Latest / Signal & Noise: Bringing Clarity to Media, Data, Tech and AI / POSSIBLE 2026 Day 2: Where AI Meets the Real Operating System of Marketing
Transcript
- 0:03This is signal and noise, bringing clarity to data, tech
- 0:07and AI. Possible 2026 Miami Day 2 with
- 0:17the one and only often spoken about David Nierenberg.
- 0:20How are you man? I love to say your name because
- 0:24I think it's frequently fucked up People best.
- 0:27You might love it, but I didn't. I don't love what people say.
- 0:30You got a multi syllabic name? My name is 2 syllables Brett
- 0:34House. And it gets messed up all the
- 0:36time. It doesn't travel.
- 0:37People call me Bread. Bread.
- 0:39Bread. Bread everywhere in the world.
- 0:41They fuck up the house. The house generally, not
- 0:44generally the house. Is all right.
- 0:46So we're looking. We're looking.
- 0:47House but but hey, great you're you're the SVP of digital at
- 0:51Intermedia advertising, right you were on the pod before with
- 0:54Sarah Caputo right there last night and you know you've
- 0:57written a lot and talked a lot in the in the industry about not
- 1:00only sort of not ad fraud, but sort of media efficiency and and
- 1:05and where where you know advertisers can be more
- 1:07effective in leveraging their and.
- 1:09Then within the efficiency, you know, you gotta have fraud.
- 1:13You got, you got different types of waste that go beyond just ad
- 1:17fraud. Yeah, you've got arbitrage.
- 1:19You know, you've got bad revenue models, you've got vendors
- 1:23trying to pull one over on their clients.
- 1:26I mean, it's just, you know, it's a fun.
- 1:28Yeah, yeah. And you've, and you've, you
- 1:30know, you've been pretty outspoken about, about what's
- 1:32wrong and what needs to be fixed.
- 1:33And and you know, it's, it's you've, you've exited
- 1:37organizations because of that outspokenness that's.
- 1:39Correct. And not then there's somebody
- 1:42has to be right because because it wants to work in an industry
- 1:45where all the great efforts that we put in ultimately don't
- 1:49matter because a few entities are corrupting the ecosystem and
- 1:55getting away with bad practices and diluting the efficacy of
- 1:58advertising, which is ultimately bad.
- 2:01And it, and it's all like a short termism, right?
- 2:03It's all about kind of a profit drive to say we need to spend
- 2:06these dollars that were allocated to me if whether I'm
- 2:09an agency or otherwise buy the brand, regardless of of the sort
- 2:13of the media buy output, right? Like if you're buying fraudulent
- 2:17inventory or just crap inventory that you know is going to
- 2:19underperform or not meet the needs of the client for that
- 2:21particular campaign, they're going to do it because they've
- 2:23got to spend the money. Is that is that?
- 2:25The well, you know, I think. Oversimplification.
- 2:28I think it's been oversimplified.
- 2:29I think maybe if that was more relevant several years ago than
- 2:33it is now to a degree, because I feel like we as an industry now
- 2:39care a bit more about media quality and media buyers are
- 2:43getting more savvy and technical and really putting in the work,
- 2:47more of the work to validate the inventory that they're buying.
- 2:50Not not everybody, of course, but there's more and more and
- 2:55there's more and more people and companies that you could put in
- 2:57that sophisticated bucket, right.
- 3:01So I think by and large, many agencies, especially independent
- 3:06agencies want to do a good job. They want their clients, They
- 3:09operate on relatively transparent media models.
- 3:13So they're the incentive. A lot the incentives aligned
- 3:15with the client incentives. Naturally, on the holding
- 3:18company side of things, things are a bit more squirrely and I'm
- 3:21always amazed at how holding companies are good at innovating
- 3:25new ways to fuck their clients. Yeah, that that's that's the
- 3:28let's dive into that. So I mean, do you think that
- 3:31that one of the causes of this is principal based buying?
- 3:34I mean, do you think that that drives or do you think principal
- 3:37based buying is a good economic decision to say we've got buying
- 3:40power? Is it, is it let's say an agency
- 3:41hold Co, we can buy at volume and then in a sense arbitrage
- 3:45off and in, in, up and sell this back to our clients.
- 3:48But when you're selling premium inventory for high CPM rates to
- 3:52your clients that cost you $0.50 CPM, that to me seems the part
- 3:57that that gets a little disingenuous, right, Because
- 4:00you're selling cheap inventory and potentially misrepresenting
- 4:04what it really is when in fact it's just sort of raw reach.
- 4:08But it may not be brand safe, it may not be may not really drive
- 4:11any outcomes because you may not be actually hitting anybody
- 4:14that's. Well, you're going to be driving
- 4:15outcomes. It's just like, how are you
- 4:16measuring those outcomes? What are those outcomes?
- 4:19Are they, are you properly assessing via an independent
- 4:23measurement system those outcomes or are you relying on
- 4:26your agency or brand or as that partner, you have their own home
- 4:30built measurement system outcomes, right?
- 4:32Grade their own homework? No, that's, that's really what
- 4:34all goes into effect. But yeah, I'll tell you this
- 4:37anecdote I heard from a friend at a holding company.
- 4:40So they had a big brand, right? And the big brand wanted to
- 4:45decrease their budget significantly year over year,
- 4:48but maintain the level of work required to manage those
- 4:52budgets, right? Yeah.
- 4:54So the amount of people that were required to manage the
- 4:58smaller budget versus the big budget was still.
- 5:00The same, yeah, but that. Has to be a smaller budget
- 5:02you're not going to be able to support.
- 5:04The same number of FT. ES yeah, so the thus layoffs,
- 5:09but the agency brought a principal based buying model to
- 5:12the brand and the brand signed off on it and they were able to
- 5:16retain all the employees. Yeah, yeah.
- 5:19And and that, you know, clearly that's that's changing with with
- 5:22the gentic with with sort of. The well, this is recently.
- 5:24Happened, yeah, this recently happened, yeah.
- 5:26And, and but you're also seeing in on the agency side, I mean
- 5:29they're, they're FT model, they're sort of that that
- 5:32revenue structure, right, supporting these accounts with
- 5:35large basically managed services that's under threat.
- 5:38It's under threat. It's not sustainable.
- 5:41In this case, you know, the brand was happy to get it, was
- 5:45happy to sign up on the principal media deal because it
- 5:47meant that there would have to be no layoffs and the level of
- 5:50quality of work would mean it would remain.
- 5:53But, you know, it was a sacrifice.
- 5:56Yeah. Like they gave him transparency
- 5:58in return for for quote UN quote efficiency.
- 6:01Yeah, Yeah. And I mean, do you think that
- 6:02that's just kind of a symptom of agencies sort of failing to
- 6:07adapt to what's happening around there And they're sort of it's
- 6:09almost like a gap filling I think size, it's like a maintain
- 6:12our, our margins, maintain our, our FDR revenue model as long as
- 6:18we can because we're not ready to make that full change.
- 6:20Right, right now, it's kind of sad.
- 6:22You get your quarterly, quarterly results that you've
- 6:25got a report and if you start to see things dipping, it's going
- 6:28to have an impact on your on shareholder value and it's going
- 6:30to have a pressure on the business.
- 6:31Right And you know, the winds of change have been fluttering for
- 6:3710 years now. You can see that anyone who
- 6:41cared enough to look and do though their homework could see
- 6:44that the legacy agency model was not sustainable long term,
- 6:48especially with brands just knocking commissions down more
- 6:51and more and more. Something had to change.
- 6:53Yet we still that the change has been a move towards the
- 6:56principle of media buying structure.
- 6:57With credit to the industry, we always find a way to move
- 7:01backwards rather than fall. Yeah, yeah.
- 7:03And it's and it adds opacity like it, it makes the whole
- 7:05process a bit more opaque because you don't really know
- 7:08the true value. And Rio and I and, and you have,
- 7:11we've all talked about this. It's like, you know, you'll
- 7:14you'll realize that the markups are are are at a rate that just
- 7:17is unacceptable, right? Right.
- 7:20So like in the case of the anecdote I just told you, you
- 7:23know, in this case, the brand got full transparency into what
- 7:27they were buying. Yeah, but they didn't get
- 7:30transparency into how much the agency paid upfront for that.
- 7:35Yeah, they saw the final price. They they got the publisher
- 7:38level transparency, but they didn't know what ultimately the
- 7:42true. CPM of that media one.
- 7:44Yeah, yeah. That was a sacrifice.
- 7:46They were willing. Yeah.
- 7:47And we've talked, we've talked about that case study where you,
- 7:48you do an audit, you realize they were charging 45X what they
- 7:52were actually paying for the media, right?
- 7:54Which, which I don't know how they were representing the in
- 7:57this particular case study, how they were representing the media
- 7:59client right now as other than just a reach play, right?
- 8:03Because was it being represented as premium audiences, premium
- 8:05inventory, premium publishers? Because in that case, you would,
- 8:08your argument would fall apart because, you know, or you, or
- 8:11you could just mask it by saying, Hey, these are, you
- 8:13know, exponentially higher CPMS than they would, than they what
- 8:16they actually paid for. I think that was the problem,
- 8:18right? Is you're buying cheap inventory
- 8:20and charging expensive, more premium premium rates when in
- 8:23fact it wasn't premium? Well, we don't.
- 8:27That's that's that's that's the question.
- 8:30Well, yeah. Well, in this case, this was a
- 8:32case study that that from Maria that that he ran and they and
- 8:35they saw it. They saw it on in the
- 8:37spreadsheet. It was a spreadsheet that was
- 8:38excellent sent to him by a client and they saw the market
- 8:42rates and they and they saw how it was represented.
- 8:45And they're like, this is complete bullshit.
- 8:46They called out the agency for let's pivot to from that to, to
- 8:49CTV because that's an area that you've got a ton of interest in
- 8:52and involvement in the CTV ecosystem from above the
- 8:55measurement perspective, from a fragmentation perspective and
- 8:58even a quality of inventory perspective.
- 9:00There's, there's challenges there.
- 9:01Tell me what you think are some of the biggest challenges right
- 9:04now with CTV as a it's clearly a growing space.
- 9:07There's a larger growing video media space.
- 9:10Like what do you think are the biggest challenges?
- 9:12With this challenge I was playing with CTV right now is
- 9:16the lack of appreciation, understanding and patience that
- 9:22brands have for it as a channel. Yeah.
- 9:25They like the performance CTV narrative has convinced many of
- 9:29these brands, especially the S and BS and the DTCS and
- 9:32something smaller ones. Yeah, they don't have massive
- 9:34budgets to spend. The CTV can perform like a
- 9:37Google or Yeah, right. With a Google or Metal, you
- 9:40expect results almost this. Yeah, yeah, yeah, exactly.
- 9:42And you see all the numbers right there and they look good.
- 9:45CTV, you need one. You can't just run a small
- 9:49button, right? You're not going to see great
- 9:51results or learn much with a small button.
- 9:53Yeah, you need to spend and you need time in market and you need
- 9:57to gather learnings and then you need to optimize.
- 10:00And unfortunately, a lot of these brands don't understand
- 10:04that they'll do a small CTV test, not see what they wanted
- 10:08to see, and then they'll turn off and be like, all right, I'm
- 10:10just going, I got to see TV is not for me.
- 10:14Yeah. And and you know, but it, but is
- 10:15that just does that misapplied expectations as to what type of
- 10:20results, what type of KPICTV is meant to deliver?
- 10:23Because if you think of traditional CTV technically
- 10:26should be served on the television glass, right, It
- 10:28isn't always that way, right, right.
- 10:29You've got a lot of cheap CP and CTV inventory that sort of masks
- 10:33the CTV that's being served on other devices.
- 10:35Battery now you're watching Google on your iPads now or your
- 10:38phone yeah, but there's not a ton of that type of inventory.
- 10:41Yeah, yeah, usually like I'll usually just go for CTV only,
- 10:45but like the old device, but I usually like it's 80% CTV, 20%
- 10:49mix of mobile, desktop, tablet. Yeah, if it's below 70% CTV,
- 10:54then something's usually. Off, yeah, something's usually
- 10:56off and you're getting kind of fake, fake out.
- 10:58Of the industry. Yeah, Yeah.
- 11:00No. So, so back to that point about
- 11:02expectations around CTV, because if I think of CTV as sort of
- 11:05sound, sight and motion, generally big screen immersive
- 11:09environments, it reminds if it's linear TV, right?
- 11:12What is linear TV for brand building, brand awareness, the
- 11:16Halo effect around brands, right?
- 11:18And then you have deeper funnel, mid funnel tactics that are more
- 11:22direct response, right? It's like you see a Toyota ad
- 11:25and then you're at the dealership level, right?
- 11:27And suddenly you're talking, you're seeing a dealership
- 11:29online, they're potentially retargeting you.
- 11:32That's a more did down funnel, like coming to the dealership,
- 11:34take a test drive. It's a different it's a
- 11:36different KPI. So is that that that's my
- 11:38question about miss expectations.
- 11:39Why would you be expecting CTV to perform like Google which is
- 11:43more of a last click attribution type of model?
- 11:46Right. Well, to me, CTV is a full
- 11:50funnel, right? Good advertising, especially
- 11:53advertising that tells a story like video on TV does.
- 11:57Yeah. Should drive awareness and
- 11:59knowledge of the brand, which long term should convert into
- 12:02outcomes intensive purchases. Yeah, the leads correct.
- 12:05IA few years ago I built an algorithm because this whole
- 12:10idea of it's either brand or it's performance, it's awareness
- 12:14or start response. It can't be both it's one or the
- 12:17other. It drove me nuts because it
- 12:18just. Doesn't make sense.
- 12:19Yeah. It's Isaac.
- 12:21Should do both. Yeah, no, you're right.
- 12:23Who just wants to run ads just to make just for quote of quote
- 12:27awareness and not drive sales? They want sales, they want
- 12:30money. Yeah.
- 12:32So like I built an algorithm that ingested a combo of Lucid
- 12:35brand lift data and conversion data from the TSP.
- 12:39Both of these, both of these data sets went into the
- 12:42algorithm and then the algorithm would identify the pockets of
- 12:45inventory that drove both high brandlet and direct response
- 12:49performance and start optimizing towards that.
- 12:52And then what we saw was within the first 3-4 weeks, you had a
- 12:57good strong lift in brand lift and recall.
- 13:02And then within week 5 and 6, that's when the direct response
- 13:06performance really picked up. You added a bit, the star was
- 13:10ramping up a bit. But then we 5:00 and 6:00 and
- 13:117:00. That's when I really kicked into
- 13:13gear. Yeah.
- 13:14And the and then the brand lift also stabilized, but kind of
- 13:18very high level. Yeah, yeah.
- 13:19Right. So the brand lift was having a
- 13:21Halo effect on the direct response.
- 13:22Response. Response rate.
- 13:24So those direct response actions might have happened in a
- 13:28different channel, not necessarily within the CTV
- 13:31channel, right? Which happened.
- 13:33You might do a search, you might do some other.
- 13:35Activity that right? Gets you to some level of
- 13:37conversion of interest, right I. Generally think that CTV is very
- 13:41much under attributed by our legacy measurement systems.
- 13:44But in this case, the my hypothesis, yeah, advertising
- 13:49drives both awareness and direct response.
- 13:51You just got to measure it correctly.
- 13:53No, that's true. Yeah.
- 13:54And then I think we, you know, we saw that kind of new start
- 13:56when I was at new start and I just interviewed Mike Finnerty
- 13:59who ran our MMM and MTA practice.
- 14:01And one thing that we we had Bank of America, JP Morgan, a
- 14:04lot of the big banks, right. And one thing we saw within
- 14:07those organizations is that 7080% of their budget was spent
- 14:10on direct response advertising. So it was the direct mailers,
- 14:13right, expensive on a per mail piece, but worth it because
- 14:17you're because of the lifetime value of a customer that you
- 14:19converted to like a credit card, for example, right.
- 14:22And so because it was so measurable, household targeted,
- 14:25they fell in love with it And they would, they would put 8070
- 14:28to 80% of their budgets into just that down funnel channel,
- 14:31right, direct mail. And they can continue, they can
- 14:33support their argument to to upper management that this was
- 14:36working and they would they would underestimate the value of
- 14:40upper upper funnel brand market. It could be CTV, could be linear
- 14:43TV. And so we ran these sort of 10
- 14:46year models, right, because it was 10 years with the data,
- 14:48because we had a long term relationship with with both
- 14:50clients and found that that the upper funnel when when you had
- 14:55more of a 5050 split in your media spin between upper funnel
- 14:58and lower funnel, you actually saw a significant increase in
- 15:01lower funnel activity. So it was the like where's if
- 15:05you kept at the 8020, meaning 20 in brand and 80 in direct
- 15:08response, right. Yeah.
- 15:11And so the Halo effect, which to your point is it's we're
- 15:14consumers, we're exposed to thing.
- 15:16We recall things we might think like, hey, I'm looking for fill
- 15:20in the blank and and you're going to think about the things
- 15:23that you know, either that or the people have recommended them
- 15:25to you and then you go and you come in an action.
- 15:27But that's brand building, that's brand awareness.
- 15:29And I think your point here is that it's critical and that CTV
- 15:33has got to be seen as a much. I think it's got to be seen a
- 15:36little bit more like linear television, right?
- 15:38I mean, do you think that I. Mean that's what I say CTV is
- 15:45CTV, the user experience. If I'm watching an ad on the TV,
- 15:50it doesn't matter to me if it's linear or CTV.
- 15:52My brain can't tell it's an ad on the on a big screen.
- 15:57And in like, what do you think about, because one of the things
- 15:59that I've seen and I was hoping that CTV would not follow the
- 16:03same path where I thought the ad load challenge with with Linux
- 16:08television where you know, when they finally implemented this
- 16:11ads across all stations. So when you wanted to flip like
- 16:14you're watching a game and you want to go to another station
- 16:17and you're hoping you can see another game, let's say it's
- 16:20like NBA and you're going to MLB.
- 16:22And when you're getting commercials all aligned.
- 16:23So I remember when that happened and I'm like, well, now I am
- 16:26forced to watch commercials, right, without subscribing to
- 16:29kind of a streaming provider, right?
- 16:30So like I, I always felt that the ad loads were too high for
- 16:34my sort of my linear television consumption.
- 16:37And now I'm feeling that CTV is just imitated, if not even
- 16:40compounded that problem, right? And and both from a frequency
- 16:44management perspective. I'm getting the same types of
- 16:46ads over and over and over again, right?
- 16:48Right. And from a fact that I'm just
- 16:50being, I think I'm being exposed to more ads on CTV than I am
- 16:53like per hour of watching I have.
- 16:56Some theories on that well one, yeah, we all have talked about
- 16:59how many people are watching streaming TV days growth
- 17:02numbers. Rarely do you see a breakdown
- 17:05between ads to your subscribers and non ads here.
- 17:08Yeah, yeah, right. So like when you see those those
- 17:12streaming numbers, you have to ask yourself how many of those
- 17:14actually have ads. Yeah.
- 17:16And what I believe is not as many as we would like to think
- 17:19they are. Yeah, yeah.
- 17:20So ultimately. The fast channels are not.
- 17:23You're saying they're not as quite as well penetrated as the
- 17:26more authenticated subscriber based ads experiences?
- 17:29Well, I, I think, I don't think the fast channels have much
- 17:32viewership as they like to claim.
- 17:34Yeah, that's one thing, right. And then two, naturally with the
- 17:38Hulu's and the Netflix's of the world, right, people are like
- 17:40there's going to be a split between people to go on ads here
- 17:42and then not. Yeah.
- 17:43So ultimately you combine the fact that I don't think there's
- 17:46a there's as many people watching these FAST channels as
- 17:48the FAST channels claim there are.
- 17:50And then you have a bite for Kated paid versus ad tier with
- 17:54the premium CTV publishers. You're going to.
- 17:57You're going to need to. Over deliver on frequency and
- 18:00get more ads in front of people than they might like in order
- 18:02to, you know, make some money. Yeah.
- 18:05Well, and do you think that's a good user experience?
- 18:07I mean, like, did CTV lived up to its sort of its hype, its
- 18:11promise or do you think we've just fallen into the same over
- 18:14advertised loop? You know, and I know we're in
- 18:16the advertising and tech industry, but you know, there,
- 18:18there's a balancing act here, right?
- 18:20The promise of CTV ultimately, I think was you could, you could
- 18:23select what you want to watch whenever you want to watch.
- 18:25You don't have to abide by its schedule.
- 18:27Yeah, it's all on demand. So I mean that that really was
- 18:31the convenience. Right.
- 18:32Yeah, there's advantages to that.
- 18:34That was the promise, yeah. What was the ad for?
- 18:37Like, what was the ad promise for CTV?
- 18:40I mean, there was always going to go in this direction.
- 18:42We were always going to fuck it out.
- 18:45Yeah. So question.
- 18:46Yeah, well, we're going to do right by it.
- 18:48Come on. Yeah.
- 18:49So you're saying that it is not, they're not delivering a better
- 18:52experience from an ad delivery perspective because you're
- 18:55seeing, you're seeing more repetition of ads, right, right.
- 18:58You're not seeing frequency capping and enforced in a way
- 19:01that you see in linear television.
- 19:03So it's a, it's a, it's a, it's a shittier ad experience.
- 19:05I mean, there's less transparency, so you can't
- 19:07really optimize. That's correct.
- 19:08So like so like, yeah, let's pivot there.
- 19:10So less transparency and and measuring outcomes for for CTV.
- 19:14Where is that breaking down for advertisers?
- 19:16Well. The way the way we like to
- 19:19measure CTV right now and made for many people is still in the
- 19:22legacy way we measure display online video via web pixel
- 19:25placed on the website. Yeah.
- 19:27And what I like to say is CTV is a much different dynamic than
- 19:31online video and display. Yes, it's digital, but I think
- 19:34because we just put the digital name on it, we then apply our
- 19:38legacy thinking and strategies that were applicable to do
- 19:41online video and display to CTV. But it's a, it's a completely
- 19:44different piece. So for instance, with the web,
- 19:47with the web, with the web pixel, with online video and
- 19:50display, right, The impression usually was served on the same
- 19:54device where the conversion occurred.
- 19:56Yeah. Right, some nuance, you know,
- 19:59some when the year of mobile kicked in, right, People
- 20:03started. Converting on mobile, right,
- 20:05using their mobile phones more and the device graphs started
- 20:09coming into play and getting more adoption.
- 20:11Yeah, but the device graph was an addition to the measurement,
- 20:15not the primary piece of primary foundation of measurement.
- 20:20With CTV where the impression happens on one device and the
- 20:22conversion will happen on a different device.
- 20:24The device graph is the focal point, is the foundation, is the
- 20:28most important piece to measure CTV correct?
- 20:31Because you can't measure a conversion happening on computer
- 20:33when the impression was served on the TV without a device.
- 20:36Yeah, Yeah, yeah. Right.
- 20:37And it's and it's linkaging, linking those devices not only
- 20:39to the household, but potentially for the individual,
- 20:41if you can do that with any level of accuracy, right.
- 20:44Generally it's at the household. Yeah, yeah.
- 20:48So you know, naturally device graphs and there's there's been
- 20:54recent research from truth sets come out that's shown how
- 20:57inaccurate the. Scott Mccarry on a little bit
- 21:00later. Oh good.
- 21:01Yeah, but I really enjoy. He was an Olympic cyclist.
- 21:06Yeah, he was. He was the he was the lead of
- 21:09the Olympic cycling team in like, it was like 1986 or 1992.
- 21:14Yeah. So he is an Uber athlete, just
- 21:17if you didn't know. So the next time you talked, I'd
- 21:18be like, I heard you were a cyclist.
- 21:20He's like, I was an Olympian. It's pretty impressive.
- 21:23Relatively unaccompanied. So yeah.
- 21:25So let me put out research recently that shows just how
- 21:28inaccurate the signals that make the linkages between different
- 21:32identifiers within a device graph are.
- 21:34Yep. So device graphs are a lot less
- 21:36accurate than we like to think they are.
- 21:38Yep. So, OK, so now you take that
- 21:39into consideration. And now again, we go back to the
- 21:42pixel pixels placed on the web page for the A conversion to be
- 21:44counted, the pixel has to fire. Yeah.
- 21:47Now, if you're running an ad blocker, which 30 to 40% in the
- 21:52US based on the research I've seen, does, yeah, pixel won't
- 21:55fire. Does it box all?
- 21:57Yeah, Pixel fires. You're running AVPN Pixel SO15?
- 22:02Yeah. Yeah.
- 22:04So that takes out your work population in a lot of cases
- 22:06percentage. If you're like Iraq, Iraq.
- 22:09If you're converting on Safari or an iOS phone, iOS strips out
- 22:18well. Safari strips out first party
- 22:21cookies after a 7 day window and then blocks third party cookies
- 22:26by default. Yeah.
- 22:27So highly likely your conversions if you're converting
- 22:29on Safari or a mobile or an iPhone will also not be counted.
- 22:33Yeah, by a pixel not to be captured.
- 22:36That's a shit ton of fucking people.
- 22:38Yeah. So you have all these different
- 22:40factors. And then again, if you, if
- 22:41you're on another browser like like Safari, like Brave, like
- 22:44Firefox, Yeah, they blocked. Firefox Pixels as well.
- 22:49We'll still use Firefox. I give that up years ago.
- 22:53So the research I saw was when 8% of the US uses Safari, 4% use
- 22:59Firefox. Yeah, 2% use break. 2% use break
- 23:02and the rest are crumb and the rest are yeah, yeah.
- 23:06Interesting. But the percent of the US has a
- 23:10iPhone, yeah. So that's all to say, you have
- 23:12all these factors convalescent to make measuring CTV out with
- 23:21the Pixel not capturing all converts.
- 23:23And we saw and we saw this, we saw this at Intermedia where we
- 23:28were running a campaign for a DTC client and we picked up
- 23:33let's just say 100 conversions for the pixel 100 sales not
- 23:37great at all. But then.
- 23:40Where were these sales happening?
- 23:42Was this an online e-commerce? Type of.
- 23:43Experience. Yeah, Yeah.
- 23:45So and then we decided to match up the IPS.
- 23:51From. Impression exposures to the IPS
- 23:55logged in their CRM because their CRM mattress the IP
- 23:59address of all converters Yeah, right.
- 24:01While people had to buy something you match those IP
- 24:04addresses Yeah, it's that it's that 1% accurate by any means
- 24:09type of manual attribute, type of attribution.
- 24:13But what we saw was that instead of 100 matched IPS slash pixel
- 24:20fire conversions, yeah, there were 1000. 10X different. 10X
- 24:25different and there's the difference between object
- 24:28failure of a campaign and a very successful.
- 24:31Yeah. So then, you know, I was like,
- 24:34wow, this is worse than I thought, right?
- 24:37And I dug in to see like try to understand how could it be this
- 24:41bad? I've heard, I've seen numbers,
- 24:44I've heard numbers that that about about what the discrepancy
- 24:47could be, but not this bad. Then when I looked into their
- 24:50data deeper, I saw that 60% of all their purchasers were coming
- 24:53from Icon, 60% of everyone buying their product and they're
- 24:57and they're moving millions a week in.
- 25:01Most cases, yeah, Yeah. So that applies to just any sort
- 25:05of contribution. It's not just CTV, right?
- 25:07It's any sort of digital interaction in which.
- 25:09Applies to a device graph heavy. Heavy device graph for live
- 25:13share. So if we're talking about also
- 25:16audio, place to audio. Yeah.
- 25:19So, so that let me just to finish that up the the solution
- 25:24automatically here and there's no perfect solution.
- 25:27We we rolled out a, we've rolled out a patchwork of solutions is
- 25:32now you're using use the browser limitations.
