk about two things. So they, they kick off the day and I've been going these for very years. Again, you know, super lucky. Thank you, Terry. With their state of digital and this one was really well done. If it's not posted publicly yet, I'm sure they will do so. And there were two things that I thought was interesting during that presentation. The first was they talked about AI. They talked about AI a lot during the day and throughout the sessions. And we should get back to the panel that you led, Ari. But they talked about the AI first startup phenomena that we're seeing right now, which is companies launching today that are AI first. They are launching with less capital, they are getting traction faster and they are likely going to get. And this is a LUMA conference, so kind of a little bit of bias here. Likely to get acquired earlier. And this is something that we're seeing in our investments every single day. We're resetting our models based on what we're seeing. I thought that was interesting that they articulated it in that way. Later on in the day there was a AI First Startup panel that you led which had Swivel Stylus Firsthand and Newton Research. A few of them are portfolio companies and I thought it was like underscoring what they were talking about, which was these companies, they're like doing very, very interesting things, they're getting traction quickly. And it really feels like the industry is starting to pay more attention to the effects of AI and what some of these startups might be doing to change the industry.
- [28:50]
Ari Paparo
Yeah, I think those startups you mentioned, they all kind of indicated that they themselves are AI first, that they're being very productive in their use of capital and their use of manpower. And I'm not seeing as much from the legacy folks who are very big, profitable businesses in some cases, but maybe aren't as aggressively rolling out AI or otherwise investing in this area.
- [29:12]
Eric Franchi
Yeah, there was a slide that made its way around the David versus Goliath that I thought captured it. Well, the other thing that I thought was fascinating from that initial presentation was they spent a fair amount of time talking about creators and there was one slide that I was standing in the back and I was like, holy cow. Kind of made my feet shift, which was Unilever said that they are going to 20x their spend on creators this year, not increase 20% 20x. And you know, like I'm, I'm trying to spend more time in this space and I want to ask you know Olivia like what you guys see around tracking like the impact of creators. I talk to marketers and those who are serving them and creators just move product in a way that is like really, really difficult to replicate in any other way. So I don't know if ad tech is going to start making its way into the creator space. I don't know what the play is there but I think it's something we have to attention to. I was like shook 20x in terms of spend. That's a lot of money.
- [30:16]
Olivia Corey
We're seeing this right now where I think Influencer Creator is the fastest growing channel if you, if you call it that on the media plan. It's tricky to measure because you can't geosegment a creator post. And so we're not geo list testing but, but it's you know these, these brands like see very clear pops when a creator is posting on their behalf. So yeah it is very fast growing. You know like I was mentioning like the number three who after Google and Meta, what's like the number three channel? Influencer is just kind of like you know, wrapped up in here in Google and Meta. It's, it's all over the place. And there are companies who are, are trying to make this, this easy. Like there's this company, I think it's called Agentio that's trying to like really scale the, the whole business of, of creators and influencers and make it more scalable because that was my experience when I was on the brand side is really tough to scale a program like that. Especially if you're a global business. You have to go into each individual country and do your research on who to be working with. So yeah, whoever can figure out how to scale this is going to be really well positioned.
- [31:27]
Ari Paparo
In a sense it's like the return of affiliate networks. These are roughly affiliate programs more or less. The creator goes on TikTok and talks about the product and then the person clicks through the TikTok store and the TikTok store is really just a really thin layer on top of someone else's distribution and fulfillment system and everyone takes a little cut. So it's brilliant. It's like the thing that always works is hearing from your friends or from.
- [31:53]
Eric Franchi
Celebrities and maybe that's the way that Applovin, going back to our early conversation could become number three is if they acquire TikTok like Adam seems to want to do.
- [32:03]
Ari Paparo
Who wouldn't want to acquire TikTok yeah.
- [32:05]
Eric Franchi
I'd like to acquire TikToks. All right, make a move. Anything from Jeff Green's talk that you want to hit on, or should we just talk about OpenSensera?
- [32:14]
Ari Paparo
Let's move on to OpenSensera, which is a little bit of an interesting side note here.
- [32:19]
Eric Franchi
So walk us through Open Sincera. Go ahead, Eric.
- [32:21]
Ari Paparo
Well, Trade Desk acquired Sincera a couple months ago. Both Eric and I were involved in that. And Sincera had this sort of blueprint of the web from a advertising perspective. They knew who's in the header, they knew where the ads were, how the ads refreshed, all this really cool data that was pretty obviously valuable to the Trade Desk for bidding purposes, anti fraud, all that good stuff. And then, somewhat surprisingly, they launched an open version of the Sincera data set that lets you download it with an API so that competitors or other people in the ecosystem can participate in the same data. So it's interesting. It's great. I appreciate the free data. I'm going to use it in some projects. I'm not really sure what the business model or if there is a business model here, or if it's just trying to make the data more standard so that. That the overall ecosystem improves. Do you know what the idea is here for making it open?