- 25:35You do that matching. If you can get the data from
- 25:38both sides, you can get the log files from your DSP and the and
- 25:42the IP addresses from the CRM. And then last but not least,
- 25:47running Internet. So like this client, this client
- 25:49was also running Prussian, which is AMM partner like a treble
- 25:53whale or house. Yep, and CTV crushed it in MMM
- 25:56and this is almost the near real time MMM like part of this thing
- 25:59that tipped me off was like we were looking great in the MMM
- 26:02great, but the pixels were barely capturing anything.
- 26:05So that's what the starting day. Yeah, I can't see more then I
- 26:08talked to you in a minute. the US president of Mute Nets, which
- 26:11is an Australian company that came here and and they they
- 26:13report back on because remember you're bringing in, it's
- 26:16anonymized data. Yep, generally, right.
- 26:19It covers a whole bunch of marketing exponents.
- 26:21So it's covering every, you know, seasonality factors,
- 26:24econometrics, any sort of economic impacts that could
- 26:27affect basically impact or, or outcomes for a particular media
- 26:32channel, right? And they, and they report their
- 26:34data out. It's all anonymized.
- 26:35They report their data out on a monthly basis because they're at
- 26:39the enterprise level. So they deal with enterprises
- 26:41and they basically claimed this are using this big strategic
- 26:44budget, budget allocation decisions.
- 26:45We're going to take this away from CTV and, and, and the, the,
- 26:49you know, you could, you could make that more frequent,
- 26:52frequent, but you're losing sort of fidelity and accuracy on
- 26:56really what's causing the impacts and where the shift
- 26:58should happen. So I mean that that'll be my MMM
- 27:01question back to you is real time MMM, I'm going to call you
- 27:03out. Is that real or you make it?
- 27:05Wow. There's a three day lag.
- 27:10Because I mean, there's, there's, there's test and learn
- 27:13experimentation, right? You know where you can test, you
- 27:16know, you can have a control group, you can have GL blocks
- 27:18right, where you have off and on and, and run small segments.
- 27:21But the question is, is that going to have the type of of
- 27:25insight to impact an entire media plan, right?
- 27:28Be able to go and say I need to shift $100 million from this
- 27:32channel to this channel, right? That's where I think MMM becomes
- 27:37much more of a predictive, you know, data heavy products,
- 27:42right? That's faster than it was.
- 27:43It used to take six months. Yeah, 12 months now.
- 27:46Now you're getting into these life cycles that are much
- 27:48quicker. It's it's largely intentic
- 27:50driven versus people driven. But but just just I thought I
- 27:53thought that was an interesting because you said real time MMA,
- 27:55but I. Thought DTC brands, They live
- 27:57and die by their MMS. Yeah, they live and they make all their
- 28:00decisions. Only based on what?
- 28:02They'll do with themselves and and what how do you what do you
- 28:04think the frequency of their media shifts are in terms of?
- 28:07Budget allocations, the patterns, but you got in there.
- 28:09Like a quarterly, monthly? No, no.
- 28:13Weekly, yeah. And this is digital channels
- 28:16obviously, because you can't make the changes as quickly, but
- 28:19they'll just make they'll make incremental changes just per
- 28:22channel seems to see how they get, you know, and these are
- 28:25direct to consumers, yes. So a lot of it's down funnel
- 28:27direct marketing, Yeah. And you're seeing a much shorter
- 28:31sales cycle. Sure, yeah.
- 28:31Exactly. That's why we're like a better
- 28:34word, these near real time MMMS, they work better for these DTCS
- 28:38because you know, I mean it's much more responsive.
- 28:42The sales are coming in much faster.
- 28:44The time to purchase is much faster often times.
- 28:48So, yeah, so what's your what's your thought?
- 28:50I'll ask you my final question because I was, I'm on the on the
- 28:52measurement kick and we don't cover it enough.
- 28:54And you and you're, you certainly love to get walkies.
- 28:56Is MTA dead? Do you think you think multi
- 28:59touch attribution has a future or do you think that do you
- 29:02think some of these MMM modern next Gen.
- 29:04MMM solutions, the rocker boxes, the Mutenex who we just
- 29:07interviewed and a few of these others are sort of replacing the
- 29:10need for sort of wow MTA as we as we once.
- 29:14Didn't consider this question is not very clean.
- 29:16But but it is. I think MTA is very much needed.
- 29:20I think that definitely has serves a purpose and has a
- 29:23place, but just the nature of the fragmentation, the industry,
- 29:28how one entity might be willing to share data necessary for an
- 29:32MTA while another will not, Yeah, makes it impossible at
- 29:35this point. Not impossible, not impossible,
- 29:38but hard and long term it'll only get harder as more and more
- 29:42entities become more and more privacy conscious and lock in
- 29:45their data. So Mtas as we know have a shelf
- 29:49life that's and will likely be phased out at some point just
- 29:53because they'll be. The nature of our walled garden
- 29:56universe, Right. Well, well, Hey, Thanks, David.
- 29:59This was awesome. Good dinner last night.
- 30:00Great having you. I hope you're enjoying Miami.
- 30:02Thanks for coming. I know.
- 30:05Yeah. We're following up possible with
- 30:07four days of Jazz Fest. So.
- 30:08Keep. Right after possible.
- 30:09Keeping the drinking to a minimum.
- 30:11Wise possible. You didn't tell me that's right
- 30:13after this. Yeah, yeah.
- 30:16You're not. You're not following the same
- 30:17same rules of engagement. My liver will not let me.
- 30:20Yeah, right. Well, thanks for joining us,
- 30:22man. It was a pleasure as always and
- 30:24thanks everybody. Illinois coming to you from
- 30:27possible 2026 in Miami on day 2. And I've got Adam Heimlich who's
- 30:31joining signals for the second time, CEO of Chalice AI.
- 30:35One thing that we talked about for those that haven't seen the
- 30:37podcast with Adam, with Rio and I, it's certainly worth
- 30:40watching. I think we would probably won
- 30:41over an hour, probably an hour and 10.
- 30:43It was a really good technical and interesting conversation.
- 30:46And I think one of the things that we started with which I
- 30:49think it would be good for those that haven't seen the podcast
- 30:50was your, your background. I mean, you came from an agency
- 30:54background. You were heavily involved in
- 30:55programmatic. You got sufficiently technical,
- 30:58you talk about how you got technical, you would learn how
- 31:00to build and, and make some of these products.
- 31:03And then you decided to go independent and form your own
- 31:06company. What was sort of that journey
- 31:07like? And in in what made you bring
- 31:10Chalice to life with your wife? Yeah, the what?
- 31:15Ally Manning is my wife and Co founder and I both realized by
- 31:19by 20/19 was that the machine learning predictive modeling
- 31:24that was offered by platforms as powerful as it was just wasn't a
- 31:28good fit for big brands. Like the things they're trying
- 31:30to do are not clicks and completes.
- 31:32They're trying to drive lifetime value and incremental
- 31:35conversions and they're trying to steal market share.
- 31:39And those signals are not what's being modeled to.
- 31:41It's not what's predicted. Yeah, it's a there's a short
- 31:43termism right in in sort of digital media.
- 31:46They're like a flattening. I think now that people are
- 31:48seeing a lot of AI outputs, right?
- 31:50The people call it slop. People are now realizing that
- 31:53average outputs, right things, things based on data pools have
- 31:57average output creative and a differentiated strategy.
- 32:00Your media buying shouldn't be this sort of flat output, like
- 32:04whoever clicks on it or whoever gives us the maturity.
- 32:07It's kind of a reach play. It's like the reach dilemma,
- 32:10like we want to reach as many people as possible.
- 32:12It seems infinite with, yeah, digital media, but I think what
- 32:16we're realizing is it may not be.
- 32:18It's much more finite from well. A conundrum is that reach media
- 32:21does better. Like YouTube does well because
- 32:23of its reach, and TV and billboards do well because of
- 32:26the reach, because in a broadcast medium at least, you
- 32:28reach everybody. Whereas if the modeling is
- 32:31wrong, it's actually an autopilot that's steering you
- 32:33away from reach and usually into your own customer base.
- 32:36So we see that. Yeah, yeah, yeah.
- 32:38Where you're targeting, you're not targeting sort of the
- 32:40movable middles. You're targeting right People
- 32:42that are already customers or in digital media.
- 32:45Often times your reach might be just slightly fabricated.
- 32:48That's right, right? Because like, no one knows,
- 32:50right? The global reach, anyway.
- 32:52Yeah, it's MFA's, it's, it's, it's inventory.
- 32:54It's IIVTS, it's inventory that's not really worth its
- 32:57grain and salt. And so you get this sort of
- 32:59infinite view. We talk to Doctor Fu all the
- 33:01time and you get this sort of infinite view of reach.
- 33:04And he's saying that there it's actually finite.
- 33:06And so one of the conclusions that came up with our own with
- 33:08our with our podcast was to your point about about like linear
- 33:11television, like some of this traditional linear reach media,
- 33:14it's actually proven to like the model seems pretty solid at.
- 33:18Least it's not an autopilot sending you away from where
- 33:21you're going. So that was the core problem.
- 33:23And you know, we're not, we're in year 7 of telling people like
- 33:26we could use your own data and model to exactly what you're
- 33:29trying to do and deploy that and programmatic and do things in
- 33:32well gardens and and. It's more it's about leveraging
- 33:35more of the the brand's data or are you actually combining that
- 33:38with the supply side as well or is it?
- 33:39Just purely the best. It's best data available.
- 33:42So sometimes brands have their own data and finance, they
- 33:44always have a lot of data, but in CPG they usually don't.
- 33:47So it's about going out to the panel marketplace.
- 33:50What I wanted to say is that it's only now 2026 and there's
- 33:53so much pressure to adopt AI and there's so much knowledge about,
- 33:58so much more knowledge about automation and AI slop that our
- 34:01message is really resonating. And we're saying this is how AI,
- 34:04its brands, it's like it's about using your own data to have an
- 34:08automated version of what you're actually trying to do.
- 34:11That's what Agent means. It means it's doing what?
- 34:14You want it to do and a lot of people have this experience of
- 34:17AI is something that you're opted into that's taking control
- 34:20away from you. And finally after, you know,
- 34:23here in our 7th year, we're saying you don't have to go that
- 34:25way. And only brands are saying, oh,
- 34:27like I now I see my choice. It's either get opted in or I,
- 34:30you know, dig into this AI thing and find out what my model
- 34:33looks. Like and do you think that model
- 34:35and we should talk about your containerized sort of model
- 34:37solution, which is relatively different than than most DSPS
- 34:40out there? And that's sort of the kind of
- 34:41the core competency, the value prop that that it's sort of a
- 34:44unique selling proposition for for Chalice.
- 34:47Tell us a little bit about that. But do you think that that data,
- 34:49when you talk about leveraging their data, does AI give brands
- 34:53and the sell side as well, publishers the ability to
- 34:55maximize a greater amount of data to be more effective in
- 34:59terms of the behavioral and intent and engagement signals
- 35:03that we're leveraging during the process?
- 35:05Absolutely. We, we talk to publishers every
- 35:07day as as much as I can. We talk to the sell side and we
- 35:10think that the the brand specific models that we're doing
- 35:14are are going to be good for publishers.
- 35:15Like we see that the platform way of doing things, right?
- 35:19Everything based on the ID, ignoring the page, ignoring the
- 35:22quality and content of that page has harmed publishers, right?
- 35:25The yeah and and it's. Publishers inability to pass all
- 35:30of their data, yeah, that they're sending in sort of a
- 35:32treasure trove of audience engagement back to the DSPS
- 35:34because it creates bandwidth issues and it cost a lot of
- 35:37money to manage that much. Basically a narrow pipe.
- 35:40The DSPS were built 10 years ago for the most part, and the pipe
- 35:43isn't broad enough to carry every URL and the quality.
- 35:47Yeah. So they miss signal from from
- 35:50the audiences that they're trying to.
- 35:51Try and the fiction is the signal isn't important, and
- 35:54maybe it's not if you don't care about quality or content, but
- 35:56we're. Still, just as consumers, we're
- 36:00still getting kind of shitty advertising right where you're
- 36:03like. How about just learning what
- 36:04your customers, even if you're if, even if you optimize to
- 36:06clicks like you want to know exactly where those clicks are
- 36:08coming from? Yeah.
- 36:09Yeah. What the pages have in common.
- 36:11Yeah. Yeah.
- 36:11And what and what kind of predictive steps led to that,
- 36:14that outcome so that you can find more people like that.
- 36:17It's not just your classic look alike modeling, but it's.
- 36:19Yeah. It's, it's audience clarity,
- 36:22right? And in clear definition in the
- 36:24ability to leverage the data to kind of do the promise that's
- 36:26been there for a long time, like the one to one in real time.
- 36:29And it just never seems to have worked, at least from a consumer
- 36:32perspective. And I think we've seen it in the
- 36:33industry. We've got to work in now.
- 36:35Yeah. So we have the case studies.
- 36:36The containerized deployment is a way to bring an entire model
- 36:41of the content and quality of the open web into decisioning
- 36:45that lives in an SSP server. And this exchange was the first
- 36:48to host it, but now it's a standard called the ARTF, the
- 36:51Agentic real time framework. It's been adopted by Open X
- 36:54still. Lives in the SSP set.
- 36:56Yes Sir, compute environment and the chalice model of the open
- 37:01web is living in containerized inside this environment
- 37:04responding in a millisecond. So it's it solves a whole bunch
- 37:07of technical problems, expense and right, that narrow pipe
- 37:11problem is completely solved. Yeah, yeah, access to the full
- 37:14bid stream. So that when you, yeah, you're
- 37:15actually taking full advantage of that.
- 37:18All the data that's available and upstream close to the
- 37:21publisher where there's 250 billion bid requests a day,
- 37:24right? So instead of it being narrowed,
- 37:26picking out of that whole stream by page with a lot of knowledge
- 37:30of what's on these pages and how well it scores for the brand and
- 37:34being able to share this information back, the results
- 37:36are incredible. That's what I'm hearing.
- 37:37Possible to say we're seeing savings up to 70% with results.
- 37:41Even better, we're getting a steady stream of learning on how
- 37:45to slice things like news to be on the news that's valuable to
- 37:47you. That being on the news, that's
- 37:49the. Critical.
- 37:50Panel. But Scala talks about that all
- 37:51the time, how news was so ignored because of just
- 37:54polarizing topics. And people are sort of it's
- 37:57chiseled because the audience is so valuable.
- 37:59Yep, we're taking a scalpel to news.
- 38:00Online video is another great place to take a scalpel.
- 38:03A scalpel? Like there's thousands and
- 38:05thousands of sites to show 2 videos at once.
- 38:08Horrible experience, but it passes the viewability and size
- 38:11thresholds of the platform. So being able to like filter out
- 38:16the real quality video, it turns out it doesn't even cost more.
- 38:19So that's why we see. And the real quality video is
- 38:22like on the device that you want it to be on.
- 38:23So if you're. Playing alone if.
- 38:25You're targeting CTV, you want it to be on the television
- 38:27glass. Or real content?
- 38:29That's the big one. Like is it before real content
- 38:31or is it just playing in a banner?
- 38:32Yeah, yeah, we could slice all that and find that the best
- 38:35video is not more expensive. So brands are getting price cuts
- 38:38as they get better results on their video campaign.
- 38:41Yeah, yeah. And we're excited.
- 38:42To but there's a trend where people say I'm going to buy a
- 38:45little, just use CTV as a video example where people, they race,
- 38:48there's kind of a race at the bottom.
- 38:50They're buying cheap inventory thinking at CTV, but it's
- 38:53actually just almost like fabricated impressions that are
- 38:55being shown across devices, domains that are being spoofed
- 39:00there. It's you know, and and so so
- 39:02that leads me to this question is, is that there is a lot of
- 39:04talk. We talked to primitive the other
- 39:06day, Joe Root, you probably know and and you're a great guy,
- 39:09super funny, super smart, lives in New York City, right, an
- 39:13Englishman and he was talking a lot about this sort of future of
- 39:16and I wanted to, you know, you guys as a DSP, kind of a a newly
- 39:19defined or differently defined DSP, I think.
- 39:21Oh. Yeah.
- 39:23Right, because you kind of live on that SSP side and you're,
- 39:26you're leveraging both sides of the data.
- 39:27Yeah. We encourage people to think
- 39:29about it as modularity. Chalice is a modular
- 39:32optimization tech. It could live in a DSP.
- 39:35It could live in an SSP. Over time, those lines get
- 39:39blurred. We're saying in the world that's
- 39:43coming, you're going to be able to take modular pieces and build
- 39:46the stack per campaign. It's almost like it's aligned
- 39:48with that composable. Trend.
- 39:49Yeah, exactly. That we see you're plugging in
- 39:52where, wherever it needs to be. So whatever the, the, the, the
- 39:54tsunami shifts in the industry that happen if you know.
- 39:57And so this leads to my question is that is that his, his
- 39:59argument is that, you know, it's probably in favor of the
- 40:01primitive business model. Is that, is that the direct,
- 40:04the, the closer you get to a direct relationship between
- 40:07publishers, the people that deliver the content, especially
- 40:10premium because they, they, they have the authenticated traffic
- 40:13and the brand. And the better you're going to
- 40:15be, right. And how that's becoming more of
- 40:17a reality. Yeah.
- 40:18I mean, that's in today's world. Where's are are all these middle
- 40:21men in the? Tax April 2026, that's where we
- 40:24are. You know, the buy side tech
- 40:26companies will respond like right now it looks like they're
- 40:30really far from the the data, the flow and they haven't built
- 40:34the environments that companies like Chalice need.
- 40:36But I'm sure they're going to come out with their next wave of
- 40:40innovation. I think part of the reason.
- 40:42You've seen that, you've seen that with like the Trade Desk
- 40:44and some open path and like they're starting to move more
- 40:46towards because they realize that's where.
- 40:48That's where the action is, yeah, we'll see more.
- 40:50But where we are right now and there's a little more action on
- 40:53the sell side. And I think it's because if you
- 40:55rewind 2 years ago, sell side ad tech was nowhere and there were
- 40:58there, yeah, they were in big trouble.
- 41:00So that, yeah. It's that.
- 41:01Pressure. Yeah, and that and the pressure
- 41:03on this on the on the sell side, right.
- 41:05Like because there's a lot of talk in the industry around and
- 41:07I've had, I won't name the the person with the company, but
- 41:10talk about, I think, I think we're, I was referencing the
- 41:12fact that I was talking to him and I referenced the fact that
- 41:14we talked and he's like, yeah, but but his chalice and other
- 41:18DSPS like fighting for kind of a shrinking pie of the open
- 41:21Internet, of the open web, right.
- 41:24I will tell you who it was. You probably could guess, but
- 41:25no. But like are they like is, is
- 41:27the entire future of of going to be either authenticated traffic
- 41:32on the premium publisher side or just behind walled gardens,
- 41:35which I guess authenticated. You know, I think there's a lot
- 41:38to like about walled gardens for the brands.
- 41:40It's very easy to use. They're they're trusted and they
- 41:42get a lot of money and they have great tech.
- 41:45But if to the extent that the open web can become easy to use
- 41:49and drive great results, we will compete.
- 41:51You know, I would say, you know, we're talking about Visa tech
- 41:53and when it's going to come back.
- 41:55Facebook now has a model called Andromeda that scans the
- 41:58creative any rate of you give them and it creates thousands of
- 42:00features about this creative Yeah and improves their modeling
- 42:03right. The more you know about what
- 42:05you're actually serving yeah, the better you can model right
- 42:08so where is the open web by side equivalent of that like I want
- 42:121000 features on creative for my data science team to be modeling
- 42:15off of Yeah, that's gonna come out right.
- 42:18We're going to compete with Facebook in the open web, but it
- 42:20takes a lot of companies, a lot of integration.
- 42:23Trying to create parity with the capabilities they're bringing to
- 42:27market. And it's like you have to
- 42:28capital society, you've got to compete with them at the level
- 42:31that they're competing and they're automating with this
- 42:32stuff the entire media buying process.
- 42:34And they had they had billions of dollars to invest.
- 42:36And the open web investment pipeline has been a little
- 42:39clogged in the in the era of of duopoly dominance.
- 42:43Yeah, yeah. Do you think it's hype?
- 42:44Do you think that the whole open web is dying?
- 42:46It's gonna, we're gonna live in a world, a walled garden world.
- 42:49Do you think that that's BS or do you think, do you think
- 42:52there's there's hope for, for an increasing and larger size?
- 42:55Let's say it could be premium, but also it could be long tail
- 42:57publishers. Yeah, I'm optimistic about
- 43:00competition. The issue is competition I
- 43:02think. The.
- 43:03Open The Open Web thrives largely on big advertisers and
- 43:07big agencies desiring an alternative to to Meta and
- 43:11YouTube, one that's data rich and and interoperable.
- 43:16I don't think that goes away even if it does shrink.
- 43:19And I think the story of shrinking is overblown.
- 43:23Like there's been a lot of issues in display and open web
- 43:26display and video that we're proving can be addressed and you
- 43:29can drive great results. And when people talk about
- 43:32people speak too broadly, like I think they should always split
- 43:35the enterprise and small advertiser market.
- 43:37Like small advertisers were never big open web display
- 43:40buyers, and the fact that they're going away and the
- 43:42publishers that supported them are going away is not
- 43:45necessarily an important. Trend and that's something that
- 43:47the Googles and the Facebooks in terms of of appealing to those
- 43:50that long tail yeah and basically giving them a simple
- 43:53easy way to do it themselves yeah they they won that a while
- 43:56back but that wasn't. Yeah, they've been winning.
- 43:58That they've been winning it for a long time and I don't, I think
- 44:01to your point, still trying to get big advertiser dollars.
- 44:04Big advertisers and big publishers like we don't need so
- 44:07many publishers like display, display publishing inventory has
- 44:10been infinite and the prices that goes down to $0.50 and like
- 44:15and people will miss that market who thrived in it, but it was
- 44:17never very important to big advertisers and premium
- 44:20publishers. And what is your perspective on
- 44:22Speaking of infinite cuz I think you could call that cuz I do
- 44:26think it's finite versus right in terms of the amount of
- 44:30inventory that's available in sort of long tail supply side of
- 44:34of the world. I mean, do you think, do you
- 44:36think that the open web still has kind of a polluted dirty
- 44:40inventory problem, right. This is kind of the doctor food.
- 44:43Let's say a pricing problem, right?
- 44:44There's a lot of $0.50 inventory that people, you know, people
- 44:48still buy for the the reach of to reach to IDs.
- 44:51And that $0.50, that price should be 0 or a penny, right?
- 44:55Like like those impressions are worthless.
- 44:56Yeah, like, yeah, it might be one, just one of the million,
- 45:00you know, recipe sites. It's AI generated right where
- 45:03it's not real human. That's all you.
- 45:04To see it looks like a crash. Oh my God, the market's
- 45:06crashing. All this $0.50 inventory is
- 45:08being reset to 0. But if you're a new site and
- 45:12suddenly there's a scalpel ability for a big B2B
- 45:15advertiser, tech advertiser to be in the news reaching
- 45:18engineers like they want to, then your CPM is going from
- 45:22$0.50 to $2.00 or $3 or $4.00. And that view doesn't get
- 45:26represented enough. Is like that.
- 45:27You know, proper pricing has, is a, is double edged, right?
- 45:30Some things will go to zero and some things are underpriced
- 45:33right now. I promise you we're staying
- 45:36massively underpriced quality inventory and.
- 45:39Yeah. And it's, it's a premium
- 45:40inventory, right, which, which is authenticated.
- 45:43You can if you're going direct to publisher, big publisher, you
- 45:45know that you're reaching the people that you're.
- 45:47Attending But it's not even all name brand publishers.
- 45:49There are good recipe sites that show clean video and have good
- 45:52layout and they've never been rewarded for their, you know,
- 45:55their commitment to having a well laid out page.
- 46:00It sounds like you're on a mission here.
- 46:02That's a good 'cause, right? Well, my mission is for
- 46:03advertisers to price things appropriately.
- 46:06Trying to tell publishers in this world, you, you can make
- 46:10more money, right? It doesn't have to be shrinking
- 46:12like the total inventory could shrink.
- 46:14That just means the money becomes more concentrated with
- 46:16fewer, better publishers. Yeah, yeah, yeah, That's
- 46:19awesome. So, So what, what do you see?
- 46:21What do you see for like the next couple of years, you know,
- 46:23at Childless AI, like you guys have you built the company, you
- 46:26ran a consulting gig for a little while.
- 46:28Like I'm sort of in that same kind of mode.
- 46:29So I have a ton of respect for that.
- 46:31You, you launch this organization with your wife,
- 46:33right? You're, you're starting to see
- 46:34some real traction in the ecosystem.
- 46:36You've got a pretty unique system that is modular like you
- 46:38talked about. What do you see in the next
- 46:40three to four years of Chalice AI?
- 46:43Well. Our technology is increasingly
- 46:46embedded in what we think will be the most important
- 46:48ecosystems, places like Index, Open X and Pigmatic, AWS, Live
- 46:53Ramp, a really important partner to us.
- 46:56And many companies that we're talking to here, we see as
- 46:58distribution points for this really flexible modular
- 47:02optimization tech, right? Paired with Live Ramp, then you
- 47:05have identity optimization conversion data, right?
- 47:09That's something very powerful. With Index, you have rich
- 47:12inventory data curation capabilities, right?
- 47:15A whole bunch of agency demand. And then our agency partners,
- 47:19right? Agencies are under tremendous
- 47:21pressure to adopt AI and show results.
- 47:23And we're showing that you don't necessarily have to fire your
- 47:27staff like another route would be driving much more value for
- 47:31your advertisers, right? If you drive enough value that
- 47:33the agency can get paid and the advertiser realizes gains.
- 47:37AI doesn't have to mean this massive cost cutting, right?
- 47:40You can get the value and the value gain that pays for the
- 47:44people. So we see that is really
- 47:45important trend and showing the service industry that this wave
- 47:49of tool adoption is a chance to legitimately earn right and on
- 47:54your expertise and save your jobs.
- 47:57Yeah. And it's, and it sounds like
- 47:58containerization is like in a sense you're building like a
- 48:00memory layer, like a memory system that's capturing more
- 48:03information, more data, more signal across the buy and the
- 48:06sell side. Yeah, they can be leveraged just
- 48:08for more and, and at the end of the day, this should drive
- 48:10better outcomes. Absolutely.
- 48:12Right. And measure even incrementality
- 48:13measurability. And as Hershey, yeah, we what I
- 48:16what I told my client Vinny at Hershey is.
- 48:19We don't have any. We have tomorrow, yeah.
- 48:21It's like we want to have like we know you make your money.
- 48:24It's, you know, it's for skinny pop, it's penetrating new
- 48:27households and for for Reese's, it's, it's really getting the
- 48:31two a month people to four a month.
- 48:33These are super hard things to model.
- 48:35Yeah. And we're gonna have models for
- 48:36you that can be dropped into your walled garden environments
- 48:39and your CTV environments. If you wanna work with live
- 48:41ramp, that'll be there. Yeah.
- 48:43And. That allow you that level of
- 48:44accuracy and that level of specificity for that that
- 48:46specific type of. Buyer and you'll be right and
- 48:49you're gonna send be sending instructions and you're gonna be
- 48:52able to compare. All right, I give my
- 48:53instructions to all these platforms.
- 48:56How did they do? What was the cost of value
- 48:57ratio? And we're saying that's that's
- 48:59the model that you're that we're going for in the future.
- 49:01Like you cannot adopt AI by going to a few different AI, you
- 49:07know feudal Lords and having them all make your.
- 49:09Your. Model.
- 49:10Your model has to sit on your side where your creative sits.
- 49:13This is your differentiation and it's it's per brand, it's per
- 49:17creative, it's per campaign. And you have to start thinking
- 49:20of. Artificial intelligence, I mean,
- 49:21this is where, yeah. What makes it AI is it's an
- 49:24extension of your brand strategy, right?
- 49:26This is automating what the brand is trying to do.
- 49:29And if someone says they have an agent and it's automating
- 49:31something you didn't choose, well, that's not your agent.
- 49:34Your agent is going to do exactly what skinny Pop needs to
- 49:37do, right? That's what we're trying to
- 49:39build for you. And you should tell the
- 49:40platforms like you could call it AI or call it agent, but I
- 49:44already have models that are designed to lift sales,
- 49:47penetrate households, improve lifetime value, optimize my
- 49:50entire chocolate portfolio. Like platforms aren't don't do
- 49:53that sort of work. Yeah, Well, this is great.
- 49:55You you heard it here first from from Adam Heimlich.
- 49:57Thank you for joining us. This is awesome.
- 49:59Thanks for having us. And but we'll have you on
- 50:01another lengthy episode, but this was great.
- 50:03Thanks everybody for joining us. This is possible event in Miami
- 50:08day #2 and I'm here with Anna from Black and Decker.
- 50:11Anna, Sir. Nice to thank you so much for
- 50:13having me. It's very exciting to be here.
- 50:15Probably possible last. Year wasn't.
- 50:17It yes, that's when we met in real life after having met at
- 50:21DNA committee, Martech committee.
- 50:24That there's no substitute for in real life I.
- 50:26Don't think? Yeah, completely agree.
- 50:28Yeah, and being in Miami is not too shabby either.
- 50:30So not at all. This view is amazing.
- 50:33It is. So I was excited to get you on
- 50:38the pod because you're so wealth of knowledge, like you've been
- 50:40working at Martech for a long time.
- 50:42You're super deep in data. I know you've been doing a lot
- 50:45of really interesting things with AI.
- 50:46So maybe we can start out by just tell the viewers a little
- 50:49bit about yourself and your background.
- 50:50I think that would be really cool.
- 50:51Sure. So first of all, thank you for
- 50:53having me. So I'm on a modal and I actually
- 50:57started my career as a digital marketing manager, actually
- 51:01evolved into analytics measurement.
- 51:04I always really liked that you could measure things in digital
- 51:07marketing. And then I ended up going to
- 51:10martech, right? And some years ago I was
- 51:13actually hired to implement a martech stack to collect and
- 51:17activate and user data. And that's how I got really deep
- 51:21into martech. And that's how I end up actually
- 51:24creating my own framework to help other marketers do the most
- 51:27out of martech and AI as well. Martech could be incredibly
- 51:32confusing, right? I believe all like I know that
- 51:34your follow Scott Brinker and you know, chief Martech and all
- 51:37the stuff they've been doing. It's gone from like 5010
- 51:40thousand to there's so many Martech things and then you
- 51:43know, we're this is an advertising event.
- 51:45I think when you fold in all it's gone on an ad tech.
- 51:49It's incredibly complex and bewildering.
- 51:51And I mean, I don't blame marketers.
- 51:53The fact that it's in fact they are confused, it's totally
- 51:56understandable. Absolutely, absolutely.
- 51:58It's very confusing. I think it's hard to know where
- 52:01to start, right? So I think a lot of marketers
- 52:04receive that directive of making the most out of technology,
- 52:07making the most out of AI, but without a lot of guidance.
- 52:11So a lot of marketers have to find their way around these
- 52:15technologies, around these platforms, both in martech and
- 52:18ad tech, to make sure that they're making the most out of
- 52:22those technologies on behalf of their companies and also on
- 52:25behalf of their own careers, right?
- 52:28Because you are exposed to new technology, you learn new things
- 52:31and you can evolve as well as a marketer, but it takes effort,
- 52:36right? It's not easy.
- 52:37And what I would recommend is that marketers really lean into
- 52:42the conversations inside their companies.
- 52:44Go talk to your IT, go understand how zero and first
- 52:48party data gets into your mark tech, where it goes in there,
- 52:52how it can be activated, right. I think having that foundational
- 52:56knowledge can really help when marketers are discussing with
- 53:00other teams, with vendors, helping select new platforms.
- 53:04If you have that view of how the data is and what you need to do
- 53:09with it and the challenges around collecting, unifying,
- 53:13activating, then you have a better idea of what is needed.
- 53:17That, at least, is my experience.
- 53:19She mentioned activation. I think this ties to an
- 53:22important point. I've always been a little
- 53:25puzzled by this. You have the world of martech,
- 53:27which is, you know, it's a big complex world.
- 53:28It's very well established. There's a lot of thought
- 53:30leadership. And then you've got the separate
- 53:32world of ad tech, right? I don't think there's too much
- 53:36cross pollination or even overlap, to be honest.
- 53:39Where I find a lot of people in AD tech, they almost scoff and
- 53:42say, oh, you know, we don't need to know martech, right?
- 53:44And people in martech are saying, well, ad tech that's
- 53:46just DSPS and not not realizing, OK, there's DSPS, ad exchanges,
- 53:50SSPS, there's clean rooms, which are kind of between these ad
- 53:55tech. There's different measurement
- 53:57devices. It's incredibly complex.
- 53:59There are thousands of ad tech tools.
- 54:00We have our own ad tech framework, that is, there's some
- 54:04overlap with Mark Tech, but not as much as you think.
- 54:07But I think those two worlds are starting to get closer together.
- 54:09So I guess my question I have is like, what do you see in terms
- 54:12of the two worlds? I don't know if they're
- 54:16converging, but at least becoming more linked.
- 54:18Is that something you're seeing? Oh, absolutely, yes, I think
- 54:21they're becoming more blank. There's more overlap, but in a
- 54:24very positive way, right, Because companies could and
- 54:28should leverage their own zero and first party data to make the
- 54:31most of our their media buy for instance, or understand the
- 54:35overlap of their own audience and the audience in other
- 54:38publications media platforms. So I think that the separation
- 54:43actually, it's no way it's official, right, Because that's
- 54:47always been overlapped. Advertising is part of
- 54:49marketing. So the separation was more
- 54:53towards, I think that the technology ended up silent a
- 54:56little more, but it shouldn't. And I think it's very positive
- 55:00to be in an event like this possible where you see both
- 55:03advertising and martech and very depth discussions of AI in both
- 55:08advertising and marketing general and where they overlap.
- 55:12So I think that's a very good first half.
- 55:14I think that's a good point. Like possible.
- 55:16Is it an event where you can see both worlds?
- 55:18And I think that there aren't too many of those.
- 55:20Can is I think the same, but this is probably the only other
- 55:23one where I've seen like all of the martech people I know we're
- 55:27all here and all the advertising and, you know, impossible even
- 55:30like there's more agency people here and more brands.
- 55:36I mean, I've seen a lot of brand marketers and brand, you know,
- 55:38marketing technologists. It's a cool evolution here.
- 55:41Oh, absolutely. And the event has been evolving,
- 55:43right? So this year you see some tracks
- 55:45and some stages that you didn't see last year.
- 55:48So I think the event is also being very smart in evolving
- 55:52with its audience. So I see that conversions
- 55:55growing into the future for this event and maybe others as well.
- 55:59I think another reason why martech and ad tech have kind of
- 56:01stayed different is I mean, for all intents and purposes, I
- 56:04mean, some some brands have in house even media buying, but
- 56:07that's pretty rare, right? I think especially for big
- 56:09global brands, you know, they're going to continue to work with
- 56:11agencies for a lot of reasons. But I think that that's part of
- 56:14the reason is that ad tech is executed.
- 56:16Many in many of the contracts even sit with agencies.
- 56:20So I think that's kind of led to this division that I mean, I
- 56:23agree with you. I'm starting to see, starting to
- 56:26see and maybe your first party data is that kind of bridge
- 56:28between the two I. Mean absolutely, yes, that's at
- 56:32least my experience is through the zero and first party data,
- 56:35right. So let's make the most out of
- 56:37our paid media budget by making sure that we start by telling
- 56:41the platform or the media company, OK, we want to find
- 56:44people like this, but not exactly these because they
- 56:47already are with us. So this helps us then go and
- 56:51expand that audience, but making sure that we also improve our
- 56:56cost per acquisition, for instance.
- 56:58So when we first started using our zero and first party data,
- 57:01that's the first benefit we saw. Targeting so you're able to like
- 57:04like better target people who you're not.
- 57:07You're saying you're so you suppressed your existing
- 57:09customer. Exactly.
- 57:10Exactly. OK, so these are like targeting
- 57:12people who look like your existing customers.
- 57:13I bet with all that incredible first party data you have, when
- 57:16you know lifetime value, you know your purchase history.
- 57:19So I bet you can really get granular and like hyper targeted
- 57:22in terms of building those audiences.
- 57:23Oh absolutely. And then you can find like the
- 57:25most engaged or the most valuable customers you have the
- 57:29ones that convert the most and say, OK, I want to start with
- 57:31those, right? So you don't have to give
- 57:33everybody together in order to do good segmentation.
- 57:36You can also segment your first party data 1st and then go after
- 57:40those specific audiences. And that's really powerful
- 57:44because it's really creates the foundation upon which we can
- 57:48then add more on top, right? So we start with the first part
- 57:52data and then you say, alright, and now I also want people who
- 57:55aren't interested in this and that, but there's an overlap
- 57:58with the first party data. So it's really helpful.
- 58:00And I think it's also very helpful for retail media as
- 58:03well. That you let you led me to my
- 58:06next question. So which like looking at
- 58:08activation like which channels like for this like using look
- 58:12alikes for suppression and using first party data for suppression
- 58:17and look alike say what channels for activation have you had the
- 58:20most success? And sounds like retail media is
- 58:22one of them. But are you doing any open web?
- 58:24Like if you collaborate on that, that would be amazing.
- 58:26So through my experience overall, I've seen more success
- 58:30with retail media and in social media platforms, right.
- 58:34But that could also be replicated in open web in other
- 58:38DSPS throughout the whole ecosystem I believe.
- 58:43And like, how does that work? I mean, do you, I'm assuming you
- 58:45work with an agency that does that, does the media bias or do
- 58:48you do that in house? For that, for some of those
- 58:53experiences, we did them the the buying house, right, but could
- 58:57also work with an agency because when we leverage the first party
- 59:01data the way we do it, we send it very much encrypted, right,
- 59:05very safely to the platform itself.
- 59:07And then the agency can see the counts but cannot see the PII,
- 59:12right. So it's super safe.
- 59:13Got it. So you're either uploading to
- 59:16some some secure method, whether it's a clean room or something
- 59:18else. Exactly.
- 59:19OK. And then is it coming straight
- 59:21from yours? You have AI, don't know if you
- 59:22have ACDP or CRM. How is that data typically?
- 59:25Where's the source data coming from typically?
- 59:28So I have experience with coming from the CDP and coming from the
- 59:32customer engagement platform. So whichever 1.
- 59:35Phrase or something like that. Exactly.
- 59:36So whichever one would work, right?
- 59:39So it's not that you have to have a CDP or have to have a
- 59:42data warehouse. No, you can start with what you
- 59:44have on your CRM, on your SAP and go from there.
- 59:47And you can start with a small proof of concept, see how it
- 59:51goes and then scale from there. That's also my approach.
- 59:55Good point about CDP is I mean I guess CDP I look at it more of
- 59:58like OK, I got almost a constellation of capabilities
- 1:00:02involving first party data segmentation, some of analytics
- 1:00:08activation that. It could be in a CDP or it could
- 1:00:11be in a cloud data warehouse plus something else.
- 1:00:13So I don't think there's, I think like the CDP world seems
- 1:00:16to be more of a consensus that there's not, well, it's not a
- 1:00:19one-size-fits-all model. Not at all, not at all.
- 1:00:21And I agree with your assessment.
- 1:00:23One of your articles where you said that we the CDP category is
- 1:00:27being deconstructed and it is. I agree with you.
- 1:00:30And I think there's a lot of overlap with other platforms as
- 1:00:33well. That's why I mentioned that is
- 1:00:35important for marketers and other stakeholders to understand
- 1:00:38what they need to do and then they go find the technology,
- 1:00:42right, Because it's not one-size-fits-all.
- 1:00:45It may be that you don't need this one technology.
- 1:00:47What you need actually it's a mix or you need a best of breed
- 1:00:51approach and you are going to supplement with something that
- 1:00:54you create inside the company, right.
- 1:00:56So it needs to be very well thought of so that you get the
- 1:01:00most out of it. How do you like when you're
- 1:01:04looking at different vendors, let's say you're looking at a
- 1:01:06category like like what are your, do you have any tips or
- 1:01:08best practices for other, let's say marketers who are not as
- 1:01:13deep as you and martech or or maybe coming from the ad tech
- 1:01:18side, like any tips or tricks like how you approach these
- 1:01:20categories? Like let's say you wanted to go
- 1:01:22get a solve first party data problems and you're maybe
- 1:01:25looking at a CDP. Any like best practices you can
- 1:01:27share? Sure.
- 1:01:29So. In our case, what happened was
- 1:01:31that we, we understood that we were collecting the data, but
- 1:01:36when we went back to our business stakeholders, to our
- 1:01:38senior leadership, they asked, OK, so how about the data over
- 1:01:42time? And we didn't have the CDP and
- 1:01:46they're like, OK, no, that over time we cannot do that.
- 1:01:49So we realized that actually it was not about having the data
- 1:01:52on, it was about collecting data over time so that we could have
- 1:01:55a clear picture over time, right?
- 1:01:59So that's what led us to say, OK, so we need a platform that
- 1:02:02gives that capability to us. And most of the questions we
- 1:02:07were getting were marketing related.
- 1:02:09So if we were not going to look for a database, that could not
- 1:02:14help marketers. Sure, right.
- 1:02:15So that's how we got to them. The CDP, that was years ago.
- 1:02:19But those are the questions that I think need to be discussed
- 1:02:22before you get to, oh, it's this acronym or that acronym.
- 1:02:27Focusing OK, what problems do we want to solve and then like what
- 1:02:30capabilities or use cases do we can we identify that that would
- 1:02:34potentially address those and kind of finding best fit right
- 1:02:36based on that and other factors too that's.
- 1:02:39Exactly it. OK.
- 1:02:41The role of procurement in like I think for marketing, like all
- 1:02:44of marketing, especially for advertising has been, you know,
- 1:02:47it comes up a lot these days. You know, like people complain
- 1:02:50about procurement all they're just evaluating based on it's
- 1:02:53everything's cost and it's in paid media, it's all CPM major
- 1:02:57pubs. I think in Martech too, a lot of
- 1:02:58it is looking myopically on OK, like on not understanding the
- 1:03:03total cost of ownership more just under OK, OK, but what is
- 1:03:06the per seat license or what's this or and I think missing the
- 1:03:09farce for the trees and making those decisions like any
- 1:03:11thoughts on that? Oh, absolutely.
- 1:03:13That happens a lot unfortunately.
- 1:03:15And I think one of the ways to counter this with procurement is
- 1:03:19to bring any impacts on the operations, marketing
- 1:03:22operations, right? Because if you may go to
- 1:03:25procurement and say, OK, so this one is cheaper in the cost per
- 1:03:29e-mail address that I have, let's say.
- 1:03:32But this one will increase our time to market because it does
- 1:03:36not solve this specific issue for us or it may increase our
- 1:03:40dependency on external vendors so that it should also be taken
- 1:03:44into consideration. So I think that's an
- 1:03:46opportunity, again for marketers to go and make sure that they
- 1:03:49explain the effects that those technologies also have have on
- 1:03:52their day-to-day and opportunities for efficiency and
- 1:03:56accuracy. Like that a lot turning to AI
- 1:03:59for a second like what are you seeing as like a marketing
- 1:04:02technologist? What are you saying?
- 1:04:03That's because there's a lot of just, and there's some gazillion
- 1:04:07start-ups, right? And they all have incredible
- 1:04:10pitches. Well, some better than others,
- 1:04:11let's say, but I just think it's, it's even more
- 1:04:14bewildering. But looking through all of this
- 1:04:16say what's what do you think is real and what's getting you
- 1:04:18excited that's actually happening there right now from a
- 1:04:21marketing technology perspective?
- 1:04:23Oh, wow, That's a great question.
- 1:04:25I think there is a lot of opportunities for marketing to
- 1:04:30improve in two ways, right? So 1 is on the operation.
- 1:04:33So how can we be more effective, more efficient, right?
- 1:04:37How can we with the same team produce more?
- 1:04:39Because there's a lot of demand for that nowadays.
- 1:04:42And the other side, of course, is how we communicate to the
- 1:04:45customer, to your end user, to your public, and you make the
- 1:04:49most out of it, increase engagement, increase conversion,
- 1:04:54right? So there are those two areas and
- 1:04:56there are many benefits on both. And I think the AI decisioning
- 1:05:01is a big one, right. So it's different from what we
- 1:05:04had in the past where we would say, OK, if there's then that
- 1:05:08now you can get give AI decisioning the objective and
- 1:05:13the AI itself, we'll figure out how to get there.
- 1:05:16Of course, you need to put guard rails in place and both without
- 1:05:19saying that you should have the date in place because otherwise
- 1:05:23the AI won't be able to really get to the objective if the data
- 1:05:27does not allow it to see how to get to that objective to begin
- 1:05:31with, right? So I see an opportunity here,
- 1:05:34first of all, for the teams that have already been working on
- 1:05:37organizing their customer data. So good for you.
- 1:05:41And 2nd on the ones that have not to really get ahead of it so
- 1:05:45that you can make the most out of AI.
- 1:05:47And then it can also be scalable because otherwise, yes, you can
- 1:05:51get gains, but then you cannot scale right in a big company
- 1:05:55that won't take you that far. So those are and avoid FOMO,
- 1:05:58right? There will be a lot of FOMO.
- 1:06:00So no FOMO, just go back to the basics, to everything we were
- 1:06:03just describing here. And then you go and check the
- 1:06:07the opportunity that AI can have for your objectives.
- 1:06:11Yes, those are real. That's really good advice.
- 1:06:13And then I guess looking like next couple days, I mean, I know
- 1:06:18we're halfway through, like what are there any sessions you're
- 1:06:21excited about? Are there any?
- 1:06:22Like, what are you really looking forward to in this
- 1:06:25event? What are you hoping to get out
- 1:06:26of? It so I'm really happy to be
- 1:06:29meeting all the brand side marketers.
- 1:06:30I like to meet my yes, and I'm trying to like meet them, talk
- 1:06:35to them because it's amazing to benchmark with my peers.
- 1:06:38That's awesome. And also a lot of discussions
- 1:06:41about AI in the future of marketing work, right.
- 1:06:44So those are some that I'm also looking forward to to understand
- 1:06:49how other companies are seeing that playing out for them and
- 1:06:53how they are working around it. I think it's it's actually on my
- 1:06:56mind as well and I think it's on a lot of marketers minds.
- 1:07:00Yeah, well, it's impressive. They've done a really good job
- 1:07:02in sending brands to come here, I've noticed.
- 1:07:04That yes, yes. And you know, this year, I mean,
- 1:07:07I'm, I'm with Omnicom and like I've noticed more big agency
- 1:07:10people. I mean, two years ago, there's
- 1:07:12very few. So it's been, it's been a nice
- 1:07:14evolution to see that. And the conference has gotten
- 1:07:17bigger. I mean, it's, I think it was
- 1:07:186000 last year. It's going to be, you know,
- 1:07:20certainly bigger than that this year.
- 1:07:21I mean, it was, it was a mob seat down in the, in the kind of
- 1:07:25reception area there by the by that bar.
- 1:07:27But I'm looking forward to the next couple days.
- 1:07:29I think it's going to be a lot of fun and I'm sure we'll run
- 1:07:32into each other. We usually do and I really
- 1:07:34appreciate you coming on. Glad to talk about this.
- 1:07:36And we definitely need to get you on for a longer like you
- 1:07:40know how we like those, you know, 69 minute discussion.
- 1:07:43We're really going to depth. I'd love to we'd love to get you
- 1:07:45back on for one of those in the near future.
- 1:07:46And of course, and thanks again for coming.
- 1:07:48And thank you for having me. From a nice beautiful overlook
- 1:07:54of Miami at possible 2026, Mike Finnerty, the newly anointed
- 1:07:59U.S. President of Mutenex, right,
- 1:08:02which is a kind of A next generation MMM or media mix
- 1:08:05modeling, marketing mix modeling, however you want to
- 1:08:08call IT company. Congratulations on the new job.
- 1:08:10And for those that don't know, Mike and I go way back.
- 1:08:13We were at New Star together for quite a few years.
- 1:08:16Dating ourselves. Yeah, we're dating ourselves
- 1:08:17before the TransUnion acquisition.
- 1:08:19So, so you know, are you enjoying you enjoying the show
- 1:08:22so far if possible? It's been great.
- 1:08:24I think Ki is a topic in everybody.
- 1:08:26'S talk track these. Days and we're taking a very
- 1:08:30aggressive approach of leveraging AI to essentially
- 1:08:33bring MMM into the current it was always supposed to bring to
- 1:08:39events. Yeah.
- 1:08:40And, and, and one thing that we that we saw and certainly and
- 1:08:43you ran, you know, a team of what 200 people, 200 consultants
- 1:08:46running the MMM practice at Newsroom marketing systems for a
- 1:08:48number of years. And that was post market share,
- 1:08:51right? Market share was purchased by
- 1:08:53new start by new start, right. And so you've seen a very kind
- 1:08:57of FTE driven model, consultant driven model.
- 1:09:00There's companies like analytic partners out there that also
- 1:09:03have you know the which it's very managed services driven.
- 1:09:07It sounds like what Mutenex is doing, which you know relatively
- 1:09:10new to the US market, you're kind of helping spearhead it is
- 1:09:13something different when it comes to how MMM analytics and
- 1:09:18strategy is actually delivered. Can you talk a little bit about
- 1:09:21that? Yeah.
- 1:09:22I would say that a few things unstructured and then we can
- 1:09:25dive into each of the pieces. I think we're seeing right now a
- 1:09:29focus on measurement and specifically like tying
- 1:09:32marketing investments to incremental growth for any
- 1:09:36marketer like. The business incrementality has
- 1:09:38been a topic. For a long time it's been a
- 1:09:39topic and it's elusive, right? It's very hard to do at scale
- 1:09:43and with the right speed to actually inform the decisions
- 1:09:46that it. Seems to make.
- 1:09:47Yeah. And that's the key point, the
- 1:09:48speed in terms of how do you actually make switch switches to
- 1:09:52your budget allocation, your media allocation, even your
- 1:09:55creative. And it's been a promise, but it
- 1:09:57just hasn't actually been delivered upon it.
- 1:09:59Isn't it's the I mean the MMM has long had the moniker of
- 1:10:04great but late yeah right. It is holistic.
- 1:10:06It's definitely a marketing mix meeting we're.
- 1:10:08Looking at yeah and econometrics, right, it's a lot
- 1:10:11of these these kind of exogenous external factors, right,
- 1:10:14anything impact the path to purchase or the consumer,
- 1:10:16consumer behavior. Anything.
- 1:10:18Drive up or down demand for a business.
- 1:10:20We want to be able to articulate somehow in the model.
- 1:10:22Yeah. It helps us better articulate
- 1:10:24what marketing versus pricing is doing and it helps us forecast
- 1:10:28out right, Cuz ultimately you want to use this to make
- 1:10:30decisions about next campaign, next quarter, next budget.
- 1:10:34And so you have to be able to tie it to predictions of what
- 1:10:37ultimately the top line of the business is going to look like.
- 1:10:39Yeah. And you have to be accurate,
- 1:10:41right? So you have to be able to
- 1:10:42articulate again what those what those trackers are.
- 1:10:44Yeah, and and you know, there's been a lot of kind of younger
- 1:10:47companies coming as a rocker box that was that was bought by DV,
- 1:10:51right, that have come out saying that they do.
- 1:10:53MMM. You've got the contraditional
- 1:10:54players like the IR eyes, the the new star TransUnion,
- 1:10:57although they're sort of seemingly exiting that business.
- 1:11:00You get the analytic partners. I mean, how do you see this
- 1:11:02landscape and where do you see do you see some of it being
- 1:11:05disrupted and being taken over by by by approaches that are
- 1:11:08more agent in AI driven? So I think the.
- 1:11:11There's generally 2 camps. Yeah, in in the MMM space.
- 1:11:14I think the the probably most well known are your
- 1:11:16consultancies. Yeah, right.
- 1:11:18And that is your analytic partners.
- 1:11:20Epis MMA Ecometrix Game Theory. Yeah, Game theory, yeah.
- 1:11:24And those, those are typically the players you'll see in the
- 1:11:26Forester wave of the Gartner. Yeah, it's a people set of
- 1:11:30Europe right there out of France.
- 1:11:32Yeah, yeah, European IT. That is what I've spent my
- 1:11:35career doing, right. It's a very people heavy
- 1:11:37delivery model, yeah. And so it's takes a lot of time.
- 1:11:41It's generally pretty expensive. Yeah.
- 1:11:43And it's slow, yeah. Right.
- 1:11:44But, and these are typically like 6, even 12 months is MMM
- 1:11:49has always been known for these sort of.
- 1:11:50Long, pretty long way. And then you use it for kind of
- 1:11:52the net usually annual planning cycles, right, versus a
- 1:11:56quarterly planning cycles? And that that here and lies the
- 1:11:58challenge, right? With the way that the ecosystem
- 1:12:01is moving, you need to get really granular because no one
- 1:12:04plans video anymore. You're talking down at
- 1:12:06publishers. You're talking about ad formats.
- 1:12:09Yeah, you're talking about audiences, right?
- 1:12:10So you have to get the analysis down to a level to actually
- 1:12:13inform decisioning. The media buying the media.
- 1:12:16Exactly. Yeah.
- 1:12:17So. To get granular.
- 1:12:19In we're talking about video can be fast, it could be CTV, it
- 1:12:24could be even social video. To like, are you running on
- 1:12:27Peacock? Where?
- 1:12:28Very specific publishers. So you have the needs to be
- 1:12:31granular enough to fit in and fuel decision making to the
- 1:12:33actual teams that plan and execute.
- 1:12:35Yeah. And it's got to be recent
- 1:12:36because performance, culture, buying behavior is incredibly
- 1:12:40fluid. Yeah, right.
- 1:12:41And so results from six months ago, six months ago or not all
- 1:12:44that action. Yeah.
- 1:12:46And so there's. It's kind of like your old batch
- 1:12:48data systems, right Where it's like, yeah, you get, you get
- 1:12:50batch data to train a model, but it's outdated the literally the
- 1:12:53net exported. Right.
- 1:12:54Right. And so it's not able to make
- 1:12:56decisions that are that are being impacted.
- 1:12:58It could be an economic shift, it could be a seasonality shift,
- 1:13:02it could be a consumer behavior trend or a competitive pressure.
- 1:13:05If you want to be proactive and take advantage of those trends,
- 1:13:08right, you use marketing as the growth driver of your business.
- 1:13:11Yeah, you can't be looking at six months, 6 month old or stale
- 1:13:14data, right? Yeah.
- 1:13:15So to bring it back, we have consultancy approach on one
- 1:13:18side. You also see like rocker boxes 1
- 1:13:20of what I would say a group of more SAS or product driven and
- 1:13:24call it MMM. Yeah.