- [33:15]
Eric Franchi
Well, I know what was said publicly, which is what you just walked through. And I asked Mike o' Sullivan immediately after it was announced he was at the conference. I asked him, I was like, is this actually the opposite of what everybody was expecting? Because what everybody was expecting was Sincera would be closed off effectively from the rest of the ecosystem. It would become some sort of proprietary data set for tt. And he said, yeah, we're going open, which I think is really cool. And to your point, everybody appreciates it. The thing that I've heard in terms of takes is the more that this gets adopted, the more this advances the vision that TTD has of what the premium Internet looks like. Could that be the real value creation for TTD here at the end of the day? Or like Sincera starts to define what's premium, what's not premium, what's implicitly good, bad. And that advances the vision of TTD for the premium Internet, the SP 500. Perhaps that's the case here, but I don't know.
- [34:24]
Ari Paparo
Yeah, if you want to put on your 3D chess hat here. The strategy that Trade Desk has been undertaking over the past 18 months or so, I characterize it as like, your ad tech tax is my opportunity. So Jeff and team have been commoditizing SSPs and they've been commoditizing TV operating systems and they're basically trying to take more of that value chain. And so one thought here is that by giving Sincera data, which is telemetry on quality, to anyone who wants, including advertisers and their agencies, they're undercutting the verification companies who are a big part of the ad tech tax. So you can pay a verification company $0.25, $0.45 to tell you that a given web page is not made for advertising or you can write a little script that takes the open source data, turns it into a whitelist or blacklist in trade desk and pay zero.
- [35:25]
Eric Franchi
Right. And then as long as you're a TTD customer, you can ensure that you are buying that inventory.
- [35:31]
Ari Paparo
Yeah, right. And they could have done that on a closed basis, but making it open maybe reduces the concern among customers about lock in and all that sort of stuff.
- [35:40]
Eric Franchi
All right, cool. The other thing that was announced at Luma, because Luma was the company that did the deal, was an acquisition. So Perrion acquired GreenBiz, which was a little known custom algo company out of France and public company numbers are public, so there's a total deal value of approximately 65 million. I wasn't familiar with green bids going into this, so it looks like it was a sustainability play that broadened its vision to custom Algos. But it seems like super impressive. And I met the founder at the vet, he was obviously really excited and it seems to be like a very interesting acquisition for Parrion And I wonder if we see more of this in the future for these custom algo companies.
- [36:30]
Ari Paparo
Yeah, I'd never heard of them either, but it's great that they had an exit. To date Cybids was I guess the only successful custom algo company to have an exit. So it's great that there's a second on that list. Perion's been known as sort of a serial acquirer of, I wouldn't call it distress, but fairly low ticket acquisitions. Except for Undertone which was the shining star of the 2000s. So I don't know what they're up to. They're like kind of trying to assemble a little RTB stack of some kind because most of their business is in rtp. I don't think it's like search and Undertone is known for premium stuff. So I watch that space, I guess.
- [37:10]
Eric Franchi
Yeah, I think it's something to watch. Perion has over the course of the past few months hired Steven Yap Consolidated, what was to Your point? Kind of a collection of acquisitions into a unified brand. They've got actually some interesting assets. Completely undertone aside this hivestack out of home platform. Like there's some interesting things there. I think you're right. Something to keep an eye on.
- [37:33]
Ari Paparo
Oh, I forgot. Yap was yapping at me at the party. So he told me one thing that he was really impressed with about greenbids was that they effectively had a money back guarantee. Like they were guaranteeing that their performance worked, which I thought was pretty nice way to sell optimization solution.
- [37:51]
Eric Franchi
So we have a couple things we should roll through, just other snippets. So Mountain is seeking a 1.24 billion valuation on the IPO. To Olivia's point, earlier TV and you know, performance TV may be underpriced. So this would be an interesting one given. Given the. All the momentum there. What do you think, Ari?
- [38:13]
Ari Paparo
You know, this comes after a dms. Like three different people came up to me to tell me the Mountain IPO is off. So it shows how much people know. Yeah, I mean, yeah. And also the. I guess we don't have it in our list here, but friend of the pod, Kevin Weatherman, launched his new competitor. What's it called up something.
- [38:36]
Eric Franchi
I think it's upscale.
- [38:37]
Ari Paparo
Upscale. Upscale AI. So now there's another company in this like, small business CTV thing and. Yeah, good for Mountain. I want to see it. I want to see it go public. Let's go for it.