- 1:13:26Capabilities and generally speaking, they have a few
- 1:13:30problems and so none of them have really been able to scale
- 1:13:34into the market all that well. Number one, they generally don't
- 1:13:36solve the data problem. They're not making it any easier
- 1:13:38for building tooling to solve the fact that you have to bring
- 1:13:413040 fifty different data sets together and you have to
- 1:13:44normalize it so that it can't. Like a back to school campaign
- 1:13:47is labeled the same in all thirty of those data sets.
- 1:13:49Right. And and and you've lived this
- 1:13:51existence. That's.
- 1:13:52Number of years and taxonomy and label.
- 1:13:54Spent a lot of like sleepless night.
- 1:13:57It is working through data unification and normalization
- 1:13:59across multiple data sets and warehouses.
- 1:14:02It is GM was a good example I always bring up at at New Star,
- 1:14:05remember? It was one of the biggest.
- 1:14:07It still is one of the hardest things to do.
- 1:14:09Yeah. Is to normalize these data sets
- 1:14:12from all of these different platforms, all these different
- 1:14:14campaigns. Yeah, and some of.
- 1:14:15The brands don't even know where all this stuff is.
- 1:14:17Yeah, I mean, if you were to ask a room full of brands who has a
- 1:14:20consistent, up to date and normalized taxonomy across all
- 1:14:24their executions, very few hands are.
- 1:14:25Yeah, Yeah. Especially the bigger brands
- 1:14:28that have multiple sub brands under them, right, Right.
- 1:14:31Different team say there's a variety of reasons why it's
- 1:14:34doesn't exist or if it existed, it's not maintained right.
- 1:14:37Well, and so, you know, generally speaking that's a
- 1:14:40problem in our mind you have to solve.
- 1:14:42You have to make data easy, otherwise you're not going to
- 1:14:44get it consistently and you're not going to get it fast.
- 1:14:46Enough. Yeah, Yeah.
- 1:14:46Yeah, we're we're definitely tackling the data problem head
- 1:14:49on. You can go into kind.
- 1:14:51Of and that's distinguishing from some of these other players
- 1:14:53that don't really yeah they'll plug on top of it but you better
- 1:14:56hope that your. Side I, I don't think of
- 1:14:57bringing an organizational level.
- 1:14:59Yeah, in my mind, you're not going to get scale and you're
- 1:15:02not going to get the speed of update and insight if you're not
- 1:15:06solving the data problem. Yeah, right.
- 1:15:08So how do you guys solve the data problem we talked a little
- 1:15:10bit about before and about semantic layers, right?
- 1:15:13And how do you interpret the data that's underneath?
- 1:15:15It's right, unstructured and otherwise, right.
- 1:15:17And then the agentic layer on top of that, that's the sort of
- 1:15:20instruction booklet, right? So how do you guys, how are you
- 1:15:22guys thinking about? I would think of it this way.
- 1:15:24There's three things that any data analyst team in a
- 1:15:27traditional sense needs to understand to bring this all
- 1:15:30together. You need to understand the media
- 1:15:32ecosystem and how data resides in your Google platforms, in
- 1:15:35Meta and TikTok across or your DSPS, right?
- 1:15:37Yeah, there's a language to how that data manifests, and you
- 1:15:40need to understand it to be able to bring it together and
- 1:15:42normalize it. Yeah, you need to understand the
- 1:15:44brand, right? How did their internal
- 1:15:46taxonomies look? What are they doing?
- 1:15:47What are their strategies? How do they Orient their sales
- 1:15:50data? What are the different P and LS
- 1:15:51that you have to report by, right?
- 1:15:52So you have a center of gravity with marketing data, center of
- 1:15:55gravity with the ecosystem. And then you have to know the
- 1:15:57models or you have to know the MMM constructs.
- 1:16:00How are you actually building different variables, right?
- 1:16:03How do you want to structure the analysis?
- 1:16:04And so just like you would train A-Team across those three
- 1:16:08centers of gravity, we've trained in a Gentic workforce.
- 1:16:11So we have the 1st industry leading end to end, MMM,
- 1:16:15magentic workforce that will bring the data together,
- 1:16:17normalize it. And it's it's almost skills that
- 1:16:20are skills that are we're typically filled by teams
- 1:16:22that's. Exactly right.
- 1:16:23Yeah. We train it the same way.
- 1:16:25It has the same guardrails, the same expectations.
- 1:16:27Here's what to do. This looks good.
- 1:16:28This looks bad, right? And those are those are skills
- 1:16:33and processes now that we continue to develop and iterate
- 1:16:37because instead of one data team being on four or five clients,
- 1:16:41which is typically how a consultancy would work, our data
- 1:16:44agents are learning from everybody all the time across
- 1:16:46the entire business. Yeah, right.
- 1:16:48So the ability to actually learn and fuel kind of the IP sitting
- 1:16:52behind these agentic workforces is an incredible.
- 1:16:55Yeah. And it and it gives, it gives
- 1:16:56the non-technical users, you know, you don't have to have a
- 1:16:59team of data scientists or analytics leaders, right.
- 1:17:01Conversational, conversational interface is that is that so we
- 1:17:05can ask questions, get immediate insights from.
- 1:17:07That's exactly right. Yeah.
- 1:17:08So there's three like broadly speaking, there's three parts of
- 1:17:11the agentic workforce, the data, which is the necessary evil,
- 1:17:15right? Again, you have to get the data
- 1:17:16right. 2 is the data science team that are actually building,
- 1:17:19tuning and deploying the models. And then the last is the MMM
- 1:17:23analyst pool, again with names you.
- 1:17:28Got to you. Got to do that exactly.
- 1:17:31That know how to essentially translate business questions
- 1:17:33like what should my upper funnel strategy be next quarter and
- 1:17:36why? You can ask it conversationally
- 1:17:38and it'll come back. Pull the analysis.
- 1:17:40It knows how to look at response curves.
- 1:17:42It knows how to pull contributions, ROI's, marginal
- 1:17:44ROI, and it'll come back with a perfectly formulated plan with
- 1:17:49recommendations summary for a lead person, right?
- 1:17:52Someone on the media team, someone in finance, someone that
- 1:17:54doesn't have a PhD in, in MMM or advanced econometrics.
- 1:17:58And you don't have to go through a translation layer anymore,
- 1:18:02right? You don't have to be
- 1:18:03bottlenecked by hey, send an e-mail to the team and guess
- 1:18:05what that team is going to have to either de prioritize getting
- 1:18:08the most recent data updated or answering these 15 questions,
- 1:18:12right? You can just get all the
- 1:18:13questions answered essentially in less than 3 minutes.
- 1:18:15That's our benchmark. Yeah.
- 1:18:17And we've seen it be a huge unlock for non-technical
- 1:18:20stakeholders in in our organization.
- 1:18:22Yeah. How long does it take to build
- 1:18:23this? Cuz it, you know, you make it
- 1:18:24sound easy and wonderful, but I've seen some of the hard work
- 1:18:27that goes by me. It's bug bug testing and fixing
- 1:18:30and test. It's always easier said than
- 1:18:32that. Yeah.
- 1:18:32So the company was founded in earnest in around 20 21 OK,
- 1:18:36which was and I think a lot of the like the precursors to a lot
- 1:18:39of the agentic workflows is something like elastic search.
- 1:18:42Yeah. And it was founded
- 1:18:45architecturally to essentially be agentically driven.
- 1:18:48Yeah. And so we've we've.
- 1:18:49So it's a native AI? It's a native AI platform, yeah,
- 1:18:52which is different than bolting on AI and try to migrate in the
- 1:18:56SAS platform. Yeah, We've been able to run
- 1:18:58incredibly fast, Yeah, because we were essentially oriented to
- 1:19:02the sygentic workforce, the get go.
- 1:19:05As. Opposed to having to like
- 1:19:06transition from like super bespoke custom work.
- 1:19:09Like everything is built to work together.
- 1:19:10Yeah. Which just means generally
- 1:19:12speaking, it's it's higher quality, it's faster and it's
- 1:19:15battle tested. Yeah.
- 1:19:16Into the into the models auto optimize.
- 1:19:18I mean, you know, because models are doing kind of weighting of
- 1:19:21different of different things that are going into the model
- 1:19:23from a prediction we're up. Updating, running with over 100
- 1:19:27brands at this point, all of which are updating at least
- 1:19:29monthly. Yeah, right.
- 1:19:31So instead of hey, I updated from six, six months ago or
- 1:19:34times even later, we're producing every single month
- 1:19:37latest and greatest update with coefficients re estimated to
- 1:19:40capture all of the latest trends that are occurring in in any
- 1:19:43business. Yeah.
- 1:19:43Is there a reason for a month specifically, is that, is that a
- 1:19:46statistically significant sample Theta?
- 1:19:48Is it? Yeah, it's twofold.
- 1:19:50I think we're always trying to push it faster, right.
- 1:19:53There is there is an intent to go faster and faster Monthly has
- 1:19:57been the cadence that I think feels most natural right now.
- 1:20:00Yeah. For a lot of brands in terms of
- 1:20:01like the decisions and the cycle times that they have to go
- 1:20:04through. Consider, yeah, considering
- 1:20:05they're making like could be making relatively big shifts in
- 1:20:08budget allocation time, Yeah, you're not going to be doing
- 1:20:11that on a weekly or a daily basis, but you're going to be
- 1:20:13doing that multiple times 1/4. Yeah.
- 1:20:15And I think the the monthly level still gives us a very
- 1:20:18holistic view of the business where you have a number of weeks
- 1:20:22that you can essentially look at performance up and down and it's
- 1:20:24a meaningful chunk of data. Yeah.
- 1:20:26I think fast forwarding what you're going to start to see,
- 1:20:29and this is something we're kind of looking into on a bunch of
- 1:20:31fronts, is less about increasing the cadence of the analysis and
- 1:20:35more about plugging the analysis into other systems that are
- 1:20:38decisioning much faster. Yeah, right.
- 1:20:40So how do you start to inform ESP strategy?
- 1:20:42How do you start to inform all of the auto campaign?
- 1:20:44And when you had a strict level of confidence or whatever where
- 1:20:47the model saying this output is that X threshold.
- 1:20:51Right. Yeah, right.
- 1:20:51Go. And you can push a decision in
- 1:20:53this DSP from a media buying perspective.
- 1:20:55That makes sense, right? Because why wait until you don't
- 1:20:57need 99% when 96% might do? All you need to do is make a
- 1:21:00better decision than you would have otherwise.
- 1:21:02Yeah, like that's the. Yeah.
- 1:21:03And and you know, when we were at New Star, we talked, there
- 1:21:06was a lot of talk about multisense distribution.
- 1:21:08You kind of came from that space, right?
- 1:21:10There was sort of the growth of what they called granular MMM or
- 1:21:14hypergranular MMM, granular. Yeah, yeah.
- 1:21:16Which was to me seemed like a positioning play, but it was
- 1:21:19like, how do you take MMM and adjust it to be more agile,
- 1:21:22faster to deploy and deliver and closer, maybe not multi touch
- 1:21:26attribution, but closer to just decisioning speed?
- 1:21:31And like, how are you seeing that world now?
- 1:21:33And what is the future of MTA? And it doesn't live in this
- 1:21:38model. I think so generally speaking,
- 1:21:41and this is coming from a few different forces that are
- 1:21:43essentially pulling apart what what the promise of MTA kind of
- 1:21:47always was like the quality of identity data, especially when
- 1:21:51you start to look cross-platform and the ability to look at a
- 1:21:54user level and tie a whole bunch of range of experiences and
- 1:21:57interactions together. Now look at it, look at AI,
- 1:22:01right? There's always a new platform
- 1:22:03that you're missing or need to integrate.
- 1:22:05I think we've taken the approach that generally speaking, there's
- 1:22:07a lot of interesting valuable data, but right now it largely
- 1:22:11resides in platform owned and operated clean rooms.
- 1:22:13Yep, like the Google clean room or the Meta clean room, right,
- 1:22:16So you can get to things like reach frequency, audience
- 1:22:19optimizations within those environments.
- 1:22:21Doing it cross-platform, the juice just isn't worth the
- 1:22:24squeeze. Yeah, that's an.
- 1:22:25Interesting point and we've taken the approach.
- 1:22:28Of. So it is the fidelity of sort of
- 1:22:30granular attribution decreases, is that what you're saying?
- 1:22:35I think so, yeah. I think the.
- 1:22:37Because it's buried behind. At a certain point, it's buried
- 1:22:40behind a walled guard. Somewhere.
- 1:22:41Correct. Whereas we're taking the
- 1:22:43approach of coming down to the levels that you need to execute
- 1:22:46from, from MMM, which is all aggregate data.
- 1:22:48It's benign. So to speak.
- 1:22:50Right. You're not dealing with anything
- 1:22:51that that is user level. Yeah.
- 1:22:52So it gives you a much more holistic cross-platform view of
- 1:22:56what's driving performance. It's much easier to incorporate
- 1:22:59new and emerging channels like any sort of paid experiences
- 1:23:03behind a GPT, right? Yep.
- 1:23:05And so we're able to just move much faster and we're able to
- 1:23:08get more relevant insights out even at a pretty granular level,
- 1:23:12like we have creatives, formats, publishers, audiences built into
- 1:23:16the model. And so you by and large, you
- 1:23:18really don't need to go through the time and effort of standing
- 1:23:21up like a cross-platform MTA like capability.
- 1:23:24Yeah, because you can get it from MMM, much so.
- 1:23:26MTA is. Dying.
- 1:23:29They've been saying that for years and it seemed like a
- 1:23:31counter position play by some providers that didn't have MTA
- 1:23:35solutions. That's what we thought about it,
- 1:23:36that we're saying, Hey, we've got MMM, we're going to position
- 1:23:39what we have and and kind of, you know, kind of poo poo the
- 1:23:44the this other solution. But you're saying that like,
- 1:23:46like everything seems to be. Converging everything is
- 1:23:48converging. I think you have MMM really
- 1:23:50asked the tried and true like cross-platform, holistic
- 1:23:55incremental impact kind of growth engine for the marketer
- 1:23:58and that's really where where we're focused.
- 1:24:00I still think there's a lot of value in user level analysis.
- 1:24:03I think it's. Largely and is that is that like
- 1:24:04test and learn experiment. Yeah, you can do audience level
- 1:24:07experimentation. You control off and on.
- 1:24:10Right. Yeah, different populations.
- 1:24:12Yeah, I think there's. And you guys help support some
- 1:24:14of that or? Is it we we take to?
- 1:24:16By and large it's a two way street with with our MMM
- 1:24:19capability. We take test results in as an
- 1:24:22informative prior, meaning it helps calibrate the model in
- 1:24:25areas. Because.
- 1:24:26Testing is a very rich source of of causal impact and then on the
- 1:24:29way out every time our model updates, it actually recommends
- 1:24:32based on areas we might see uncertainty or things that are
- 1:24:35relatively new to the plan. It actually will recommend and
- 1:24:37pump out a test design that the brand and their agency partners
- 1:24:40can take and put into market. So we ingested to help make the
- 1:24:44models stronger, more calibrated based on in market
- 1:24:46experimentation and then to recommend areas where testing
- 1:24:50would help shore up and and bolster the analysis.
- 1:24:52Yeah, yeah. So, so 1, you know, before we
- 1:24:54finish, one area that I thought would be fun to cover was, you
- 1:24:57know, another bane of our existence is dealing, dealing
- 1:25:01with the analysts vendor ranking reports, you know, like the Mmos
- 1:25:04wave, which is by Forrester, which was their measurement
- 1:25:07thing, measurement vendor ranking, you know, and we did,
- 1:25:10we just did a podcast on the CDP Gartner Magic Quadrant, right.
- 1:25:14And generally similar to what we experienced on the measurement
- 1:25:17side, on the CDB side, there were, there were it completely
- 1:25:21missed the sort of mid and long tail of CDPS, right?
- 1:25:24And had, you know, sales force and in some of these big
- 1:25:27players, Oracle in places that that most of the industry was
- 1:25:31kind of up in arms going. What this doesn't really reflect
- 1:25:34what's happening in this industry right now, right?
- 1:25:37There was there was kind of an agentic play, there was sort of
- 1:25:39a a traditional SAS play and there's a bifurcation in the
- 1:25:42market. How do you think the foresters
- 1:25:44of the world are they still doing the MMOS waves?
- 1:25:46So they are There was 1. Both.
- 1:25:49Gartner, they put out a Magic Quadrant and forced to put out
- 1:25:52an OS wave. I think largely it's still
- 1:25:54focused on what I would call like the legacy of the more
- 1:25:57encompassed. So they had quite adapted to.
- 1:25:59Yeah. You don't see a lot of the newer
- 1:26:01players that are approaching it differently than your legacy
- 1:26:06consulting model, essentially. I think that'll change, to be
- 1:26:08honest, because now we're seeing a ton of engagement from the
- 1:26:11market and we're growing to the point where you're going to have
- 1:26:14to be on the wave, like one way or the other, yeah.
- 1:26:16Which is good. Yeah, and it's important.
- 1:26:18I mean, you mentioned that that being, being important,
- 1:26:20especially in this space, being considered in that top five from
- 1:26:23a vendor selection perspective for our RFPs is still a critical
- 1:26:27play. I mean, despite we talked about
- 1:26:29this in the pod yesterday about how well, you know, RFPs can
- 1:26:32often times be largely unsuccessful.
- 1:26:34Like there's like an average industry rate.
- 1:26:36I don't know if this is for a minimum specifically, but about
- 1:26:3810% hit rate reading meaning like you'll apply, you'll,
- 1:26:41you'll fill out, you know, 10. Yeah.
- 1:26:45For, for Mutinex, we've done, the team has done an incredible
- 1:26:48job really scaling out and are they are the predominant, we are
- 1:26:52the predominant player in the Australian and broader 8 pack
- 1:26:56market. So I think it's an interesting
- 1:26:58journey for us now taking a tried and true enterprise grade
- 1:27:02AI workforce essentially deliver and deploy and and bring MMM to
- 1:27:07life. And we're really just
- 1:27:09transplanting it and growing it within the US and more, yeah,
- 1:27:13essentially across the globe, so.
- 1:27:14And it seems like it'll have a disruptive, potentially a
- 1:27:16disruptive impact on the market from what I know about the
- 1:27:20vendors in the space. And what, you know, I think
- 1:27:21marketers, yeah, marketers need to reassess their expectations
- 1:27:25around how fast the the economics around what MMM needs
- 1:27:29to be right. The bar is much lower with an
- 1:27:31agentic workforce, Yeah, in terms of just cost and then to
- 1:27:34the relevance and speed that they can get insights, Yeah.
- 1:27:37And drive better decision making across their organization.
- 1:27:39Yeah, the time to value that's, that's a big well, Hey, that's
- 1:27:42thank you Mike for for joining us.
- 1:27:43But that was awesome. It's always good to to walk out.
- 1:27:46And it's a topic that we sort of skirt around.
- 1:27:48I've certainly had experience, but we don't really dive deep.
- 1:27:50And so we should we should have you on the on a full podcast
- 1:27:53session to keep going further into sort of the MMM world.
- 1:27:55Measurements having a moment I would love.
- 1:27:57Yeah, yeah, it's, it's a moment that I've been involved in for
- 1:28:00years. But thanks very beauty for
- 1:28:01joining and we will. We'll see you next.
- 1:28:03Time here, signal and noise and we are a possible event in Miami
- 1:28:08and we are midway through day #2 another beautiful day here in
- 1:28:13Miami Beach. And a special guest is Andrea
- 1:28:15Coats, Auto CRO at Aquifer. Welcome.
- 1:28:18How's it going real? It's going real well, yeah, good
- 1:28:21to have you I know we talked about doing this for a while and
- 1:28:23I hope we can make it happen. Impossible.
- 1:28:24So maybe for people who don't know you and don't know Aquifer,
- 1:28:29maybe start off by telling the guests a little bit about
- 1:28:31yourself and your background. Yeah, absolutely.
- 1:28:33So I'm actually fairly new to Aquifer.
- 1:28:35I got here in October of last year and I spent most of my
- 1:28:38career on the DSP side of the business.
- 1:28:40So, you know, way back when at value Quick Media, which became
- 1:28:44Epsilon and had all sorts of stuff there and then rocket fuel
- 1:28:47pre IPO all the way into the seismic years saw that DMP, you
- 1:28:51know, ad server, the Pier 39 data business, all sorts of
- 1:28:55different things within the market.
- 1:28:56And then more recently, right before the pandemic, I went to
- 1:28:59the beeswax, helped grow out kind of the tech and media
- 1:29:02vertical and then the agency vertical.
- 1:29:04And then after we got acquired by free Roll, I ran sales there.
- 1:29:07Did the free will identity with your Lux Nari and all those
- 1:29:10guys. Yeah, exactly.
- 1:29:11Small world. Yeah, Joe is my counterpart on
- 1:29:13the brand side baton from Ari and Stride check himself.
- 1:29:19Yeah. Yeah.
- 1:29:20And it was fun. It was fun because it was my
- 1:29:21first exposure to the supply side.
- 1:29:24Everything I'd done, you know, I touched CTV LA, but just to be
- 1:29:28at the epicenter screaming was fun.
- 1:29:30Spend a little time before hired a new CEO.
- 1:29:33I was brought in as CRO and we're really, really chugging
- 1:29:37along, which is a lot of fun. I think your new CEO is actually
- 1:29:39out in Denver. We're going to try to get
- 1:29:40together over the next next few weeks.
- 1:29:42Schedules haven't aligned, but looking forward to that.
- 1:29:44Yeah, Bill and I are both pandemic escapees from New York.
- 1:29:47So, you know, we we tell people we're there because we're there
- 1:29:51so much. But you know, he's in Denver and
- 1:29:53I'm now in Philadelphia. Honey, I still have a 917 number
- 1:29:56despite I've been in Denver. It was pretty pandemic.
- 1:29:58I went out. It was in about 10 years kind of
- 1:30:02the first wave of my wife is from Denver, which is how I
- 1:30:04ended up out there, but first wave of kind of people from the
- 1:30:06East Coast going there. But definitely a lot of us now
- 1:30:09have escaped New York and, you know, go to Denver for the
- 1:30:11mountains and a lifestyle. But yeah, it's a good place.
- 1:30:13But I'd tell some a millennial and I tell people for most folks
- 1:30:17who are millennials, you can tell where people lived at the
- 1:30:20end of college when they got off their parents cell phone plan
- 1:30:25because their area code is too funny from that location.
- 1:30:28And, you know, I'm from Boston and I'll see a 617.
- 1:30:31He's like, oh, you're from Boston.
- 1:30:32It's like, no, no, no, I was just there when I got my cell
- 1:30:34phone. Now, you know, now it's like
- 1:30:36your Social Security number follows you.
- 1:30:38And we'd never want to get rid of that.
- 1:30:40Right. Yeah, Well, it won't hold it
- 1:30:41against you from Boston being a New York sports fan.
- 1:30:44But I did go to school in New England.
- 1:30:46So in the belly of the beast. But I think a lot of people like
- 1:30:49Aquifer like works with some incredible companies, including
- 1:30:53lots of the agencies and the big hold codes.
- 1:30:55But I think a lot of people maybe like, because you work
- 1:30:58behind the scenes, don't know what Aquifer does.
- 1:31:00So maybe you can spend a moment or two telling viewers about
- 1:31:03that. Yeah, absolutely.
- 1:31:04So Aquifer has actually been around for a while.
- 1:31:07It's a 7 year old company. And at a high level, what we are
- 1:31:10is a marketing data engine for ad tech and martech.
- 1:31:13I was initially founded by the CTO of Yahoo and the chief Data
- 1:31:16Officer of AOL C Real entrepreneurs, senior people
- 1:31:20have been in the space forever. And they found that they kept
- 1:31:24solving the same data problems for everyone.
- 1:31:26And at Tech Mart Tech, and they said, what if we could find a
- 1:31:30single piece of technology that could solve those problems?
- 1:31:33Because it doesn't matter whether you are a data company
- 1:31:36or you are a publisher or you are an agency holding company or
- 1:31:40you're DSP or an SSP, you all have the same problems.
- 1:31:45And So what they built is Aqua for Fusion.
- 1:31:48And what Aquifer Fusion effectively does is it allows
- 1:31:51folks to deploy new features that are better up to 10 times
- 1:31:55sooner, 20 times faster, and at half the cost because down on
- 1:31:59cloud compute costs a lot, right?
- 1:32:02Yeah. And that's a big kind of helpful
- 1:32:04component. And actually, you know, we've
- 1:32:06kind of decoupled the cloud piece from the rest of the stack
- 1:32:09because the aquifer of a few years ago, you know, we ran
- 1:32:12everything. We happened to be an AWS shop.
- 1:32:14We ran everything on AWS on our instance, but over time, a lot
- 1:32:18of our clients because it's completely private label, right?
- 1:32:20One of the reasons why you don't know us is that everything we
- 1:32:23have in market, you know, 60 different companies in market,
- 1:32:25we are the tag that powers them, for instance, right?
- 1:32:28And one of the reasons why you don't know is that we actually
- 1:32:31deploy in their cloud instance. And once we started doing that,
- 1:32:34we said, well, we don't only have to work on our own ad tech,
- 1:32:37big data stuff. We could do any job that saves
- 1:32:40some money on cloud computing. We were talking before this, the
- 1:32:45term 0 copy came up a little bit and I know there was some
- 1:32:48debate. Do we ask about it?
- 1:32:49Do we not? But I think that there's like it
- 1:32:52comes up from time to time. I think there's been a trend
- 1:32:55maybe the last year where it's become coming up more and more
- 1:32:58pitches like what Like maybe tell the viewers, what is it
- 1:33:02like? What is like real zero copy
- 1:33:04versus people just saying a zero copy?
- 1:33:06And why is this become such an important issue for a lot of
- 1:33:08brands? Yeah, good question.
- 1:33:10So I have a little bit of kind of imposter syndrome talking
- 1:33:12about it because I'm not an engineer, right?
- 1:33:14But it is an area where I've learned a lot, right?
- 1:33:16It's it's funny, we're talking about my background, right?
- 1:33:18I've spent my entire career in the and I feel like, you know,
- 1:33:21to the degree that people on the commercial side are experts, I
- 1:33:24feel like a real expert in the space.
- 1:33:25I know a lot coming to aquifer. We don't touch media or the big
- 1:33:29stream at all, but all of our clients do, right?
- 1:33:32And so I've learned a lot about the space to your questions
- 1:33:35specifically. So zero copy is this idea.
- 1:33:38If I have if I'm a multi cloud enterprise for instance, I might
- 1:33:41have some data in AWS, I might have some data in GCP, I might
- 1:33:45have some data on Prem with a client if I want to query all of
- 1:33:49that data. A lot of times when you're doing
- 1:33:51a big data job, you create a temporary copy somewhere where
- 1:33:55you can get it in a single schema.
- 1:33:56Another example might be maybe I have attribution data from an AD
- 1:34:00server, or win logs and bitstream data from my DSP.
- 1:34:03There might not be a common key between those two and I have to
- 1:34:07create a temporary copy to actually make that connection.
- 1:34:10The idea of 0 copy is can you actually in storage right?
- 1:34:15Can you, in storage query all of those databases and treat all
- 1:34:20that information as singular and bring it together without moving
- 1:34:23data or shipping too? I think that's the key, right?
- 1:34:25And to your point about like what 0 copy is real and what 0
- 1:34:29copy isn't, the reason why it's so important we call it compute
- 1:34:31forward to be in a client's VPC. There are a lot of companies
- 1:34:34that claim to do 0 copy, some really large, you know,
- 1:34:37hyperscalers, but because they're on their own cloud, they
- 1:34:41do zero copy. Guess what, the first step they
- 1:34:43have to do before. So that's a copy of the data to
- 1:34:46their ecosystem. So it's not really zero copy and
- 1:34:49it's not really powering the client's actual interface.
- 1:34:52And so the fusion technology, at least on our side that we've
- 1:34:55built, there's a lot of reasons why it's so much more effective.
- 1:34:59But the two things that it's able to do 1 is true zero copy.
- 1:35:02And two is most big data jobs are done in something called
- 1:35:06Spark, which is kind of a lingua franca for big data, at least in
- 1:35:09adtech and martech. Spark usually is done in memory
- 1:35:12and RAM and we have a way to do it with 0 copy, but then also in
- 1:35:16storage. And so we just use a ton less
- 1:35:19compute as we're doing that. Actually, you do a lot of work
- 1:35:22with Databricks as a result of the.
- 1:35:24Yes, yeah, exactly. Just Spark.
- 1:35:25OK. And one of my favorites rules
- 1:35:27and both business and just in like theoretical economics is
- 1:35:31Jebbins paradox. I don't know if you're familiar
- 1:35:33with it, but it's the idea that actually as something gets more
- 1:35:37efficient, it's used more. So the old Industrial Revolution
- 1:35:40example is there were factories that were on rivers with water
- 1:35:43wheels, right? And factories with not very
- 1:35:46efficient steam engines that had to use coal away from that.
- 1:35:49And the water wheel guys were like, well, I'm not going to
- 1:35:51build a steam engine because I've got the water wheel.
- 1:35:53And then as things got more efficient, the coal companies
- 1:35:56were like, no, we're not going to use this much coal, right.
- 1:35:59But it actually turned out it got so cheap that the water well
- 1:36:01guys moved to steam engines, right, doing more coal use.
- 1:36:04The same thing is true for someone like data bricks.
- 1:36:07We're able to get data bricks so efficient on Spark that our
- 1:36:11clients are actually using more of it because the utility of it
- 1:36:14has way gone up. And so we actually, you know,
- 1:36:17are working with data bricks to do a lot of that implementation,
- 1:36:19you know, not just for data bricks, but across, you know,
- 1:36:22any cloud or or big data processing environment that our
- 1:36:25clients use. This just shows that there was
- 1:36:26all of this pent up demand, right?
- 1:36:28For you for that processing power, right.
- 1:36:30And for, for, for the storage. And then it was just too
- 1:36:33expensive right now the costs have fallen.
- 1:36:35I think people have seen utility.
- 1:36:37Yeah, I think we're seeing a lot of that with, as AI impacts
- 1:36:40different tasks and jobs as well that, you know, especially
- 1:36:43engineering, for example. I think we're saying that
- 1:36:45there's a ton of pent up demand for engineering and that, you
- 1:36:47know, OK, just because a lot of code is written, it doesn't mean
- 1:36:51that every developer suddenly going to be out of work.
- 1:36:54And what it probably means is that there's going to be just,
- 1:36:58if you look at the graph number of apps shipped, websites built,
- 1:37:03it's just gone like thermonuclear rights, like a
- 1:37:05hockey stick growth because of this.
- 1:37:08And everyone's benefiting from that.
- 1:37:10So, yeah. And I see it especially in
- 1:37:12prototyping, like a lot of my front end folks, like our head
- 1:37:16of applications in AI and generic, for instance, he has
- 1:37:20built out a whole bunch of prototypes because the viewers
- 1:37:23can probably understand, like we do a lot of cool stuff, but it
- 1:37:27takes a little bit of time to explain.
- 1:37:28And So what we really wanted was a set of demos so folks could
- 1:37:32touch and feel it. But over time the demos our
- 1:37:36clients have seen them and they're like, and I and I get
- 1:37:38that, you know, it works specifically on our own stuff
- 1:37:43and and in general folks will bring third party apps to power
- 1:37:46their AI. Like we work with some of the
- 1:37:48best and brightest in the industry.
- 1:37:50We have our remote MCP server that can hit, but in some cases
- 1:37:53for purpose built stuff, we are starting to production Alize
- 1:37:56some of our apps, and I don't think that would have happened
- 1:37:59without well, they are authentic in nature.
- 1:38:02So of course it wouldn't have happened with AI.
- 1:38:03But even sort of like the quote UN quote vibe coding life work
- 1:38:06style where, you know, my front end engineers can now production
- 1:38:10as something on our infrastructure that it has
- 1:38:13utility for clients in a relatively short period of time.
- 1:38:16That's something people don't fully understand or there maybe
- 1:38:19they're not being honest about is, you know, because criticism
- 1:38:22is all these vibe coded apps, they're not production ready.
- 1:38:24They're of course they're not right.
- 1:38:25It's like it's a it's a demo. You put it together in two days,
- 1:38:28But guess what, Like I've worked in like, you know, product
- 1:38:32quote, UN quote for for almost 20 years working with design
- 1:38:34teams, engineering teams that would have taken weeks before
- 1:38:37you would have had to requirements gathering, build,
- 1:38:40build user stories, have your have your epics lined up.
- 1:38:45It would have. You're going to transfer one
- 1:38:47team to the other, going from prototype to clickable prototype
- 1:38:50to, you know, them pulling, I mean.
- 1:38:53You're. Two days now.
- 1:38:55I mean, it's not production, but it's something you can get in
- 1:38:57someone's hand, you get feedback and to your point, the client
- 1:38:59gets excited. Yeah, and it's, it's a new power
- 1:39:02dot and you know, like that part of it.
- 1:39:05So to your .10 years ago, APM would have drawn it on a
- 1:39:09whiteboard. An engineer would have tried to
- 1:39:11prototype it, bring it back like you said, bring in the
- 1:39:13requirements, talk to the commercial teams, etcetera.
- 1:39:16Designer built, right? Design the front end, slice it
- 1:39:19up. Someone can have an idea, make
- 1:39:22it look roughly like it should, and then if everyone's like,
- 1:39:27yeah, that's what it should be, or you edit it a little bit,
- 1:39:29then it goes off to design and it goes off to the engineering
- 1:39:32team to actually hook it up right.
- 1:39:34And so, you know, this idea of engineers are going to go away
- 1:39:38because so much of this can be done here.
- 1:39:40It's like, no, no, no. The the new paradigm is things
- 1:39:43are going to ship really quickly.
- 1:39:45Like things are going to ship at the speed that as a salesperson
- 1:39:48I always thought thought they should, knowing nothing about
- 1:39:50how it was right. And so for that reason, you
- 1:39:55know, you're going to need all those folks still, but they're
- 1:39:57going to need to be a much more higher value task because the
- 1:40:01person who actually has the idea and understands, you know, the
- 1:40:04Nexus of everything can actually shift the prototype.
- 1:40:06It's a good point. I mean, if your job today is
- 1:40:08mostly doing things that are wrote repetitive, can be easily
- 1:40:11automated, then I think those types of jobs are probably more
- 1:40:14at more risk. But like if that stuff is.
- 1:40:19Think you do it because you need to, but what you really want to
- 1:40:21do is the higher value stuff that I think those people will
- 1:40:23be totally safe. They'll like a little more time
- 1:40:25to do that, which would be good to looking at AI.
- 1:40:27Like what? Like is anything that you're
- 1:40:29you're working on aquifer that you think is interesting or even
- 1:40:32maybe not even add just generally any like, I don't know
- 1:40:34what you're able to share for your road map.
- 1:40:36Any new features or any interesting things you can tell
- 1:40:39the viewers about? Yeah, absolutely.
- 1:40:41So traditionally Aquifer, so we kind of think of the data
- 1:40:46ecosystem in three different layers, right?
- 1:40:49So there's sort of the operational layer, which is
- 1:40:51like, where is my stuff stored? What's the governance around it?
- 1:40:55Like what am I allowed to do? Right.
- 1:40:58And whatever jurisdiction you're in.
- 1:41:00Then there's the processing layer.
- 1:41:01That's really where we are, right?
- 1:41:03We have our underlying fusion technology.
- 1:41:05I don't think I really talked about our modules for ad tech
- 1:41:07and MAR tech that are built on that, But we have data
- 1:41:10collection ingestion, which is our tagging and ingestion.
- 1:41:14We have an identity piece which dissimilar to most, most of the
- 1:41:18identity folks out there have a graph and that's their IP.
- 1:41:21We don't own any data, right? For us, it's choose your cloud,
- 1:41:24choose your spine, 30 days, you have a graph, you can use our
- 1:41:27model or you can renovate it or write your own, right?
- 1:41:30And then the third piece is our enablement suite.
- 1:41:33And the best way I can describe our enablement suite is, you
- 1:41:37know, if you're a brand and you want a CDP, probably want a
- 1:41:39license to CDP, there are a lot of great CDP's out there despite
- 1:41:43what, you know, the market sometimes says.
- 1:41:45However, the dozens of CDP's out of that is, you know, probably
- 1:41:4910:15 that are sure. I think they're, you know, have
- 1:41:52these market share very good. But you're right, there's plenty
- 1:41:54of good solutions out there regardless of what happens to
- 1:41:55the category. And I think you can even debate
- 1:41:58is it one category. But you're right, there are
- 1:41:59plenty of great solutions out there.
- 1:42:01Yeah. And we can talk more about the
- 1:42:02category in a second. But I do think and our POV is
- 1:42:07that there are a lot of other companies that have other
- 1:42:10pockets of IP and pureplay CDP's were ahead of the market and now
- 1:42:15the incumbents with other parts of IP are wanting to in house
- 1:42:18CDPS capabilities and that's where we come in.
- 1:42:21That's where like it's still there UX, it's still their
- 1:42:25cloud, it's still their IP. We're just kind of connecting it
- 1:42:28all process. And then the top layer is the
- 1:42:30interaction layer. And traditionally again that the
- 1:42:34old joke used to be aquifer. We took the UI, EU and the I out
- 1:42:37because the UI is yours. We don't say that so much
- 1:42:40anymore. It's kind of a joke.
- 1:42:41Was that actually done on purpose or was that just, you
- 1:42:43know, I don't know whether that was a joke internally.
- 1:42:47Again, I haven't been here that long or whether they did that on
- 1:42:49purpose, but it is kind of a fun thing to say, especially if, you
- 1:42:53know, our buyers want to situate themselves.
- 1:42:56I always wondered why the Valves machine?
- 1:42:58OK, yeah, there's a lot of, there's a lot of talk about that
- 1:43:01now too. But what's funny is like a lot
- 1:43:03of people are going from a traditional UX to a gentic
- 1:43:06workflows. So I said we're doing some of
- 1:43:08this stuff internally. We have a tool, for instance, a
- 1:43:11lot of our clients bring us really on structured data.
- 1:43:13So one of the first agents we did was like, I think we have 13
- 1:43:16standard schemas internally. I was like, and agents should be
- 1:43:20able to get us as close to one of those 13 standard schemas as
- 1:43:23possible. I remember back, you know,
- 1:43:25beeswax was always like, oh, our partnerships team needs to make
- 1:43:28sure data Rama as a bead beeswax adapter and you know, look at as
- 1:43:31a beeswax adapter and so fight against abuse.
- 1:43:33You have to maintain all those, right?
- 1:43:35You have to maintain those. Somebody has to do the contract.
- 1:43:37Somebody has to build it. It's just like this column is
- 1:43:39this imagine you're managing 200 of those, right?
- 1:43:42I mean, it's just that's, that's a business almost in your other
- 1:43:45cell, right? It is.
- 1:43:46And just to be able to adjust that and have an agent be like,
- 1:43:49this is probably right and it is probably right most of the time,
- 1:43:52which is great. We have another cool thing we
- 1:43:54call Talk to My Data where a lot of clients bring and house a ton
- 1:43:59of different marketing data with us.
- 1:44:00And mostly, at least historically, it's been either
- 1:44:03their analytics team running SQL or their data science team
- 1:44:06running Python. We now have a chatbot or any
- 1:44:08idiot like me can say like what's my best performing
- 1:44:11campaign and that'll come through.
- 1:44:13It'll also write the sequel on the side, right?
- 1:44:15And so we have some other things like right graphing and
- 1:44:18discovery. One kind of road map thing I can
- 1:44:21share is a lot of our customers, we work in very sensitive
- 1:44:25industries like finance and pharma, for instance.
- 1:44:29And a lot of times they say we want to use an LLM but we don't
- 1:44:32want to contribute the IP. Of the LLM.
- 1:44:37So we have created this thing which so far is really just a
- 1:44:42training tool. We call it Aquabot.
- 1:44:43It's in our Slack and it trained me, but it's the Acrobat is the
- 1:44:48Slack interface. But it's Peace the private AI
- 1:44:51chat environment where we're right now working in a claw, but
- 1:44:54it could be really any LLM and we can actually take these
- 1:44:59different tools we've built as skills within this chat
- 1:45:02environment and discuss and nothing gets back to the cloud
- 1:45:07model. It's a good containerized
- 1:45:08version where like you're interacting with your local
- 1:45:10model yeah, you're using the LLM, but like nothing's been
- 1:45:13you're not sharing that in fashion.
- 1:45:14I can see why for brand like you're right, I mean like
- 1:45:17foreign agency, right like ours yeah, especially I mean like
- 1:45:21meant like, you know, some of them might not care.
- 1:45:23I think generally they will care right.
- 1:45:24They OK, this is our IP is their clients.
- 1:45:26I think we either don't want to or can't share this for sure.
- 1:45:30But one of the things like I talked about how companies are
- 1:45:33becoming CDPS, I think one of the major fundamental shifts
- 1:45:36that's happened, it's right like Omnicom merging with IPG with
- 1:45:40Axiom Epsilon going to Publisa, Smirkl going to tend to do.
- 1:45:44But even some of the independent shops like Blue being built by
- 1:45:47Horizon or Alley Feeling built by PMG.
- 1:45:50You know, it's really important that it's not just a services
- 1:45:54layer because folks need a lot of help on the services side.
- 1:45:57But you know, if you're a brand, it makes sense that you'd want
- 1:46:00everything in one place and we'd want a services partner who can,
- 1:46:04you know, treat your data properly end to end.
- 1:46:07And so a lot of those platforms, you know, have kind of grown up
- 1:46:11relatively quickly and, and need sort of the underlying CDPS
- 1:46:16functionality that powers. Thanks see for an agency too.
- 1:46:19I mean, if like the, if the goal is to big add more value, bring
- 1:46:24your own IP to clients. I mean, like all these
- 1:46:26investments, you call it, whether it's Horizons Blue or
- 1:46:28you know, what we're doing at Omnicom with Omni and like
- 1:46:34acquiring systems integrators, you know, like Cordero, for
- 1:46:37example, where I am or like Sapien publicist, right, who can
- 1:46:40bring the, the technical acumen, the engineering skills, but then
- 1:46:45also building your own kind of infrastructure and IP.
- 1:46:47It's almost like when I wrote about this a few weeks ago, the
- 1:46:49agency is more becoming, I mean, I think the goal is to be more
- 1:46:53of an operating system that can bring the classes here to run on
- 1:46:55us, right? And it also creates A Moat,
- 1:46:57right? Because another thing we've seen
- 1:46:59is brands are changing agencies a lot more kind of, you know, I
- 1:47:03don't want to say keeping more arm's length, but particularly
- 1:47:07on the media side, you know, there's, there's a lot of
- 1:47:10pressure there, right? And so I don't disagree with
- 1:47:12you. Yeah.
- 1:47:13So being able to have a value proposition around what you're
- 1:47:18doing with the media is really important.
- 1:47:19I think there's more pressure, there's certainly margin
- 1:47:22pressure. There's a there's interest in
- 1:47:25having more transparency. And I think with more
- 1:47:27transparency, if you give clients more transparency,
- 1:47:31they're smart clients and you see the smart ones ask for
- 1:47:34transparency, they're going to find things they don't want to
- 1:47:36pay for and your margins probably will go down.
- 1:47:38So I think that the premise would be OK as an agency, you
- 1:47:40have to have other services you're charging for, right?
- 1:47:43And maybe you know that, you know, I mean, looking back,
- 1:47:46remember HST split off the media business like what, 80s
- 1:47:50nineties, right? So yeah, I've met kind of coming
- 1:47:52full circle now where you have to have creative, you have to
- 1:47:55have strategy, you have to have analytics and you know, media is
- 1:47:57part of it, right? But if you really want to drive
- 1:47:59towards outcomes, you need the whole picture.
- 1:48:01It actually is funny because I feel like you watch Mad Men,
- 1:48:05right? And they're doing some really,
- 1:48:07yeah, really interesting strategy work.
- 1:48:09And at least when I got into the industry, working with a lot of
- 1:48:12media agencies, they seemed pretty removed from the
- 1:48:14strategy. It was just like, you know, the
- 1:48:17planning team was as strategy as we got, and that it was just
- 1:48:20like there's pretty senior people doing the media planning,
- 1:48:23but near the throat, over the wall, find the vendors, get the
- 1:48:27message out there and get the broadcast rates as low as
- 1:48:29possible. But what's funny is now that the
- 1:48:32data lives with the agency, they actually have the best.
- 1:48:35Fuel to drive the strategy. And so when I'm sure of that
- 1:48:39loop, correct. Yeah, totally is like a lot of
- 1:48:41these shops are servicing some really cool insights in a way
- 1:48:44that, you know, it would take a, you know, white shoe strategy
- 1:48:49firm like a Bean or a McKinsey a lot of work to to get those
- 1:48:54insights to even do it. I think that's why a lot of them
- 1:48:56are trying to move like down from like big ass pure strategy
- 1:49:00into more operational and data and tech stuff.
- 1:49:02They're trying. I mean, I know you know, PCG
- 1:49:04has, you know, I think this gamma was there like their data
- 1:49:06science division. You have definitely run into,
- 1:49:08you know, Mckenzie's doing more agency selections and marketing
- 1:49:11transformations. You know, they have a publisher
- 1:49:13offering. So I think a lot of retail media
- 1:49:14as well, they do and like they're doing retail media
- 1:49:16strategy. Not I mean, you could debate
- 1:49:18whether they should be or if they're any good at it, but I
- 1:49:21don't know like that. But I know I've, I've had many
- 1:49:23clients who work, work with them on that.
- 1:49:25But it is interesting how the how the space is changing and I
- 1:49:30don't know. I mean, like, I don't know, when
- 1:49:31I got into media, I started as a media buyer, right?
- 1:49:33You know, it's my first job out of grad school.
- 1:49:37You know, these are these are not high paying job.
- 1:49:39You realize, like that's how media is actually bought and
- 1:49:41trafficked, you know, and inactivate.
- 1:49:43It's just a bunch of young people working, working crazy
- 1:49:45hours doing this. I skipped this part of my
- 1:49:47resume, but my first paid job in media was an intern at an
- 1:49:54Omnicom on PR firm in DC. Actually, it was.
- 1:49:58It's better to fail fast. You know, I don't know, I
- 1:50:01thought, I thought I just, I didn't get it.
- 1:50:04Well, PR back then was just writing press releases and
- 1:50:06stuff, right? I mean, well, it's funny because
- 1:50:09I came on the scene right when social media and the Internet
- 1:50:12were really becoming a thing. And what I saw was a whole bunch
- 1:50:16of vice presidents that were tearing their hair out because
- 1:50:18they used to be able to call The Washington Post, New York Times
- 1:50:20and kill a story. And then it would pop up on
- 1:50:22Facebook. And they're like, why did this
- 1:50:24happen? I told them not to let it out.
- 1:50:25And it was like, because they're not in control either.
- 1:50:28And you've lost control of the narrative, right?
- 1:50:30It's so funny. Yeah.
- 1:50:31But it's it's been it's been wild just seeing all the
- 1:50:34different iterations and things. I don't know.
- 1:50:36Where do you where do you see it going?
- 1:50:38Like being as a consultant, but within a traditional media
- 1:50:42agency. I mean, there's got to be value
- 1:50:45to that, right? Because you're, you've been
- 1:50:46there less than you. Yeah, I joined Kodera, which is
- 1:50:50you really the Omni kind, like Omni comms advisory arm.
- 1:50:53You got an engineering arm just kind of like a consultancy.
- 1:50:55What you see like at the committee version of extension.
- 1:50:57But I think we're probably less focused on like offshore
- 1:51:00delivery. We have some offshore teams, but
- 1:51:02the premise here would be like clients need help with
- 1:51:05engineering and Omnicon meets to bring a more rounded package
- 1:51:09that includes like partnerships with hyper scalers.
- 1:51:12Like, yeah, I almost think it's table stakes that you know, how
- 1:51:15to work with AWS, you know how to work with GCP, you know how
- 1:51:18to work with Azure and and the big hyper scalers, you know, how
- 1:51:21to build technology solutions, whether it's for advertising use
- 1:51:25cases, marketing use cases, because The thing is the
- 1:51:29traditional agency model where you have a bunch of senior
- 1:51:31people planning, junior people executing.
- 1:51:34I mean, who's going to do that, right?
- 1:51:35There's just I don't I just don't think it's a model that
- 1:51:37lets yourself all of that because it wasn't designed that
- 1:51:39way. So then you look at the systems
- 1:51:41generators, accentures the Deloitte's I can't I came from
- 1:51:44Salomon kept Gemini before this. I mean, that's just what they do
- 1:51:47for a living. So they're good at it.
- 1:51:50And I think competing in that space is going to be
- 1:51:51challenging. But the what I think is
- 1:51:52interesting, the valve prop of what Publicis is built.
- 1:51:55What we're building an Omnicom is, you know, you take the
- 1:51:59traditional agency, Business, Media, strategy, creative, you
- 1:52:03bolt on the data. Side of that.
- 1:52:05Axiom, right. And then what we're building in
- 1:52:07Omni and our own platform and then you add systems
- 1:52:10integration. There should be a flywheel
- 1:52:14there, right? It should be very compelling
- 1:52:16where we can if clients need better technology in order to
- 1:52:19get better results and outcomes from media, we can be there for
- 1:52:22them as well. So that's the premise is
- 1:52:23working. It's working well.
- 1:52:24It seems so obvious. Your data is already there.
- 1:52:26You might as well, you know, rely on your expertise to to
- 1:52:30surface it. Yeah, create some signal from
- 1:52:34the noise of you 100% right. Yeah.
- 1:52:36So that's what's it's working well.
- 1:52:37And you know, the space is lots of evolution.
- 1:52:39It's exciting. I mean, like it's funny to like,
- 1:52:41you know that think some Goodfellas.
- 1:52:42Every time I try to get out, they call me right back in like
- 1:52:45started media, got into consulting, got into Martech and
- 1:52:49then it's funny then, you know, here I am like a couple decades
- 1:52:52later, you know, I'm working with a lot of the same people,
- 1:52:54which is great. I mean, I, I love people in
- 1:52:56media and people on that tech. I think it's the most the most
- 1:52:58fun community even a part of. And I've gotten to work with
- 1:53:01great people like, you know, like you folks at Aquifer, you
- 1:53:04know, it's been fun and I'm excited about your new role and
- 1:53:06what and what you're doing. Yeah, me too.
- 1:53:08I mean, what I've found is at first, like I said, we don't
- 1:53:14have AUI, we have demos and stuff.
- 1:53:16But when you have a platform, you sort of have a decision
- 1:53:19Treaty of things. It's like, OK, they slot in
- 1:53:21here, they slot in here and aquifer, it's so much more about
- 1:53:25just sitting there and listening because every problem does
- 1:53:29roughly fit into a module. But there's much less lands than
- 1:53:34it is, you know, like a river with with pipes that can kind of
- 1:53:39meander. And so it's a breath of fresh
- 1:53:43air that every single conversation is it's a real
- 1:53:46challenge and you're helping people create something within
- 1:53:50their own product that will be their intellectual property.
- 1:53:53I'm a big believer that in AT tech and Martech, there are five
- 1:53:56things that truly differentiate companies.
- 1:53:59Unique supply, unique data, unique decisioning or modeling,
- 1:54:03some unique user experience, whether it's a genetic or
- 1:54:06traditional or services, right? And what I loved about Aquifer
- 1:54:11and what I love about what I do is everything we do, it's kind
- 1:54:15of in the palm here where we are just powering those things, but
- 1:54:19we don't actually own any of that ourselves and we don't
- 1:54:21touch any of that ourselves. I like that.
- 1:54:23I think that the fifth one, the services 1 is frequently
- 1:54:26overlooked, right? But I think that as we go into
- 1:54:28more AAI and a gentic like being, I mean, you look what
- 1:54:31Palantir did with the forward deployed engineers.
- 1:54:33I mean, there's a reason why that works so well because if
- 1:54:35you're going to be taking incredible technology that's
- 1:54:37changing quickly, match it with, match with unique data sets like
- 1:54:41that, like, you need that in order to make it happen.
- 1:54:44So, so well. It's exciting.
- 1:54:47Well, thanks again for coming over and joining us.
- 1:54:49This is a good conversation. And this could be a lot worse.
- 1:54:53Yeah. I don't think they can see our
- 1:54:54view, but this is. Yeah, we'll give them a preview.
- 1:54:56But, yeah. But life is good down here in
- 1:54:58Miami. Thanks to everyone for tuning
- 1:54:59in. Thank you so much.
- 1:55:02OK. So hi, Jill.
- 1:55:06Hi, Jill Randall. So I'm very glad to have you on
- 1:55:09the podcast today. We are sitting here in the
- 1:55:11trestle tower above the Fontainebleau Hotel looking out
- 1:55:15across Miami is really beautiful, and we're very lucky
- 1:55:18to be here. Thank you so much for joining
- 1:55:20us. Jill, you're from Fuse, am I
- 1:55:22right? Yes, use use dot IS.
- 1:55:26And a fuse in a minute because I just, I read your website.
- 1:55:29I've been talking to you about it as well.
- 1:55:30I want to actually just start. With you.
- 1:55:33You've obviously had a the storied career yourself, you've
- 1:55:37been a big tech, and now you're in the entrepreneurial sort of
- 1:55:41startup. I want to hear about just a
- 1:55:43little bit about your career, actually, but also how you're
- 1:55:45finding the big scary world of startup life as well.
- 1:55:50Thank you. Yes, actually my career kind of
- 1:55:53in a way helped prepare me for start of life.
- 1:55:56I worked at or started at Google back in 2002 under Sheryl
- 1:56:00Sandberg. So we were a small team.
- 1:56:02I think there were maybe 400 people in the company at a time
- 1:56:05and really exciting explosive growth.
- 1:56:08And I got to be a part of that and through my role in that team
- 1:56:13realized how important agencies were to the ecosystem for Google
- 1:56:17and we had no special support. So that prompted my move to New
- 1:56:20York to start the first agency partnerships team.
- 1:56:23And you know, from there I did have a stint on a team that was
- 1:56:27like building a product from scratch within Google.
- 1:56:29So that was again, kind of what helped prep me for where I am
- 1:56:32today. But but did spend about half my
- 1:56:34career working with brands and agencies at Google, from
- 1:56:37performance to media buying to creative.
- 1:56:40Wow. Yeah.
- 1:56:41So kind of ran the gamut. That's awesome.
- 1:56:43And so where was Google at this point?
- 1:56:45They must have been in a totally different place where they are
- 1:56:48now, right? Very, very, very different hyper
- 1:56:50growth stage. The culture was just incredible.
- 1:56:53Learned so much, met so many incredibly smart people, really
- 1:56:57amazing people, and I'm forever grateful for my time there.
- 1:57:00Yeah, I can imagine, if I can imagine you've taken some, some
- 1:57:04lessons into a new business from all that experience is 1 then
- 1:57:07you moved to. Then I moved to Twitter, yes.
- 1:57:09So I I took on a global role working with group MWPP and
- 1:57:14creating strategic relationships there.
- 1:57:16So continued on my learnings from Google into Twitter, which
- 1:57:20was also in a state of growth at the time in terms of like
- 1:57:23expanding video offerings and all of that.
- 1:57:25So it was also a fantastic experience.