- [38:48]
Eric Franchi
The headline is funny. The headline on Reuters for the story, Ryan Reynolds, Mountain 6211.
- [38:56]
Ari Paparo
I was trying to work a Blake Lively reference into my little talk, but I could. So, Olivia, these platforms, Mountain and tv, Scientific, et cetera, they're all grading their own homework. They're all using pixels and IP address and stuff like that. What's your take? Have you done any research on them? And what's your take?
- [39:14]
Olivia Corey
Yep, we see a lot of Mountain. We see a lot of Tatari and then tenuity, which is something that's formerly kind of Bliss Point. So I see those three all the time. We do a lot of tests. Mountain, I think, is primarily connected tv. They don't really play in a linear space. Is that right?
- [39:33]
Eric Franchi
Yeah, it's mostly CTV and online video.
- [39:36]
Olivia Corey
Yep. So that's super easy to geotest. So, you know, we have a lot of data here. And the trickier channel is linear because you can't just easily. Similar to influencers, you can't just easily convert a national remnant cable buy to like, you know, a local buy. And so it gets. It comes with its own host of complications which we're working through and you know, doing the best we can on. But yeah, we have quite a bit of data on ctv again, this may be my next appearance. Will be bringing some data on these channels that you all are talking about.
- [40:12]
Eric Franchi
That would be awesome.
- [40:13]
Ari Paparo
Yeah, we love it when guests bring the data break news here on the podcast.
- [40:17]
Eric Franchi
Yeah, come with gifts. Hey, speaking of data, there was an article, I think Adweek had the exclusive. I'll read the headline. Group M's US staff told up to 45% will be impacted by a restructure. Sidebar. It's unclear if affected means layoffs, reassignments or something else, but that is a big number on a big company.
- [40:41]
Ari Paparo
Yep, yep. Please meet your new boss. It's a chat prompt@DMS.
- [40:50]
Eric Franchi
I had a conversation with an agency exec who did not work at Group M and he said that, man, if you were to just really get aggressive, you can run that with 2,000 people. And then he thought about it for a second. He's like, I bet you could run group M with 200 people. Which I think is like crazy talk. But to your point, I think that the AI train is not slowing down and we'll see a lot of shifts over the course of the next year or so.
- [41:24]
Ari Paparo
Yeah, I think I talked about this on the pod. I can't recall, but there's generally a feeling that on agency renewals with brands that are coming in 20% lower on headcount for the same business and it's AI, AI is really driving that. It's not just cost savings and that's just the first level. That's not really. That's not really with the maturity of the technology into every aspect of their organization. Also at dms, was it the. I forget the fellow's name. One of the senior execs from publicist was there and he talked about how Penske. Yeah, and they're taking their whole budget from can, not going to can, and instead investing it in Claude, their AI system. It's a very agency like announcement to make, but still it's kind of cool.
- [42:10]
Eric Franchi
Yeah, absolutely. Hey, let's end it on a more positive note. So apropos of Olivia being on the pod, Netflix, their ads tier continues to grow. So they added 94 million. So they're up to 170 million people. Which according to them means that they reach more 18 to 34 year olds than any broadcast or cable network.
- [42:30]
Ari Paparo
Yeah, just to be clear, I think they said they had 94 million subscribers, which is 170 million people.
- [42:35]
Eric Franchi
I'm sorry. Yes.
- [42:38]
Ari Paparo
Yeah, it's big. I mean, I still don't know what sort of freak. I'm not on the ad here. Maybe I should downgrade my Netflix as a science experiment so I can see the ads. Olivia, do you pay?
- [42:48]
Olivia Corey
Of course I pay. And I think that's, you know, once you're used to not having ads, it's really hard to downgrade. But yeah, it's still lower than YouTube in terms of, in terms of viewers. That's, that's the stat I keep seeing. YouTube reference is they're bigger even than Netflix.
- [43:05]
Ari Paparo
Olivia, do you think they ever do the Amazon thing and just put an ad in the, in the paid service?
- [43:11]
Olivia Corey
Like, like that Black Mirror episode. Did you guys. Have you guys watched the most recent episode, the most recent season of Black Mirror?
- [43:17]
Ari Paparo
There's an ad, there's an ad angle. I haven't seen it.
- [43:20]
Olivia Corey
There's the whole. It's like the Eric Suefer. Everything is an ad network. There's a whole episode about how they just weave in these ads to your day to day life. So. Yeah, I mean, they have like some sort of product placement team now, right?
- [43:36]
Ari Paparo
Sure, sure. And also the live. That's part of the live event thing because you can't have a live event without ads. If they start shoving ads into their premium content, there'd be like a revolution. It would be bad.