- 1:57:28And my last role was actually agency side.
- 1:57:30So I felt a lot of the pain points that Fuse is solving for
- 1:57:34when it comes to understanding. I love data, and one of our pain
- 1:57:39points from an agency perspective is that, you know,
- 1:57:42we have a lot of. People right out of.
- 1:57:44College you don't necessarily you know have that training and
- 1:57:47understanding of what is an insight versus a data point and
- 1:57:51so sometimes the default is to send out data points science and
- 1:57:57the clients are like well I don't what do I do with this
- 1:57:59information so so we. As a human job wasn't.
- 1:58:03It, it's a human's job and it still happens today, even
- 1:58:07though, you know, I think there are tools and, and AI is
- 1:58:09obviously helping supplement that workflow.
- 1:58:12But you know, humans are human and we have our patterns and
- 1:58:15people stick to what's comfortable.
- 1:58:16And so there's that, that innate, that training that you
- 1:58:20need to go through to help, you know, junior employees kind of
- 1:58:24get into that mindset of. I need to.
- 1:58:27Help my clients. What do I do next with my
- 1:58:30marketing so that that's what we're trying to solve for?
- 1:58:33It's, I mean marketing. Isn't.
- 1:58:35That's one of the key learnings that I've taken away throughout
- 1:58:38my career is that you're always at the behest of your customer.
- 1:58:41You have to think about what they want, what they need, what
- 1:58:44their problems are, and I think adtech and martech is all too
- 1:58:49product salesy for that. There's a lot of businesses and
- 1:58:52products that tell you what they do, but So what you know, that's
- 1:58:56that's what is a really interesting range to put over
- 1:58:58the top there. Exactly.
- 1:59:00And I think, you know, dashboards have their place in a
- 1:59:02way, but the the challenge with dashboards is that they're
- 1:59:05static and they don't give you the context of why is this trend
- 1:59:09happening and what is the anomaly that I need to address
- 1:59:12right away and and helping you take that extra step.
- 1:59:15Like they, they just don't have that intelligence.
- 1:59:18Yeah. And so that again, that's kind
- 1:59:19of what we're trying to do. Yeah, and, and I suppose AI is
- 1:59:22a, adds a lot of the, you know, the ability to ask it, the
- 1:59:26ability to load it with an anchor or a bit of context that
- 1:59:31changes everything. And that can happen in a
- 1:59:33self-serve way as well that that the human was doing manually
- 1:59:36before. Exactly.
- 1:59:37Yeah. I mean, AI is a fantastic tool
- 1:59:41in that it helps save you all this time and there's these
- 1:59:44amazing cool things you can do with it.
- 1:59:46But it's also you have to be a little careful in terms of how
- 1:59:49you deploy it because you can't get into trouble, especially if
- 1:59:53you're just kind of blindly uploading things to trap GPT or
- 1:59:58any take your pick of a bunch and you're not like using common
- 2:00:04sense and double checking. You can of course have
- 2:00:07hallucinations, which is good for anybody.
- 2:00:10OK, so I see a problem here, which I think you're going to
- 2:00:13talk about a little bit, which is in one hand, someone can ask
- 2:00:18it anything, and on the other hand it hallucinates.
- 2:00:21This sounds like a challenge to overcome.
- 2:00:25It's a huge challenge. I think I, and I don't quote me
- 2:00:28exactly, but I believe I saw a stat that it's something around
- 2:00:3113% of. Of.
- 2:00:35Responses to prompts are hallucinated marketer and you're
- 2:00:39trying to make budget decisions based on bad data.
- 2:00:43You know you're, you're not necessarily going to make the
- 2:00:46right decisions for your business.
- 2:00:47Yeah, I mean, that's a scary thought in a world where, you
- 2:00:51know, we already sort of we already that the famous saying
- 2:00:55is what we don't know which 50% works.
- 2:00:57And you know, all those funny things that that come up that
- 2:01:01just point to bad planning or bad measurement on on one or
- 2:01:05either end of the spectrum. And yes, for sure, AI overall is
- 2:01:10improving, going to improve productivity and data and
- 2:01:13insights over the course of time.
- 2:01:14But yeah, I can see how it just, it might just amplify a lot of
- 2:01:18rubbish as well. Exactly, you know the whole
- 2:01:20expression, garbage in, garbage out.
- 2:01:22You know, we need to eliminate the garbage altogether.
- 2:01:26And, you know, it starts with the data and it starts with
- 2:01:29making sure the data is accurate and clean and organized in a way
- 2:01:33that allows you to, you know, have that analysis and conduct
- 2:01:37that analysis. And so that's, that's where we
- 2:01:39start with just verifying the data.
- 2:01:41So we, we pull it live directly from the platforms and then have
- 2:01:47this whole verification layer so that you can trust that the data
- 2:01:51is clean and ready and organized and, and, you know, ready for
- 2:01:55that next step of the analysis. And a lot of other tools that
- 2:01:59are especially the ones that are legacy, they're not AI native.
- 2:02:02They're, you know, they're not doing it.
- 2:02:04And therefore, there's a lot of mistrust in in, you know, you
- 2:02:08talk to any marketer, you talk to any CMO, they're skeptical.
- 2:02:11You know, they're like, I know I need to be using AI, you know,
- 2:02:15in, in, you know, more robust way to help me provide better
- 2:02:21outcomes for my business. But how do I do that?
- 2:02:23Where do I start? And the tools that I have used
- 2:02:26don't work. Yeah.
- 2:02:27So there's a lot of mistrust. And that's where we're trying to
- 2:02:30build back that trust with your data tools and data analysis
- 2:02:35tools because we've built in that trust layer.
- 2:02:38It's so like it's a really interesting concept because I
- 2:02:41think that if you're a person, personal user, I've seen time
- 2:02:46and time again that a good looking dashboard and UI always
- 2:02:50is always appealing to someone to just get a quick job done.
- 2:02:53But when there's money on the line, when there's a business
- 2:02:55outcome on the line, there must be so many leaders that don't
- 2:02:59even know their, their data is not trustworthy.
- 2:03:02So is it? So obviously you're trying to
- 2:03:06sell to people that or customers that get the problem, but there
- 2:03:09must be also so many customers that don't realize what they're
- 2:03:12looking at is the problem. Yes, yeah, yeah.
- 2:03:16Conversation. Or is it?
- 2:03:18Or do you find ways of like? Moment like I was actually I was
- 2:03:22like right after CES and former colleague who's the CMO at a
- 2:03:26very large enterprise now she's she just she's like day one I
- 2:03:30met with my team and they presented all these like
- 2:03:34dashboards and she's like, I was just like what what is this?
- 2:03:37What like how do I make sense of this?
- 2:03:39And you know, for her, you know, especially as a brand new person
- 2:03:42walking in, like how, where do you even start?
- 2:03:46Like how do you know which metrics are important, which
- 2:03:49ones are driving business and you know, kind of building a
- 2:03:52strategy from that? Yeah, for sure.
- 2:03:55That, I mean, I that's a really interesting point to make about
- 2:04:00what questions you ask the leader as well.
- 2:04:03And I also wonder, is there a way of you, I guess, getting
- 2:04:07them to stress test the problem themselves with AI?
- 2:04:10So like, ask it the same question three times, see what
- 2:04:13it comes back with. Another actually gave me totally
- 2:04:16different answers. Yeah.
- 2:04:18Yeah, I know. That's exactly right.
- 2:04:19We've had we've had our current clients.
- 2:04:21So we'll start off with the first phone call and they're
- 2:04:23skeptical and they're like, I've tried these things and they're
- 2:04:26inaccurate. And then they try ours or they
- 2:04:28compare it side by side. Even Google's own tool, isn't it
- 2:04:32quite as accurate as you would expect it to be?
- 2:04:36And when you do the side by side comparison or like you said,
- 2:04:40take the same question multiple times and see the consistency
- 2:04:44and responses, that builds the trust.
- 2:04:46That's like the trust layer of no hallucinations when you
- 2:04:50actually verify. And if you want to go back and
- 2:04:53verify it on your own, just to, you know, check it, everyone
- 2:04:56comes back and they're like, wow, like I cannot believe how
- 2:04:59accurate and close it is, but I do.
- 2:05:02I'm going to touch on the your point about what questions to
- 2:05:04ask, because that's, that's actually what we're hearing a
- 2:05:08lot from some of our clients and potential clients where they're,
- 2:05:12I don't even know where to start with the questions.
- 2:05:15So we will. Before.
- 2:05:17Well, it depends on the expertise levels.
- 2:05:19So if we're talking to an expert marketer, analyst, they already
- 2:05:22know what. Questions to ask.
- 2:05:23They're they, they have no problem just jumping right in
- 2:05:27and asking questions. But if it's like ACEO or, you
- 2:05:32know, maybe a startup founder who doesn't have a marketing
- 2:05:35background, those are the people who need a little bit extra
- 2:05:38guidance when it comes to OK, like, I know, I, I know this is
- 2:05:42important, but like, where do I start?
- 2:05:44What are the right questions and what's important?
- 2:05:49So, so we, we have some preset prompts.
- 2:05:52We're also providing some more proactive, you know, health
- 2:05:55checks or, you know, score cards that help you identify it more
- 2:06:00quickly. What are the key critical areas
- 2:06:02you need to address in your marketing?
- 2:06:04So we're working on ways to be more proactive.
- 2:06:07But if you already have that marketing expertise, people love
- 2:06:11our products. I can imagine.
- 2:06:13Yeah, as an ex Google as well, it must be cathartic, not in a
- 2:06:17like a in an US versus STEM way, but just in a wow.
- 2:06:20It's so challenging that even their technology finds it a
- 2:06:24struggle. So it must, it might, you must
- 2:06:26enjoy that process as well of fixing that issue.
- 2:06:29Yeah, it is very rewarding. I mean, it's harder than it
- 2:06:32sounds. I mean, I think there's a lot of
- 2:06:34like, oh, I can just go and, you know, buy something in, you
- 2:06:38know, minutes. And it's like, wow, it's not
- 2:06:40quite that easy. There's still that contextual
- 2:06:43layer and that proprietary keys, that layer that we've built in
- 2:06:48specific for marketing. So like we're vertical in terms
- 2:06:51of, you know, we have that expertise in the performance
- 2:06:54marketing area. Yeah.
- 2:06:55And I guess, I don't know, this is maybe just me with a
- 2:07:00harebrained idea, but there, I guess it would be really
- 2:07:04interesting to see if you could quantify the scale of the error
- 2:07:08when you're using the wrong data in, in terms of like, because
- 2:07:12that's when it becomes really real to someone who is, you
- 2:07:15know, just got their interns buy code dashboard and, and
- 2:07:19connected a brittle API to, you know, their meta ads or
- 2:07:22whatever. And, and they're going, OK, I'm
- 2:07:24getting what I want. And, and all of a sudden they
- 2:07:28see something that's like, well, firstly, you're not.
- 2:07:30And secondly, the scale of the problem has, you know, the
- 2:07:35consequence as well. Yeah.
- 2:07:36And that consequence sometimes is we're seeing that Meta can
- 2:07:40shut down your ad account. So in terms of quantifying, OK,
- 2:07:43what does it mean to your business if Meta suddenly shuts
- 2:07:47down your advertising account because you didn't have the
- 2:07:50right connections into their platform and therefore it was an
- 2:07:53unauthorized use of or unauthorized use of of their,
- 2:07:57you know, data. So, you know, making sure you're
- 2:08:00working with a tool that has that approved API access, then
- 2:08:05you don't have to worry about that happening.
- 2:08:06Yeah. Look, I think, I think that's
- 2:08:08actually a really understated future challenge for the next
- 2:08:12couple of years because as the world gets more API connected
- 2:08:17and as the world or as you know, our industry lessens in human
- 2:08:21resource, I think that there's going to be so many attempts at
- 2:08:25just plugging that into that and that into that.
- 2:08:27And and so I think that that narrative and that capability of
- 2:08:32having strong connections into the data so many all systems
- 2:08:38running around now. That's exactly right.
- 2:08:40Yes, that's exactly right. And and it's also, you know,
- 2:08:43just it's like that continuous loop.
- 2:08:45So you make a decision, you implement it, which by the way,
- 2:08:49we are working towards that we call it fuse execute where OK,
- 2:08:52you've got the recommendation now you just click a button, you
- 2:08:55approve it and you don't even need to log into your account to
- 2:08:58make the change. So that that's coming.
- 2:09:00But but yeah, I mean, in terms of of that loop, you've executed
- 2:09:07it. Now that data goes back into
- 2:09:08FUSE and then it's it's going to take that into account for the
- 2:09:12next analysis. And, you know, making sure that
- 2:09:15in real time, like your budgets are optimally deployed across
- 2:09:19all the channels because things change, things happen.
- 2:09:22There's, you know, seasonality, there's who knows what's going
- 2:09:25to happen in the world. And so to be able to have that
- 2:09:28agility is really important as a marketer.
- 2:09:31Absolutely and and Jill, tell me, what was the I guess did you
- 2:09:35the journey towards starting fuse?
- 2:09:38Was it born of an aha moment? Was it born of a pain that just
- 2:09:42was happening there? It's just this just needs to be
- 2:09:45fixed. What was yours down?
- 2:09:46Here, give credit to my Co founder for that.
- 2:09:50This, this was kind of the original actual concept of FUSE
- 2:09:55was to combine both the automation of like creating
- 2:09:58slide decks and presentations to present either to clients or to
- 2:10:01your internal teams or your CEO, CMO or CFO.
- 2:10:05And, and you know that, that analysis and ability to
- 2:10:10holistically analyze across all data sets.
- 2:10:13So, so it was experiencing that pain point ourselves at, at
- 2:10:16different agencies, but the original idea was to automate
- 2:10:19that visualization so you can communicate to your stakeholders
- 2:10:24what those recommendations are and what the insights are.
- 2:10:26We are now actually, we do have that visualization capability
- 2:10:30with FUSE. So yes, to answer your question
- 2:10:33was born from a pain point that we've experienced our entire
- 2:10:36career. But you know, from for me, I'm
- 2:10:39terrible at creating PowerPoint and designing slides and all
- 2:10:43that. And now thankfully we have AI to
- 2:10:45help us with that. And so kind of combining those
- 2:10:47two together, like we do have the ability to create
- 2:10:50visualizations using, you know, Google's data banana for the
- 2:10:54technology piece to help communicate those, you know,
- 2:10:57what is that data story that needs to be told to the decision
- 2:11:00makers to approve those decisions?
- 2:11:04Because that's what really what it comes down.
- 2:11:06That's kind of the whole point with creating all these slides
- 2:11:08is to to communicate. Here's the analysis, Here's the
- 2:11:12why. Here's the story.
- 2:11:13Here's what we need to do. Next.
- 2:11:15Yeah. And and that sort of comes full
- 2:11:17circle to the So what you know what, what is.
- 2:11:20The point? Of the the thing for that
- 2:11:23customer of that marketer. Yeah, yeah, that's exactly,
- 2:11:26Yeah. We have the the what?
- 2:11:28The what, which is the data point, the.
- 2:11:30So what does this mean? Why is this important, which is,
- 2:11:33you know, our secret sauce and then the recommendations of now
- 2:11:37what? What do I do next with my
- 2:11:38marketing? And that's kind of why we've
- 2:11:41created that framework to, again, help marketers make
- 2:11:44better, better decisions. Well, that's also, it's a 'cause
- 2:11:47that I think, you know, the idea of making data more believable
- 2:11:51and reliable in an AI world is something you're going to fight
- 2:11:54against tide going that way. And I think it's a really good
- 2:11:58journey to go on and it's a really good outcome.
- 2:12:00So thank you so much. Are you just, are you here for
- 2:12:03the rest of the week? Are you?
- 2:12:04Here yeah, I'm actually here for the week.
- 2:12:06I'm just staying through for F1. So if anybody wants to get in
- 2:12:10touch, you know, please reach out either LinkedIn, Jill
- 2:12:13Randall or Jill at fuse dot IS or yeah, we might come find me.
- 2:12:19I'll be probably hanging around the fountain glue.
- 2:12:22Awesome. Well, thank you so much Jill for
- 2:12:24your expertise and your. This is great.
- 2:12:27Thank you so much. Thank you.
- 2:12:30Here we're a possible 2026. This is not just getting towards
- 2:12:34the end of date #2 and I hear with Doug Lauritano from Tuple.
- 2:12:38Great to see you as always. Good to see you.
- 2:12:41So get a lot to talk about today.
- 2:12:43You launched a new company recently, didn't you?
- 2:12:45I did 60 days and more or less. All right.
- 2:12:48And you know, just public service announcement.
- 2:12:50He is no longer the ad tech pizza guy.
- 2:12:53I'm taking a break at least probably, maybe permanent but
- 2:12:57not sure. All right.
- 2:12:59We'll see where it goes. But love to hear about Tuple and
- 2:13:02like a little about the company, what you're trying to solve it.
- 2:13:04But maybe before we jump into that, maybe a little bit about
- 2:13:07yourself. Sure.
- 2:13:09Laura Tano, born and raised in Brooklyn, NY.
- 2:13:11Guy all my life. What neighborhood, by the way,
- 2:13:15did you Brooklyn? Bay Ridge.
- 2:13:16Nice. Yeah.
- 2:13:17Bay Ridge in Brooklyn married a Long Island girl 18 years ago
- 2:13:22and she dragged me out there kicking and screaming, and I've
- 2:13:25been there ever since. What's how Long Island feel,
- 2:13:28mind? Cold Spring Harbor.
- 2:13:29OK. Yeah, North Shore.
- 2:13:31It's a nice place to raise a family and it's it's not quite
- 2:13:36New York City, so. It's that amazing whaling museum
- 2:13:40there, isn't there? Do you have a whaling museum?
- 2:13:42That's that's a heck of a pole there.
- 2:13:44And it's really cool. I went there I think in in fifth
- 2:13:46grade on a good field trip. And it was, that was pretty
- 2:13:49cool. So a lot of history, I mean,
- 2:13:51it's like goes back. I think it's founded like 1680s
- 2:13:55or something like that. It goes with that, right?
- 2:13:57I mean. You know much better than I.
- 2:13:59You did research or you just know a lot about Long Island?
- 2:14:03I spent 10 years Long Island growing up.
- 2:14:05I had a teacher who was just really into that.
- 2:14:07So yeah, it's pretty, pretty part of the world.
- 2:14:08So. But yeah, so like moved out to
- 2:14:12Long Island and I you were you're so before found in this
- 2:14:15company, you were CEO of another firm, right?
- 2:14:17Yeah. So five years ago I was hired to
- 2:14:20start the advertising business for Civic Science, which at the
- 2:14:24time was a research company, had some access to pretty
- 2:14:27interesting first party data. Obviously that data all around
- 2:14:31declared consumer intent could be very valuable in advertising.
- 2:14:34So they hired me to start up that business and recently
- 2:14:37recently left after the company more or less pivoted into being
- 2:14:40an advertising company in total. So as the founder of Civic
- 2:14:43Science says, I left the company much better than I.
- 2:14:47Then I came into it and. What you did became the
- 2:14:49strategy, so yes. Yeah.
- 2:14:51And it's really, it's actually really interesting approach.
- 2:14:54They have voluntary holes in and around premium publisher
- 2:14:59content. You get some really interesting
- 2:15:00signal users share who they are and what they like to do, how
- 2:15:04they like to travel, what they like to buy, where they like to
- 2:15:07eat, and all of that data can be compiled into profiles and can
- 2:15:11build really interesting custom segments for advertisers.
- 2:15:14So it's a pretty interesting business and pretty proud of
- 2:15:17what we did there. Nice.
- 2:15:19So two months in, so this, but this was stealth mode for a
- 2:15:22while I'm guessing right? Yeah.
- 2:15:23So two months in, officially launched a couple two months
- 2:15:27ago, been building the platform for about a year.
- 2:15:29Part of my engagement with Civic Science was over the last year
- 2:15:34basically a very long transition and what got us to what really
- 2:15:39drove the approach and how we're launching Tuple was actually a
- 2:15:44significant problem that we found while building Civic
- 2:15:46Science, which Civic Science is really interesting data set.
- 2:15:49Instead of plugging into every single DSP to give advertisers
- 2:15:54access to it, we were able to leverage curation working
- 2:15:57directly with some SSPS to create deal IDs around these
- 2:16:00custom segments. Sure.
- 2:16:02What we were seeing from our buyers was that we would send
- 2:16:06them 10 million impressions a day in this in this custom
- 2:16:09package and in their DSPC they would only see 4 million
- 2:16:13impressions. So at first we thought this was
- 2:16:16like a, you know, some sort of a glitch started working closely
- 2:16:19with the SSPS to try to figure that out.
- 2:16:22Was seeing this across all our customers that we were working
- 2:16:26with. So it's consistent.
- 2:16:27Consistent and what we ended up realizing and this is tons of
- 2:16:32conversations with different folks across the industry at
- 2:16:35platforms, even at publishers. What we realized is that the
- 2:16:39traffic shaping and throttling GPS limits, all these things
- 2:16:43that platforms both on the sell and buy side were doing in order
- 2:16:47to to minimize or or to reduce their own infrastructure costs
- 2:16:53was basically leaving really good impressions on the cutting
- 2:16:56room floor. Like I said, Civic Science was
- 2:17:00building these custom segments. The users we were building them
- 2:17:03around where you know, had had persistent ID's.
- 2:17:07This wasn't, you know, like, you know, some uncooked or you know,
- 2:17:12some this. Wasn't like cookie list idealist
- 2:17:14traffic these these were legit. This was this was not only legit
- 2:17:18but very valuable impressions. So, and like I said, that was
- 2:17:23that was about two years ago. So because that ad just DSPS
- 2:17:26weren't aware, there's no bidding going on, ads weren't
- 2:17:29served up, it could basically. DSP, so buyers in their DSPC
- 2:17:33were not aware that there were another six million impressions
- 2:17:37that they could be bidding because basically it's a huge
- 2:17:40number and and that 40% number was pretty like that.
- 2:17:44That was around the median of what we were seeing when when we
- 2:17:47saw just cosrepancy. Limits and like as you said,
- 2:17:50like the potential, like the cost because of the number of
- 2:17:54bits going through. Because of that, the, so that,
- 2:17:57so if you think about the last decade plus of programmatic with
- 2:18:03the, with header bidding really becoming front and center and,
- 2:18:06and just everybody working through the header, there's just
- 2:18:09so much duplication of inventory that ADSP needs to, they need to
- 2:18:14be able to, to run their business and not just be this
- 2:18:17massive infrastructure play. While at at the same time they
- 2:18:22have certain clients that want to want to achieve supply path
- 2:18:28optimization. But I actually view it as many
- 2:18:32of them want supply path optionality and they want to be
- 2:18:37able to buy from many different supply sources, whether it's
- 2:18:41from a business relationship or some other preference.
- 2:18:45So many of these, many of the like the legacy DSPS aren't able
- 2:18:49to say, hey, I'm only going to work with this smaller set of
- 2:18:53SSPS because they already have clients that want this, want to
- 2:18:58be able to operate. Brands they're working with,
- 2:19:01they all header bidding, yeah. So I mean like break your number
- 2:19:05down which is not an option for that.
- 2:19:06Right. Yeah, so.
- 2:19:08There's a tons of traffic and it's traffic that's it might be
- 2:19:11even be the same bit coming through like duplicated sent
- 2:19:14across multiple SSPS that they come to hit them at the same
- 2:19:16time, right? So basically what I think on
- 2:19:19average what a legacy DSP sees or what a buyer in a legacy DSP
- 2:19:24sees is 40% or so of unique impressions duplicated dozens of
- 2:19:30times. I've heard it might be up to
- 2:19:33only 20% unique some in some instances like duplication.
- 2:19:37Maybe that's a little. I'm saying I'm saying 40% of the
- 2:19:40total universe of unique and fresh so, but then that 40% is
- 2:19:47duplicated dozens of times. That's wild.
- 2:19:49So if you if. You want to say like, what is
- 2:19:51actually like unique impressions?
- 2:19:53It's maybe even maybe. Less than that, right, Maybe as
- 2:19:56low as 15% or 10%? Right.
- 2:19:58So if you think about it, if you want to it.
- 2:20:01So you're saying this culling process is kind of normal again,
- 2:20:03I have to. Say, well, yes and no.
- 2:20:07Is there a better way, which is, which is a good segue into how
- 2:20:11we how we think about what that better way could be with, with
- 2:20:14tuple. So tuple is you're building an
- 2:20:17enterprise level DSP and Omni channel ESP.
- 2:20:20But we're building it on what we believe is a better foundation.
- 2:20:23And so if you think about, you know, everybody in programmatic
- 2:20:26wants to talk about competing with the walled garden Google or
- 2:20:29or Meta when there's an impression on one, one of their
- 2:20:34one of their properties. Owned and operated in one of
- 2:20:37their properties, one. Of their properties on on
- 2:20:39YouTube or on Instagram or what have you like a buyer can access
- 2:20:44that impression right. A buyer also is not like there
- 2:20:48there, there isn't dozens upon dozens of duplicates of that
- 2:20:52impression. For a system to have to call
- 2:20:54through to be able to recognize that impression, well, it's.
- 2:20:59Their inventory, it's their DSP, it's their impression, yes.
- 2:21:02So they control it A. 100% I think about it as like a one to
- 2:21:04one, right? Well, we should try to try to
- 2:21:10copy that, right? Like if you want it, like if you
- 2:21:12want to start a DSPD today, which I started 60 days ago.
- 2:21:17But if you wanted to start that, how would you build it?
- 2:21:21So that's what. The open web like how would you
- 2:21:23open? How would you do that?
- 2:21:25So what we're basically doing is we're we are working with a
- 2:21:29small number of supply partners those supply.
- 2:21:32Partners are SSPS. SSPS.
- 2:21:35So we're working on a small number of SSPS.
- 2:21:38We are working with SSPS that are that can enable us to add
- 2:21:42logic by being within a container on their side.
- 2:21:46So we Co locate with the SSP so that we're able to de duplicate
- 2:21:50and we're able to ensure that the entire bit stream from each
- 2:21:55SSB we can see, but we don't need to push into our platform.
- 2:22:00Because everybody might everything.
- 2:22:02Yes, all. Everything from bits.
- 2:22:04Yeah, a lot of information. We can see it, right?
- 2:22:06We can see it, but it doesn't need to move from their platform
- 2:22:10to acts that looks. Like a problem for a lot of DSPS
- 2:22:12that there's no way they can handle the volume of contextual
- 2:22:15and identity. Right, right.
- 2:22:17So we're able to limit the infrastructure.
- 2:22:20Costs of that. Because we're only working with
- 2:22:22two to three SSPS and because there's there's no push from
- 2:22:25their side to our side unless it's it's something that that is
- 2:22:29necessary not to. And so that is going to create
- 2:22:33as close to a one to one walled garden like access to inventory
- 2:22:39as there has been in programmatic.
- 2:22:41I was just different for like let's say what the trade desk is
- 2:22:43doing with open path. How can you compare the 2?
- 2:22:46So I think open path, forget all of the, you know, controversy
- 2:22:53around different things regarding it and and having
- 2:22:55different parts of the industry A1 to 1, they're creating a one
- 2:23:00to one, right. I think that's the right way.
- 2:23:02The now part of the issue that that and some of the issues I
- 2:23:08think behind some of that controversy is that they have
- 2:23:11these long term, long time partners in the SSPS and they're
- 2:23:18basically funneling money in One Direction versus the other and
- 2:23:22they're also taking control away from their buyers control is
- 2:23:26another like kind of tentpole. Control as a decisioning or
- 2:23:29control as in. Decisioning, but like like
- 2:23:32basically like they're deciding where the where the money goes
- 2:23:36when I know that some buyers don't necessarily want that, but
- 2:23:40I think but on on as a concept, open path I think is the right
- 2:23:45way now with tuple building out direct relationships with every
- 2:23:52publisher. It's not in the cards, not in.
- 2:23:56The millions of publishers? Sure.
- 2:23:58Right. So we can't pursue that.
- 2:24:01So I think the approach that we're taking is we're leveraging
- 2:24:04some pretty unique things that are going on in the industry
- 2:24:08today. So I mentioned the
- 2:24:09containerization before. One thing that we're using is
- 2:24:12the agentic real time framework from that it's kind of a
- 2:24:16standardized the way that different players can work
- 2:24:19within a container. You know, most of the
- 2:24:21containerization partners to the actual curators are custom
- 2:24:24Algodo folks and DSPS. But I think the like having all
- 2:24:31of the logic and the decision and being being able to take
- 2:24:34advantage of the full, the timing of the round trip of an
- 2:24:38impression just puts us in a pretty interesting spot.
- 2:24:41So is the decision might like not happening in the DSP, it's
- 2:24:43actually happening in this containerized environment or?
- 2:24:46So we we have some of our decisioning happen happens in
- 2:24:50the containerized environment. Now the reason why all of it is
- 2:24:54not happening there or more of it is because you do have a
- 2:24:56limited amount of time. So it's usually.
- 2:24:58Alone, if that gets right, yeah. A little 5 milliseconds, which
- 2:25:01is. You know, that's pretty fast.
- 2:25:03Yeah, pretty fast. But basically, when you have
- 2:25:06that additional 5 milliseconds to make a decision that that can
- 2:25:11greatly affect your own infrastructure costs and the
- 2:25:14ways. And you know, I'm describing all
- 2:25:16this. What I think is more unique
- 2:25:19about it than necessarily the technology is the kind of
- 2:25:22partnership and cooperation between the SSP and DSPI think
- 2:25:27part of the reason we've gotten to the place we have with
- 2:25:30programmatic is because the SSP and the DSP relationship has
- 2:25:36devolved in into this, you know, really contentious.
- 2:25:40Level audition shit. It's a problem, yeah.
- 2:25:43And it's, you know, not only, not only are there, you know,
- 2:25:46you know, swipes taken on conference stages and in the
- 2:25:49press, but like, you know, an SSP doing bid stuffing where
- 2:25:54they're sending, you know, multiples of the same bid to A,
- 2:25:58to a DSP because that DSP is, you know, doesn't respond the
- 2:26:03first time. Or if they do like like so the
- 2:26:06SSP does things in the way that they send things to a DSP that
- 2:26:11the DSP would not approve of the and the DSP does things with the
- 2:26:15SSP. They'll do other things like
- 2:26:17with cookies or with. All sorts of different things
- 2:26:20and. Or.
- 2:26:21Adjusting Jake rates ID bridge. Yeah like who's?
- 2:26:23Shady behavior for sure. I could, I could be out on this
- 2:26:26balcony all day talking about stuff like that.
- 2:26:29It's drives me nuts but. I think there's no.
- 2:26:31Shortage of questionable behavior, that's for sure.
- 2:26:33But there's a better way to do things.
- 2:26:35Well, like what we are, what we aspire to do is to have a much
- 2:26:40better relationship with the SSP because the way that we see it
- 2:26:45is that. Yes, the SSP needs to.
- 2:26:48Represent and and work for the publisher in to some extent, but
- 2:26:52the SSP is also beholden to to the the ultimate, you know,
- 2:26:58funder of all this, which is the advertiser, right?
- 2:27:01They have the dollars and the SSP has a relationship with the
- 2:27:04advertiser often time. So if we if we are trying to
- 2:27:08make the same cohorts happy, right, we're trying to fund, you
- 2:27:12know fund content creation and and the open.
- 2:27:15Advertisers keep them going right?
- 2:27:17Keep monetization high? Sure.
- 2:27:19We do have the same, you know, intentions.
- 2:27:21Then we should be able to collaborate and work and work
- 2:27:24closer together and not get to a point where either one of us are
- 2:27:27doing things that are not in each other's best interests.
- 2:27:33How would you compare what you're doing to some of the SPL
- 2:27:36or supply path optimization work that's been going on?
- 2:27:38I mean we just similar. So that the thing about supply
- 2:27:45path optimization is that I feel like a lot of a lot of that is
- 2:27:51when a third party outside of an SSP or DSP tries to find the
- 2:27:57best way. Now, I'm not, I'm not saying
- 2:27:59that those third parties can't be can't be right and they can't
- 2:28:02be good at what they do, but this is trying to solve for a
- 2:28:06problem that was needed like and and just shouldn't charges like
- 2:28:15like there, it's an unnecessary problem.
- 2:28:17You know if a if an if an SSP is working with a bunch of
- 2:28:22resellers which creates creates a.
- 2:28:25Mess of certification and I didn't.
- 2:28:27Lack of transparency like that? Sure.
- 2:28:30But at the same time, I would also an interesting kind of
- 2:28:33ingredient to all this is that part of the reason why SSPS may
- 2:28:38may have, you know, tolerated more resellers, some of the ones
- 2:28:44that may be more aggressive kind of duplication themselves.
- 2:28:48All these different things is because most legacy DSP
- 2:28:52platforms are designed to send a higher bid response rate to to
- 2:28:59more GPS. So with that basically means.
- 2:29:02The incentivize just to thank you, the higher volume, just the
- 2:29:05way they're more GPS. So the same way the same way
- 2:29:09that a a publisher makes more money when they add one more
- 2:29:13header bidding a SSP makes more money when they send one more
- 2:29:17impression typically. I mean, yes, but ultimately this
- 2:29:21is a law of diminishing returns though, right?
- 2:29:23So publishers keep adding more header bidders, the key adding
- 2:29:25more SSPS that at some point, OK, like history has shown,
- 2:29:29maybe that's not the best strategy, so.
- 2:29:31Well, it depends on what you want the outcome of that
- 2:29:35strategy to be like. Do you want your your as a
- 2:29:38publisher? Do you want your website to take
- 2:29:39forever to load and to have a terrible consumer experience?
- 2:29:42Just the outcomes we do that, sure, yeah.
- 2:29:44So like, yes, there's, there's very, there's, there's bad
- 2:29:48outcomes, but at the same time, a lot of a lot of publishers are
- 2:29:53under not only like quarterly revenue pressure, somewhere on
- 2:29:57many are under daily revenue pressure.
- 2:29:59I mean, it's daily now with especially with the like the
- 2:30:02declines we're seeing in referral traffic.
- 2:30:04I mean, it's twenty 3040%. I'm dinner with the big
- 2:30:08publisher tonight and out there they've been telling me things
- 2:30:10like this. So it's getting bad and people
- 2:30:12realize there's a big revenue stream for pubs 100%.
- 2:30:15And, you know, having seen that get cut down by maybe up to 50%
- 2:30:19and with no end to sight, right, with the rise of like of AI
- 2:30:23search. So it's going to be yeah.
- 2:30:25So I think there's definitely need for some solution like
- 2:30:29this. So well, it's exciting.
- 2:30:30So two months in, are you selling, maybe it's too early to
- 2:30:35say, we were able to disclose, Are you selling direct to
- 2:30:37brands, partnering with agencies?
- 2:30:39What's your go to market so far? Yeah.
- 2:30:41So right now our pilot partners, two of them are curators.
- 2:30:46So curators. So we're partnering together to
- 2:30:49be able to allow their their clients to be able to buy more
- 2:30:53of what of what the the curator is is.
- 2:30:56Packaging so the package according to your ID and then
- 2:30:59OK. Yeah, 1 is a direct brand and
- 2:31:02one is an independent agency. So it so it's a little bit
- 2:31:05across the board. I don't, I don't foresee us
- 2:31:08being a, you know, a focus for a holding company to be honest.
- 2:31:14You know, I know that they've lots of different needs that I
- 2:31:18think we're not going to focus on and some.
- 2:31:21More than a pretty robust like immature curation capability
- 2:31:24stuff. Yeah, I think look, the, what I
- 2:31:27think is unique about what we're doing is I think Arie Arrow
- 2:31:33wrote in one of his newsletters recently that ADSP needs to
- 2:31:37either have unique inventory or unique data, right.
- 2:31:40He was basically saying that without unique inventory or
- 2:31:43unique data. There's we'll start.
- 2:31:47You know, Emily and I spoke a lot before I made the final
- 2:31:50decision to to start Tuple. We, we have unique access to
- 2:31:55inventory. We, we give the buyer unique
- 2:31:58access to. Now should that, you know, is
- 2:32:01that inventory something that other platforms should be able
- 2:32:03to offer access to? Yes, I I truly believe that.
- 2:32:06Like with your approach, like you have it, you have it now
- 2:32:09like his. People could copy this approach,
- 2:32:11but your approach is giving you access to the.
- 2:32:14Right. And again, like we, we our
- 2:32:16intention is to is like we want to be an enterprise level DSP
- 2:32:20and Omni channel DSP, not just the DSP that is trying to buy
- 2:32:25this this other half of impressions.
- 2:32:28But he felt like the foundation of what we're building on needed
- 2:32:33to be really strong and shouldn't be shouldn't have the
- 2:32:36same issues that the the industry as a whole.
- 2:32:40The the. Ecosystem as a whole has ranked.
- 2:32:42I bet building a DSP was not easy.
- 2:32:45You know, it, it, it's, it's there's a lot of.
- 2:32:49Latency requirements and with I mean, it's you know, I think
- 2:32:54people don't realize because all tech is hard, but I, I think
- 2:32:57DSPS, you know, from what I can, I've never built one, but I'm
- 2:33:01guessing it's a difficult thing to build.
- 2:33:03Well, look, I'm not a developer, so technically I didn't build
- 2:33:07it, so not with my own hands, but I have really, really smart
- 2:33:11folks that I, we're, we're, we're building something that's
- 2:33:16going to continually improve over time.
- 2:33:18I can't, I can't sit here and pretend like, you know, two
- 2:33:21months in that we have, we have the absolute best DSP that's in
- 2:33:26the market. But I think our initial approach
- 2:33:28does set us apart. And I think over time that one
- 2:33:31to one access that we have will allow for performance that is
- 2:33:37much closer to the to the walled gardens than it than it is to
- 2:33:40typical programmatic. You wonder like in like if some
- 2:33:45of the functionality has been in DSPS migrates out and I think
- 2:33:48that's already happening, right? Like assuming that continues
- 2:33:51like end of the day, like does it become easier to make DSPS?
- 2:33:55I think in theory, yes, but I think it also becomes harder to
- 2:33:58run a DSP business model. So like, you know what I'm
- 2:34:02curious what the future role in terms of like it's an
- 2:34:05interesting approach because this is a really fresh take on
- 2:34:07it, right? But this fresh take is like
- 2:34:09actually going into supply and rewiring the way people work and
- 2:34:14having a fresh way of identifying what you're saying
- 2:34:17is just some, like inventory that was just sitting there that
- 2:34:19no one even, yeah, was able to tap into.
- 2:34:21And, and I think, you know, I give a lot of credit to a lot of
- 2:34:25to be SSPS. Who?
- 2:34:28You know, five years ago or so, like.
- 2:34:32They wouldn't have done this, would they?
- 2:34:33Well, no, I, I also, they wouldn't have done it maybe.
- 2:34:36But also I think a lot of the a lot of the trade press was
- 2:34:40talking about how the SSP is dead right like and, and you
- 2:34:43know, between curation, the sell side decisioning and
- 2:34:47containerization, like really allowing third party to, to get
- 2:34:52closer to the impression to start making decisions.
- 2:34:56Breathe. Is breathe new life another
- 2:34:57business model? There's no doubt, right?
- 2:34:59Yeah. I mean, you're right.
- 2:34:59Five years ago, people were, like many people were, were
- 2:35:01questioning if SSPS even exist, right?
- 2:35:04Yeah. And and you know, and now people
- 2:35:07question about DSPS, right, which look, I think there's a
- 2:35:12healthy programmatic ecosystem has the two different entities.
- 2:35:16Again, we think that there should be a lot better
- 2:35:18coordination and there shouldn't be this angst between both
- 2:35:23sides. You know, I would, I'd be happy
- 2:35:25if that that spread and you know, you know, old, old enemies
- 2:35:32between became new friends. Yeah.
- 2:35:34I'm not. I'm not.
- 2:35:35I'm not too confident that will happen.
- 2:35:37There are no permanent enemies, only permanent interests, right?
- 2:35:41That's fair. But I, but I think, you know,
- 2:35:44that there's this kind of collaboration, like a lot of us
- 2:35:48want to talk about, you know, all the dollars going to the
- 2:35:51walled gardens and how programmatic we as like a whole,
- 2:35:56we need, we need to, you know, we need to be better, right?
- 2:36:01I think, I think unless we're all working together and, and
- 2:36:05both sides of the equation are really working together, that's,
- 2:36:08that's just not going to happen. So, you know, I think we, we
- 2:36:12want, we want to be and we want to drive that and we think that
- 2:36:15it's going to be an advantage for us.
- 2:36:17And you know, I'm a, I'm a relationship guy.
- 2:36:19That's been my career partnership.
- 2:36:21There you are. So I'm hoping that I'm, I, you
- 2:36:23know, I can, I can help drive that from a people perspective
- 2:36:29and you know, from a technology perspective, I think all the,
- 2:36:33it's all there to happen and I'm pretty excited about where
- 2:36:36things are going. Well, completely agree, you're a
- 2:36:40great relationship guy has been it's been cool building a
- 2:36:42relationship with you and hanging out, whether it's over
- 2:36:44pizzas or over over this site. I love to maybe check in in a
- 2:36:47couple months, see where you are and you know, on the regular
- 2:36:50signal and noise. We do as you, as you know,
- 2:36:52longer, longer form hour, even 90 minutes.
- 2:36:55So I'd love to get you on and when you're ready and you've got
- 2:36:58some, maybe some more interesting news to share to
- 2:36:59talk more about about what you're doing.
- 2:37:01But thanks a lot, Doug. It's cool to see you.
- 2:37:03Thanks for having me. All right, Rio.
- 2:37:06Signal and noise were possible. We're almost at the end of the
- 2:37:10day too. And I'm here with a special
- 2:37:11guest, Nish Gore, Chief Dad Officer at Zeta Global.
- 2:37:16See you my friend. Thanks for having me, yeah.
- 2:37:17I thought you made it yes. I was really stoked to get you
- 2:37:20here, you know, wanted to spend some time with you.
- 2:37:25Your team was great by the way, in terms of getting to set up,
- 2:37:27but wanted to hear a little bit about you and I, I think Zeta is
- 2:37:31a really cool like story that a lot of people heard of but don't
- 2:37:34really know enough about you. So I know our listeners are
- 2:37:39gonna be really excited to learn more.
- 2:37:41So maybe we could start out by just tell us a little bit about
- 2:37:44yourself. Sure.
- 2:37:45My name is Nish Gore. I'm the Chief Data Officer at
- 2:37:48Zeta. I sold my company alongside our
- 2:37:50CTO to Zeta about nine years ago and that platform became what is
- 2:37:55today Modern Zeta. Prior to that, I'm a serial
- 2:37:58entrepreneur. I've had several companies,
- 2:38:00predominantly in the martech space, but also done a lot of
- 2:38:02fun offline businesses. I started a coffee roastering
- 2:38:05company. Owned.
- 2:38:07A couple bars in San Francisco for about 10 years, so did.
- 2:38:10A loan to bars or. No, I sold the bars, you know,
- 2:38:12when I got to about 32, I think I had enough of the late nights
- 2:38:16and. Bunia Bar is a late night
- 2:38:18business. Is a late night business and
- 2:38:19it's really fun. I was doing it with my some of
- 2:38:21my close friends so but always been very close to technology
- 2:38:25engineer from Cornell historically so and that's been
- 2:38:29something that's propelled me through my career.
- 2:38:31The company. That you sold the Zeta?
- 2:38:33What type of firm was it was the core?
- 2:38:35Tech, yeah. So the core technology was
- 2:38:37actually machine learning, big data pipeline or
- 2:38:40recommendations. So we were using it in a
- 2:38:43marketing capacity for content recommendations and it was
- 2:38:46actually doing things like automating emails and automating
- 2:38:49media. This is back in 2017 when we
- 2:38:52sold. We started working on it a few
- 2:38:53years before that. So at that time, we were
- 2:38:56basically one of the front runners of machine learning for
- 2:39:00marketing and media that existed in the ecosystem.
- 2:39:02When I say that today you're like, well, everyone's doing a
- 2:39:05piece of that, but back then. People were not.
- 2:39:07You're right. Yes.
- 2:39:07So when Data acquired, our platform quickly realized that
- 2:39:12we had pointed our platform at the wrong set of customers
- 2:39:15having the wrong problem. But the opportunity was really,
- 2:39:18really massive. What customers were you pointing
- 2:39:20it out? What are you?
- 2:39:21Trying to solve, yeah, we were really pointed at the publishing
- 2:39:23community and automating content recommendations across e-mail
- 2:39:28and other channels that they run.
- 2:39:29As opposed to like journey orchestration?
- 2:39:31Yeah, there was journeys, but it more so you know, if you if you
- 2:39:34look if you look at the money and marketing and marketing
- 2:39:36technology, you got to be focusing on retailer, you got to
- 2:39:39go to insurance, you got to go to telecom, you got to go to
- 2:39:42auto throw the money. This is no money.
- 2:39:44Yeah, so. It's not on the publisher side,
- 2:39:46unfortunately. Unfortunately, they're being
- 2:39:47squeezed again, but it's not on the publisher side.
- 2:39:51But I will say that even then the technology also needed to be
- 2:39:56upgraded to become more enterprise grade.
- 2:39:58We're targeting mid market Zeta soundly plays in enterprise we
- 2:40:04have. 51% of the Fortune 100 use us in a very meaningful way and
- 2:40:0824% of the Fortune 500 now. So I.
- 2:40:11Think that's interesting and we could probably circle back on
- 2:40:13that about like how like Zeta has such like a really
- 2:40:16impressive market share among among a lot of big companies and
- 2:40:20people may not even know that you're there.
- 2:40:21But like before we go to that like the decisioning journey
- 2:40:25orchestration, you were early. So I remember around this time,
- 2:40:28maybe around 20/18/2020, there was a bunch of, and I know Adobe
- 2:40:34ended up building theirs like you were, you were early to
- 2:40:37market. Like what?
- 2:40:38Like what? And granted you said you focused
- 2:40:41on what you felt was the wrong type of customer, but like what?
- 2:40:44Why did? Like why?
- 2:40:45What made you think that was the future?
- 2:40:47So I had historically owned an ESP and that was a function of
- 2:40:51me owning bars and not knowing how to get access to my own
- 2:40:55customers. So that's when I started to
- 2:40:58realize that if you think of the average customer, you need to
- 2:41:01understand two things very well. Where are they reachable and
- 2:41:04where they likely to respond. And people respond across
- 2:41:06multiple channels in multiple ways, different economics.
- 2:41:10You solve those two problems, you have the ability to cut
- 2:41:12through the noise. So at Boom Train, what we were
- 2:41:16doing was really solving that problem, right?
- 2:41:18We were trying to figure out where are you reachable?
- 2:41:21Where are you likely to respond? What are you going to respond
- 2:41:23to? And then using essentially a big
- 2:41:26data pipeline that ingested everything we knew about a
- 2:41:29customer and then made it available to the enterprise.
- 2:41:32Then action on marketing. The interesting thing today is
- 2:41:35that although machine learning has become AI, you still need
- 2:41:38the data pipelines. You're still answering the same
- 2:41:40questions. You just have a different
- 2:41:41approach now, right? The world's become generative.
- 2:41:43So many of the same principles and the modern stack that we had
- 2:41:47put in place has transpired into what we have today and has made
- 2:41:50Zeta really a leader in from the technology space in our in our
- 2:41:53opinion. I like that you said that.
- 2:41:55I mean, and I think a lot of people who have gotten into AI
- 2:41:58the last two or three years don't realize a lot of what we
- 2:42:01call AI now really like we might even calling machine learning.
- 2:42:05Five years ago, generative AI now.
- 2:42:07But many, many AI systems still run on machine learning.
- 2:42:09That's true. Take a look at Zeta.
- 2:42:11I remember I first I used to do a lot of work in e-mail space.
- 2:42:15You know what marketing automation like you know, Fender
- 2:42:18section free SPS and we're running into you guys a bunch of
- 2:42:20times like a lot of trouble. If I remember a lot of like
- 2:42:23really big travel hospitality clients had worked with work
- 2:42:27with Zeta for many years. That was my first time I ran
- 2:42:30into you, but it's been cool to see this evolution from what you
- 2:42:34know, it seemed like a lot of transactional e-mail, guest
- 2:42:37relations, things like that. But then really you've added a
- 2:42:40bunch of stuff over the years that it's been cool to chart
- 2:42:42that evolution. Love, you can maybe talk a
- 2:42:43little bit about that. Yeah.
- 2:42:44So when I joined Zeta in 2017, I think we made two very critical
- 2:42:49decisions about how we thought marketing and advertising were
- 2:42:52going to evolve. The first decision we made it
- 2:42:55was that we believe that acquire, grow and retain,
- 2:42:59Traditionally, 3 use cases that are run across different types
- 2:43:01of platforms needed to be in one platform.
- 2:43:05So you needed to be able to acquire, grow and retain in one
- 2:43:08place. This is not the composable model
- 2:43:11necessarily. But what this model is more akin
- 2:43:13to is where the world is moving. It's a decision system because
- 2:43:17your next best customer might be someone who already bought from
- 2:43:20you previously. So that was decision #1 acquire,
- 2:43:25grow, retain in one platform. We're going to work towards
- 2:43:27that. Decision number 2 is that we
- 2:43:29felt that the identity and data enrichment asset that we could
- 2:43:34put together could power acquire, grow and retain in the
- 2:43:37same way that wall gardens power their environments.
- 2:43:41So we subsequently made acquisitions like discuss the
- 2:43:44largest in commenting platform online live intent the largest
- 2:43:47ad exchange for e-mail, and we cultivated data and identity
- 2:43:51asset that now basically ends up in the form of intelligence and
- 2:43:55intelligence is now actionable by AI.
- 2:43:57So, you know, I like using the metaphor of Zetas like a walled
- 2:44:00garden for the open web because you get very similar mechanics
- 2:44:04and around measurement, around identity, around enrichment that
- 2:44:08you'd find in a walled garden. We made that available across
- 2:44:11channels where it historically has not been available.
- 2:44:14Seems like you're also pulling in features from CRM, features
- 2:44:17from ESP, marketing automation in order to hit those 3 phases.
- 2:44:20That's correct, the. Power of Zeta data, meeting
- 2:44:23customer data in a siloed environment that's available to
- 2:44:26them is what unlocks the value. So if you think about the core
- 2:44:29technologies that underpin Zeta, it really is 3.
- 2:44:32There's a managed media solution.
- 2:44:35So we are a big provider of basically think of it as like
- 2:44:39intelligence driven media display CTV audio, all the
- 2:44:42channels. So the buys actually do you
- 2:44:45connect to DSPS? You have a DSP.
- 2:44:47You have a DSP, OK. Yeah.
- 2:44:48We, we acquired the assets from Seismic Rocket Fuel and then
- 2:44:52obviously needed a lot of retrograding and upgrading, but
- 2:44:54that became a foundation and now we're very much.
- 2:44:57Working on back in the day, Sure.
- 2:44:58Yeah, makes sense, yeah. And now we're very much moving
- 2:45:00in the eugenic space. The other core technology behind
- 2:45:02Zetas is CDP, so we have a data foundation.
- 2:45:06Was that homegrown or was that an acquisition?
- 2:45:07It was homegrown and that was much more adjacent to our
- 2:45:10marketing automation stack. And if you follow Forrester,
- 2:45:12we've been the number one rated marketing automation platform
- 2:45:15for owned media, namely e-mail, SMS website for the last three
- 2:45:19years, so. You were, it's like this the
- 2:45:21last four or five years has been this big push for OK, activating
- 2:45:25first party data, you know from CRM and owned media on paid.
- 2:45:29You were really early to this. Very early, yeah.
- 2:45:32And our ability to, you know, resolve identity across the open
- 2:45:36web to resolving identity that comes from our customers
- 2:45:40identity they don't even know is probably the best in the world.
- 2:45:43So just I can give you a data point on it.
- 2:45:47A visitor on the open web in the US not authenticated, we can
- 2:45:50likely resolve to our graph with the first name, last name,
- 2:45:53postal address, phone number about 50% of the time.
- 2:45:56So you can imagine the resolution rate we can bring to
- 2:45:58enterprises that we. IDR service that you built.
- 2:46:01Yes, it's, it's, it's baked into our platform, but we wouldn't
- 2:46:03sell identity resolution as a stand alone service.
- 2:46:07The way that Zeta brings our solutions to market is you would
- 2:46:10use this for media activation, right.
- 2:46:12So either own media activation, data infrastructure and CEP or
- 2:46:16or paid media activation and then you get our platform
- 2:46:19solution that underpins it. And and really, if you think
- 2:46:22about the last 10 to 15 years, you know, the consolidation onto
- 2:46:26data bricks and snowflake have been the setup for what this is
- 2:46:30today because now marketers have brought their data together.
- 2:46:33Shouldn't that be data be used in one place to acquire, grow
- 2:46:37and retain the AI can work off of it on in one place.
- 2:46:40So all of these pieces have been coming together in a way that
- 2:46:44candidly like we didn't even expect it to be this quick, but
- 2:46:48the the tailwinds that have been created over the last few years
- 2:46:50are really powering Zeta's growth.
- 2:46:53Do most of your customers have a cloud data warehouse like a
- 2:46:55snowflake or data bricks they're using?
- 2:46:56Yeah. Generally, they're connecting to
- 2:46:58a Cloud Warehouse 0 copy into Zeta.
- 2:47:00And then obviously we provide them with a lot of identity
- 2:47:03enrichment. We provide them with data
- 2:47:05enrichment. And then really the core of
- 2:47:07those, we provide them with intelligence to acquire, grow
- 2:47:09and retain in ways that they wouldn't be able to do without
- 2:47:11us. There's some like one year.
- 2:47:14I mean it's a really comprehensive solution.
- 2:47:16Do you usually sell it as a suite together or if you go in
- 2:47:18and like some clients will say I want to CDP, some will say I
- 2:47:21want e-mail, like how do you usually?
- 2:47:22Approach, it usually starts with a use case so it's either one of
- 2:47:25the three but interestingly enough, our happiest customers
- 2:47:29are the ones that do the most with us, which is a really good
- 2:47:31sign because when you're using us as your system of truth for
- 2:47:35your CDP and for your own media and then you start doing paid
- 2:47:38media with us, that gives us a lot of potential to grow.
- 2:47:40It also gives the customers a lot of potential to really make
- 2:47:43their marketing efficient. And in this new world I like
- 2:47:46using the expression context is the new content.
- 2:47:49So if you think of the old world you'd have hey, I have a profile
- 2:47:53with 300 attributes on it. I'm gonna use that through a
- 2:47:57machine learning model to determine what I should send to
- 2:47:59somebody. But in the new world, because AI
- 2:48:01can operate off of real time data and your context about what
- 2:48:05you're trying to do right now in this moment determines what you
- 2:48:09are should be shown and presented with and where data's
- 2:48:12graph is real time because our code lives expressively across
- 2:48:16the entire open web, right? Think of very large publishers
- 2:48:19to very small publishers, and we're able to really decipher
- 2:48:22what the context of an individual is in that moment and
- 2:48:25then use that for. Solution.