- [43:47]
Olivia Corey
I have a question for you guys. Going back to the. We should wrap up Netflix, but going back to the agencies laying off 45% of their staff.
- [43:56]
Eric Franchi
Yeah, go ahead.
- [43:57]
Olivia Corey
I'd love your thoughts. Like what I'm experiencing right now is we know AI is going to make folks more efficient and more productive, but it hasn't happened yet. Where it's like the new GPT images, it's like close to being useful in terms of replacing some of the creative workflows in production, but it's not there yet. So you can't fire your designer or like web dev. It's close, but it's not there yet, so you can't fire your dev agency. Are you guys feeling this, this way where it's like this frustrating moment where it's like you see the advancements but you actually can't make the team more efficient. Like there's nobody you can actually let go of in reality?
- [44:35]
Ari Paparo
Yeah. I was thinking about firing Eric from this podcast, but the tests have been inconclusive. No, I totally get what you're saying because I'm a pretty heavy user of these things. And what I'm finding is that it's helping me do work that couldn't have been done otherwise. So it's increasing my productivity, but it's not in any way able to replace annoying tasks or time consuming tasks that are my regular day to day. There's always something in the way of doing that. And I think the image example is great because if I didn't have ChatGPT, my newsletter just wouldn't have an image. I'm not going to hire a illustrator for a thousand bucks a week to give me an image. So now I have an image. And if it's a little crappy, I don't care because it's an image. I'd rather have it than not have it it. But I'm nowhere near the point where ChatGPT is writing my newsletter or organizing my thoughts even for the newsletter, just because it would take more time to do that than it's worth.
- [45:30]
Eric Franchi
Yeah, a couple thoughts here. I would ask you to consider how much your workflow is being augmented today through AI versus two months ago, six months ago and a year ago. I would presume that it's, you know, change things very, very quickly and you're using it a lot more. Right. So it's like number one, we're still early in the cycle, but it's moving very fast. Number two, we're seeing the capabilities, you know, coming online faster and faster. So I think it's only a matter of time before the tools catch up with our expectations in terms of the work. But I think affecting organizations, that's going to be a, that's going to be a more trickier one. Right. It's going to, it's going to take time. Right. Like it's manifesting itself in layoffs. Today it's manifesting itself less in, you know, the AI tools starting to replace and augment things. But Joe Hirsch from Swivel was on Ares Pod and he spoke to some examples. They help publishers and streamers, Automate adopts and account management. And he pointed to some examples where tools are being adopted and they do the work of 40 to 50 people. So I think it's only a matter of time and it's a matter of organizations trying to figure out exactly how they adapt. But I would imagine in one to three years everything looks completely different.
- [46:58]
Olivia Corey
Yep. I feel like we're on the brink of just something massive. But there's like no very clear, obvious, like, oh, I can go replace this team with AI tomorrow.
- [47:09]
Ari Paparo
I think there's also an issue where the skills of the people who are responsible don't match what's needed of them. So you might have a very good middle manager who runs a group in a media company, and that's not the right person to do a business transformation using AI. There's nothing against that person. So that actually comports pretty well with what Joe from Swivel said, because bringing in a new vendor is something that person is good at, and if the vendor can eliminate the jobs, that's great. But expecting that person to imagine using Cursor and chatgpt to write a new, you know, module that does certain automatic things is just, you know, it's not the kind of work that that person is trained to do.
- [47:50]
Eric Franchi
Yeah, such a good point. I think it also then goes back to what we said in the in the beginning with this notion of AI first startups, they have just such a massive advantage of starting today and being able to design their organizations, being able to fundraise, being able to just like be thoughtful about using these tools from day zero. And I would imagine consultants might have a field day helping with AI transformations over the course of the next one to three years.
- [48:16]
Ari Paparo
Prime time for consultancies. All right, let's call it there. This was an amazing conversation. Olivia, Corey, thank you so much for being here and sharing all of your research and other insights.
- [48:26]
Olivia Corey
Thank you, guys. Big fans, so this is a lot of fun.
- [48:29]
Ari Paparo
All right, Eric, always a pleasure.
- [48:31]
Eric Franchi
Thanks. Thank you, Olivia. See you later, everybody. Thank you for subscribing to marketecture.
- [48:36]
Ari Paparo
New interviews are added every week at marketecture TV and your favorite podcast app. Thank you for listening to the marketecture podcast. New episodes come out every Friday and an insightful vendor interview is published each Monday. You can subscribe to our library of hundreds of executive interviews at marketecture tv. You can also sign up for free for our weekly newsletter with my original strategic insights on the week's news at News Market tv. And if you're feeling social, we operate a vibrant Slack community that you can apply to join@adtechgod.com.