- 2:48:27Someone could come in on one channel.
- 2:48:29Let's say you could, you could you could come in a web form or
- 2:48:32you could do a cookie, a cookie sync and match them.
- 2:48:34And then your user identity graph you identify enriched if
- 2:48:40need to be started to see who they are.
- 2:48:42And then you then you can build out like you could, you know,
- 2:48:45wherever they are, you're going to be able to touch them.
- 2:48:46Right. We can touch them anywhere.
- 2:48:48We can also tell you their patterns across different
- 2:48:50competitors. Are they shopping more with you
- 2:48:52or your competitor? How would you get that person to
- 2:48:55shop more with you? If they're with your competitor,
- 2:48:57what are their drivers? What are their needs?
- 2:48:58States our. Our graph exposes all of that
- 2:49:00type of. Information who you usually
- 2:49:03competing against when when your competitive sales are these like
- 2:49:06these like CDP, RF, PS are these CRM?
- 2:49:10Are they, I mean, because you could you do a lot of things.
- 2:49:12Yeah. So it's a good question.
- 2:49:15You know, it depends on the in Rd. that someone has taken to
- 2:49:18get to Zeta. So on the, you know, the RFP
- 2:49:21side that generally will control CDP or marketing automation.
- 2:49:25We see all the players, the big guys, right, the Adobe's, the
- 2:49:27sales forces of the world on the media side is typically a share
- 2:49:31shift, right. So it's, hey, I'm using Zeta now
- 2:49:34because Zeta delivered better results and I was able to
- 2:49:37measure the outcomes with more efficacy.
- 2:49:40So it's usually not, you know, winner takes all in the media
- 2:49:43space. It's there's a collection of
- 2:49:44partners, but Zeta is taking more media share from others
- 2:49:48because we're delivering. Performance media definitely
- 2:49:51more fractured complex. Who's doing the buys the DSP?
- 2:49:55Is it typically an agency? Is it a like a brand in house
- 2:49:58team or is it, it's probably all over the place, right?
- 2:50:01It's it's all over the place. We have a very large agency
- 2:50:03business. We platform their businesses, we
- 2:50:05give them access to our tools and they can actually operate
- 2:50:08within our platform. We're also seeing if you start
- 2:50:11on the enterprise side with our CRM solutions or CDPR marketing
- 2:50:15automation stack, you very quickly move into the media side
- 2:50:18there too. And sometimes that's an analyst
- 2:50:19team or sometimes they will direct their agency to work with
- 2:50:22us. So it can go a number of
- 2:50:23different directions. And is the is a lot of the media
- 2:50:28businesses in more social? Is it open web?
- 2:50:30Is it like? Yeah, it's, it's, it's
- 2:50:32definitely more open web and and certainly more CTV these days,
- 2:50:37right. So CTV display are probably the
- 2:50:39two largest channels there. We do get directed to do social
- 2:50:44and and by the way, some people are reachable on social and
- 2:50:47likely to respond on social. But it's part of an integrated
- 2:50:50strategy that our AI systems will help determine.
- 2:50:53Are you able to share anything from your road map that's
- 2:50:55planned out over the next few months, years you're really
- 2:50:58excited about? I think the main thing and it's
- 2:51:01on stage here at Possible that will be on stage in Cannes in a
- 2:51:06in a couple weeks, is Athena, right?
- 2:51:09Athena is our decisioning AI that sits directly within our
- 2:51:14platform. It can talk to the Zeta data, It
- 2:51:17can talk to customer data. It can help you navigate the EO
- 2:51:19likes to use the metaphor that we built a 747 customers know
- 2:51:24how to fly a Cessna. Athena can help bridge the gap
- 2:51:27to get maximum value from Zeta and take an ROI that we're
- 2:51:30already delivering and make it even greater for our customers.
- 2:51:33And it is a amazing capability that's embedded into the
- 2:51:36platform. And I know, you know, if you go
- 2:51:38to websites today of Zeta and even our competitive set, it all
- 2:51:42sounds the same. You know, everyone sounds like
- 2:51:44they're doing the same thing. I encourage.
- 2:51:46There's a lot of smoke and mirrors.
- 2:51:47Yeah, there's a lot of smoke. I encourage everyone to actually
- 2:51:49go and try to ask for a demo, see the product, even interact
- 2:51:54with it yourself so you get first hand experience of like
- 2:51:57what it means to be agenetically turbocharged in a marketing
- 2:52:02capacity. Now, what can you do that's
- 2:52:03faster? How does measurement become
- 2:52:06easier? How does data analytics become
- 2:52:08easier? These are all the things you can
- 2:52:10actually do in the platforms. I encourage everyone to try it
- 2:52:12out. Any thoughts on, I mean, there's
- 2:52:15been a lot of talk about how OK, as these AI tools get
- 2:52:20capabilities get better and tools get rolled out that, you
- 2:52:24know, because you can connect by MCP server, you know, maybe the
- 2:52:27UIS and the front ends of these things, you know, you know, head
- 2:52:30for Salesforce just went head for the other day, for example.
- 2:52:32Like thoughts on like how that might impact Zeta?
- 2:52:37Like are you preparing for that? I'd love to hear your thoughts.
- 2:52:40On that, yeah. So I think that the main thing
- 2:52:41right now, and I, I wrote something on Medium about this
- 2:52:44few weeks back, is that there's a lot of noise about who gets
- 2:52:48disrupted, who gets to stay and why, right, the sass apocalypse.
- 2:52:52But the reality is, is that if your decisioning system is set
- 2:52:56up correctly for the enterprise, it provides them with so much
- 2:52:59value that the friction to leave is too high.
- 2:53:03And so the thing I love about Zeta's model is that we can
- 2:53:06easily work with enterprises and their vendors, but we
- 2:53:10consolidate intelligence into our platform in a way that makes
- 2:53:14us very sticky. And it's not about a channel,
- 2:53:16it's not about a media strategy. It comes back to having a system
- 2:53:20in place that helps an enterprise do more.
- 2:53:22And by system, you can't do one thing.
- 2:53:25You have to be able to acquire, grow, retain.
- 2:53:26You have to be able to provide enrichment, provide identity.
- 2:53:29Those are all the pieces and building blocks that end up
- 2:53:32being very, very hard to replace in the market.
- 2:53:34I think it's a good point. I mean like I guess the way I
- 2:53:36look at it is if you build really solid capabilities, you
- 2:53:39can look at them almost as engines, right?
- 2:53:41Like regardless of who agents they are and maybe brands will
- 2:53:44even have their agents. The agencies, I don't really
- 2:53:46think it matters, but if you have these agentic, either
- 2:53:50agents or agentic workflows you're going to cross to build
- 2:53:54across things, if you're building the best capabilities
- 2:53:59that work the best, yeah, that's where the action still owns the
- 2:54:03brain totally. That's the thing, you gotta own
- 2:54:05the brain, and Zeta's system is set up to essentially be the
- 2:54:08brain of the enterprise. I really like that a lot.
- 2:54:11How's the show been for you so far, by the way?
- 2:54:12It's been hot, but I've been it's.
- 2:54:14Getting hotter here in the sun. That sun keep stuff.
- 2:54:15Yeah, beautiful view, really enjoying the show.
- 2:54:18You know, possible is one of the the touch points of the year
- 2:54:21that I am really excited about because get to meet all of our
- 2:54:23HD partners, get to meet some important data venture partners
- 2:54:26of ours. And then customers, you know,
- 2:54:28you know our, our direct enterprise customers are here
- 2:54:30and they're all anxious to learn and in the, in the Miami Heat.
- 2:54:35It's been cool watching possible evolve.
- 2:54:38This is my third year here. I missed the first year, but I
- 2:54:40know it's, I think it started very ad tech heavy, became let's
- 2:54:44say more brand heavy last time today, you know, they weren't
- 2:54:47sending brands to come here, offered to even pay their
- 2:54:49tickets if that went to three vendor meetings.
- 2:54:51And then, you know, look, I'm with Omnicom, There's a ton of
- 2:54:54Omnicom and other AGC people here that I did not see that
- 2:54:57before. So the whole thing, and I guess
- 2:54:59it makes sense, you get the brands, the agencies will
- 2:55:03follow. So of course, so it's been cool
- 2:55:04and you know, the, you know, the tech vendors I think have been
- 2:55:06here from the start. So I've had some incredible
- 2:55:08meetings here with people including including this.
- 2:55:10So you know, it's been a lot of fun and I wish you lots of
- 2:55:13success in the show and I'd love to love to connect up.
- 2:55:16I can again too. We'd love to.
- 2:55:18I'll be there, would love to do another one of these and tell
- 2:55:20you what's changed. And we're going to be announcing
- 2:55:22some new stuff it can. So really looking forward to
- 2:55:24that as well. You're interested, we do longer
- 2:55:26like 6090 minute long form podcasts on on the regular pod.
- 2:55:29You know what, like you know, let's keep in touch and when you
- 2:55:31get a bunch of stuff you're we want to talk about it and go
- 2:55:33deep on let me know. We can schedule another 100%
- 2:55:36Nish. Thank you so much.
- 2:55:37Great to see you. Great to see you.
- 2:55:38Thanks, Rio. Noise at the at the possible
- 2:55:42conference on day 2. We've actually moved inside
- 2:55:45because it's piping hot out there on the balcony.
- 2:55:48We've got Tom Koch here from 12 Labs and yeah, thank you for
- 2:55:52joining us. I want to hear a little bit
- 2:55:54about you're coming in really excited to hear about what you
- 2:55:57guys do and how it's playing into the media space.
- 2:56:00But First off, a little bit of an intro to yourself.
- 2:56:02What's wherever you come from? What's your background in the
- 2:56:04industry? Yeah, well, it's it's Tom Cook
- 2:56:06to start. I'll protect my German brand
- 2:56:08last name, but yeah, I gotta get that out for the industry.
- 2:56:10But yeah, you know, I did a decade an ad tech at App Nexus
- 2:56:14and the whole ATT's Andrew Warnermedia acquisition thing.
- 2:56:17Then Microsoft for a while started in more technical
- 2:56:20consulting and programmatic who've been the enterprise
- 2:56:22sales, the intersection of big data, clean rooms, identity,
- 2:56:25enterprise technology for media buyers and sellers there for a
- 2:56:28long time. And then found this company
- 2:56:29called 12 Labs, which is doing really cool things in the AI
- 2:56:32space. It's a small research company
- 2:56:34that is focused on building foundation models for multimodal
- 2:56:38video understanding. So unlike LLMS that were built
- 2:56:41for texts and they're amazing at reasoning on PDF files and texts
- 2:56:44and all the information, we build models that T hear and
- 2:56:47understand and reason on unstructured video data.
- 2:56:49I'm looking through decades of content and I'm looking for
- 2:56:51specific moments or actions or expressions, machines that can
- 2:56:55and comprehend sort of what we're watching and seeing on a
- 2:56:57screen the way that our brains think about it.
- 2:56:59Yeah, wow. OK, so this is, this feels like
- 2:57:01already the next iteration of what what AI is promised to be,
- 2:57:05right. It's it started off as becoming
- 2:57:07as reading a book or reading the Internet and it will end up in
- 2:57:11real world. And this feels like somewhere in
- 2:57:13between that. Yeah, I like, I didn't know this
- 2:57:16about sort of the world of data before I started this company,
- 2:57:19but if you look at it by density, video is 90% of the
- 2:57:22world's enterprise data set. And there's been a lot of
- 2:57:24attention on LLMS in the world of text and sort of reasoning to
- 2:57:28become more efficient in our sort of modern day-to-day.
- 2:57:31But you think about all of the media operations teams that sit
- 2:57:34at some of the largest studios in the world that can't find
- 2:57:36things like Steamboat Willie and the Disney Archive because
- 2:57:40nobody tagged that 70 years ago. Yeah.
- 2:57:42So there's enormous opportunity for the editing community,
- 2:57:45production community, the advertising industry that thinks
- 2:57:48about how do I segment ads more effectively, how do I make sure
- 2:57:50that content is brand safe and compliant to bring similar
- 2:57:53automation efficiency in the video space.
- 2:57:55And that's the research that we're really focused on.
- 2:57:58Wow. OK.
- 2:57:58And it's all coming out of like you said, you're a research lab.
- 2:58:01It's an interesting positioning for a business that's also then
- 2:58:05going in and segmenting media and and and doing that.
- 2:58:07So how does that do you find that that those two narratives
- 2:58:11help each other or do they do you find that they interrupt
- 2:58:14each other? It's really funny, right?
- 2:58:16Like we in the ad tech space have like a lot of great market
- 2:58:19decks and like build a lot of applications.
- 2:58:21But you know, we're an infrastructure company that's
- 2:58:23kind of showing up on the scene for, you know, relatively the
- 2:58:25first time I joined the company in July to start really like
- 2:58:28building a presence for the business and media advertising.
- 2:58:30But our Co founders were PhD track students in the United
- 2:58:34States doing machine learning degrees that were originally
- 2:58:36born in Seoul, South Korea. So they went back after their
- 2:58:39studies to do Cyber Command defense work, looking at large
- 2:58:43scale CCTV and drone footage, and realize that like the early
- 2:58:46Transformer model research around what LLMS are, it's not
- 2:58:49very good at understanding video.
- 2:58:51The way that an LLM that's built on text looks at video is to
- 2:58:54take pictures of frames every six to 12 seconds, convert that
- 2:58:57into tokens and try to guess what happens.
- 2:59:00Do you want to like get better at the models performance?
- 2:59:02It's obviously not a great way to understand what's happening
- 2:59:04in video content. You take 4 pictures and more
- 2:59:06frames and that's really expensive.
- 2:59:08So we have this like amazing sort of roundup research
- 2:59:11background to build a business and then have gone out into
- 2:59:14horizontal industries where we have to think about how do we
- 2:59:16productize this, how do we commercialize it?
- 2:59:19And AD tech. That's how do publishers think
- 2:59:21about landing segmentation and AD servers and SSPS and DSPS.
- 2:59:25The world that I come from. So it's really fun space.
- 2:59:27I'm kind of out add tech, but still kind of back in it and the
- 2:59:31creator economy is exploding. You need to imagine that one of
- 2:59:33the biggest problems that creators and networks have is
- 2:59:35managing the explosion of UGC data, right?
- 2:59:38There's more than 500 million hours of, you know, video
- 2:59:41content getting uploaded to YouTube like on a daily basis.
- 2:59:43How do you manage all that? OK, so that like I have AI had
- 2:59:47actually an interesting or I thought was an interesting point
- 2:59:51earlier, which was I see a lot of obviously in AD tech market,
- 2:59:54you see a lot of businesses that you know, will feel that promise
- 2:59:57of classify that being segmented and then, you know, push it out
- 3:00:00for sales, contextual businesses, right everywhere.
- 3:00:03And so you you've sort of alluded to the fact that as a
- 3:00:06research and infrastructure heavy company, the methodology,
- 3:00:09the magic is in the methodology almost as a sort of maybe a more
- 3:00:14blunt sort of instrument on the other side, which is just
- 3:00:17classifying a video by words and.
- 3:00:19By its metadata, yeah. Is that right?
- 3:00:21Yeah, I think the way I think about it as like a OG ad tech
- 3:00:26person is we started treating video when like especially when
- 3:00:29CTV became a thing in like 20/17/2018, then exploded in
- 3:00:322020 was like treating it like text, like treating it like
- 3:00:36banner ads. You know, we have like genre
- 3:00:37classification and shallow signals about what's happening
- 3:00:41in content, but we have no scene level intelligence about what's
- 3:00:43actually happening on the screen.
- 3:00:45And so we didn't like re architect the infrastructure for
- 3:00:47delivering ads around what was built for the open web.
- 3:00:50And that's where I see a lot of the opportunity is to get really
- 3:00:52deep about the content expression, who it's built for,
- 3:00:55to start to personalize that. And like the first thing that we
- 3:00:57focused on in like addressable TV and CTV inventory, it was
- 3:01:01really just like replacing cookies, Like how do I use clean
- 3:01:04rooms to match data sets? And that's not personalizing on
- 3:01:07contacts, that's personalizing on the user.
- 3:01:09And so I look at a lot of the disruption that will come in the
- 3:01:12video space is going to be around understanding how to
- 3:01:15match content and creative to what's actually sort of built in
- 3:01:18the production house for the consumer itself.
- 3:01:20Yeah, right. And and the technology because
- 3:01:23of what it is, it will when once you start doing that, it will
- 3:01:25start, start to uncover maybe 234 layers of intent rather
- 3:01:30than, you know, rather than just a broad sweeping one.
- 3:01:32Yeah. I think one of the ones I get a
- 3:01:33lot, you know, like the, the pushback I get on this and
- 3:01:36there's, there's companies in the market that are doing this
- 3:01:38is like, and I get so granular in the content, I'm going to
- 3:01:41really start to shrink the audience that I'm trying to
- 3:01:43address. The very basic use case is like,
- 3:01:46you know, for larger, you know, VOD libraries that contain
- 3:01:49sports documentaries and sitcoms, sports really monetizes
- 3:01:52better at a higher CPM for the publisher.
- 3:01:54And there's so many moments that I would watch in Modern Family
- 3:01:57of like a football game or a basketball game that goes into
- 3:02:00an ad break that doesn't get classified as sports to be
- 3:02:02rolled into that package because it's classified as a sitcom.
- 3:02:05And so that's a really basic thing where it's just I'm going
- 3:02:08to understand the inventory that I actually sit on top of to
- 3:02:10start to package it differently for media buyers to the benefit
- 3:02:13of the brands. Those are great moments within
- 3:02:15warm living room scenes that you know Nike could participate in
- 3:02:19if they knew it was sports. Yeah, OK.
- 3:02:21That makes sense. The, I mean the, I have one
- 3:02:24question that sort of relates to for we get to the creator
- 3:02:27economy, which you mentioned and I really wanna jump into as
- 3:02:30well. But you you mentioned earlier as
- 3:02:32well that you it's a, it's a developer ecosystem really comes
- 3:02:36to you, right. How does that change your go to
- 3:02:38market motions and how does that change how you find customers
- 3:02:42and how you, you know, as it yielded any surprises?
- 3:02:44It must, I guess if you see organically who's using your
- 3:02:47products, you must find a few little surprise Nuggets in there
- 3:02:50that that you can then blow up into a real commercial strategy.
- 3:02:54Yeah. So like our first customers
- 3:02:56found us. So we've been working on this
- 3:02:58for six years. It takes a long time.
- 3:03:00I got a question. A lot is like, you know, are you
- 3:03:02related to 11 labs? 12 labs, actually.
- 3:03:04Yeah, I was gonna, I was gonna ask that question.
- 3:03:06I thought you'd have had it too many times.
- 3:03:08It's a like native multimodal video.
- 3:03:10Understanding is one of the hardest science problems in the
- 3:03:13world because video has multiple modalities that you need to
- 3:03:16track over space and time for a machine to understand.
- 3:03:18So that example I get of an LLM that treats it like, you know,
- 3:03:21pictures every six to 12 seconds doesn't work at scale to
- 3:03:25understand what a machine what's actually happening.
- 3:03:27So we had to like literally build ground up, you know, a
- 3:03:30model that can take content, index it, create vector
- 3:03:33embedding spectrum. Embeddings are the data format
- 3:03:35machines speak to that track vision, audio, like the sound
- 3:03:39that's spoken, what's playing in the background, transcription,
- 3:03:42how that moves over sequences of time.
- 3:03:44That's like the cognitive reasoning that we have.
- 3:03:46What's cool is like the first biggest customers that found us
- 3:03:49were the largest media companies in the world.
- 3:03:52One of the examples that we've talked about on stage with the
- 3:03:54NFL is like nobody thought to start logging many people
- 3:03:58sitting watching football games every year.
- 3:04:00Nobody started tagging when Travis Kelsey was making hard
- 3:04:02signs of Taylor Swift. That's when that go retrieve,
- 3:04:05right with a viral cultural moment to participate in that
- 3:04:08nobody could have monetized. Nobody could have gotten out to
- 3:04:10social. So from there, right, it's like
- 3:04:12we launched an AWS bedrock last year that's like kind of the
- 3:04:16supermarket for AI models in AWS.
- 3:04:17Yeah, that's where we never have to move their data.
- 3:04:19They can start using our models. We start seeing smart home
- 3:04:22manufacturing companies, automotive manufacturing
- 3:04:24companies, commercial security companies using our models and,
- 3:04:28and what's hard is like a small revenue team that's product LED
- 3:04:30is like we want to know who those people are and figure out
- 3:04:32what they're building. So we're starting to do that,
- 3:04:34which is really exciting because AI is only so good as the
- 3:04:37business value it delivers. So we want to get better at
- 3:04:39going and consulting them how to use the model, how to build
- 3:04:41products to make them more intelligent in their video
- 3:04:43workloads. And I presume that when you're
- 3:04:45talking about smart home and these sorts of vendors, they're
- 3:04:47not talking about monetization the same that a media company is
- 3:04:51a very. Different world for me, right?
- 3:04:53Like they're thinking about how do I understand that, you know,
- 3:04:56the edge device that I have captured somebody delivered my
- 3:04:59groceries. How do I like rapidly sign the
- 3:05:01push notification to somebody that's sitting at the office
- 3:05:03what, you know, was delivered? But anyway, I should be really
- 3:05:05good at that. Let me transcribe it.
- 3:05:07Maybe the groceries weren't delivered on the front doorstep.
- 3:05:09They were like left beat the stairs.
- 3:05:11Models can do that. But that's completely different
- 3:05:13Hughes case than going and talking to a streaming publisher
- 3:05:15about how to monetize more effectively.
- 3:05:18Yeah, fantastic. And that's again, that's
- 3:05:20something that you're having a developer ecosystem and an
- 3:05:22infrastructure based product would give you, give you those
- 3:05:25little seeds of clues that that you may have not thought of
- 3:05:29yourself, you start. Seeing mass and then maybe we
- 3:05:31should go invest in more people and more time making this work.
- 3:05:34And that's kind of how we've grown through different cycles.
- 3:05:37It's yeah, it's really interesting.
- 3:05:39And it and it that sort of, yes, brings US full circle to this,
- 3:05:42this topic of the the creator economy, which is a vertical
- 3:05:46though if you call it a vertical, but it's an area you
- 3:05:48look after in your, in your role.
- 3:05:50And you, you mentioned obviously just Taylor Swift and Travis
- 3:05:53Kelsey, which is that moment, that viral moment, that
- 3:05:56unexpected live moment that just would not have gone monetized in
- 3:06:00a traditional format with traditional data collection and
- 3:06:02a traditional sort of media company.
- 3:06:04In that way, the creator economy is going to be full of that,
- 3:06:08right? So is the is the objective to
- 3:06:12earn in real time these new moments or for these businesses
- 3:06:16and turn that into an entirely new kind of monetization
- 3:06:20package, product etcetera? I think we've started like
- 3:06:23really following the path from like content being created to
- 3:06:26like monetizing that, finding demand for advertisers to match
- 3:06:30to that to measuring campaigns and then these new like AI
- 3:06:33licensing revenue streams. So to start from the beginning,
- 3:06:35like the most exciting thing that we announced at NAB last
- 3:06:39week was we built what we see in the like research community as
- 3:06:42the first multi step reasoning agent for video data.
- 3:06:45So mostly we're an API, we're a developer product, but we built
- 3:06:48an application for creators, post production teams to drop
- 3:06:52and their whole library of footage, chat with it like
- 3:06:54Chachi, BT and get to a quick sizzle reel that goes in like
- 3:06:57multiple vertical video formats. So if I'm a creator and think
- 3:07:00I'm like Mr. Beast, right? Like I just shot, you know, 3
- 3:07:02hours of content. If I'm signaling noise, I just
- 3:07:04shot interviews all day. I want to drop all of that in
- 3:07:07and get to like a one minute sizzle reel and have that in
- 3:07:09multiple aspect ratios. So that's like content
- 3:07:11production and what we're doing. Then it's like, you know, I need
- 3:07:14to go and connect this to brands.
- 3:07:16I want to make sure that that's brand safe and compliant.
- 3:07:18People didn't smoke or swear things that, you know, a Nike as
- 3:07:21an advertiser wouldn't want the content.
- 3:07:23You can use AI to automate all of that and look at rules you
- 3:07:26want to do about matching brands to creators, right?
- 3:07:28Like so many problems in networks is that all the
- 3:07:31networks uses how many followers somebody has and what they
- 3:07:33generally talk about. But there's nothing that can
- 3:07:35actually take an RFP from advertiser and search within the
- 3:07:38library to say maybe this nano creator is actually the perfect
- 3:07:41brand for perfect creator for me.
- 3:07:43So now I'm helping creators monetize more effectively in an
- 3:07:46agency paradigm. Then I want to measure things.
- 3:07:48When did the logo appear? What was mentioned?
- 3:07:50What time stamp did all these things happen that starts to
- 3:07:53inform post campaign reporting? And then like I'm a big network.
- 3:07:56I have all this data. Like all the big AI labs like
- 3:07:59Open AI and Anthropic, they're trying to buy data and train on.
- 3:08:01So I have all this video, you know, just like Rod dump a data
- 3:08:04file, I need to annotate that. I just say this is the value of
- 3:08:08the data you're looking for. I can search for all of that.
- 3:08:10I can annotate it and now I can market or commercialize it.
- 3:08:13So we're starting to like get it right.
- 3:08:14Like we kind of have video pane in every step of the workflow.
- 3:08:18Yeah, all those things are either like human computed or
- 3:08:20there's computer vision technology which can tell you
- 3:08:23what's in a frame, but I can't tell you what happened in the
- 3:08:25scene. So that's really where we start
- 3:08:26to step in and differentiate. Yeah.
- 3:08:28And I, what I think what fascinates me most about what
- 3:08:31you're saying there is like just the caliber of tools that are
- 3:08:34available to, you know, not just multi $1,000,000 enterprises,
- 3:08:38but you know, individual creators sometimes who are
- 3:08:41solopreneurs. And I could, I could hear Rio
- 3:08:44and Brett's ears prick up as well for, for signal and noise
- 3:08:47when you're talking about talking to your video library.
- 3:08:49So it's, you may have another customer there, but that's the
- 3:08:52idea, right? You can.
- 3:08:53You can almost democratize the idea of talking to video or
- 3:08:57doesn't matter how you know, big or small you are.
- 3:08:59I think the big changes that we see are both in performance and
- 3:09:02cost. So computer vision was a great
- 3:09:04solution and we still work in conjunction with computer vision
- 3:09:07models a lot. Think of like brand sponsorship
- 3:09:09measurement. I was fascinated to learn this,
- 3:09:11like when Coke does and using general, you know, figures like
- 3:09:15when a brand does a deal with the NBA Finals to integrate into
- 3:09:17the game. People are still hiring a
- 3:09:19measurement companies like people around the country to
- 3:09:21watch that broadcast, count how many times the logo appeared and
- 3:09:24that is your reporting panel. Like you should not be like
- 3:09:26hiring people to watch basketball games.
- 3:09:28Like imagine how many, you know, college students are watching
- 3:09:30that game and just saying, yeah, 45.
- 3:09:32Like, you know what? So, you know, I think like
- 3:09:34computer vision though is really good at identifying it was in
- 3:09:37this frame. And what you don't want to do is
- 3:09:40have a computer vision model that's pretty expensive.
- 3:09:42Sample every frame in a three hour NBA Finals broadcast.
- 3:09:45You know, you should be using more effectively to get to a
- 3:09:48more cost effective pipeline. Now the performance is there
- 3:09:51where like models can natively understand things like our brand
- 3:09:54that I can start to Add all of these features and capabilities.
- 3:09:57What didn't exist three years ago was both like the
- 3:09:58performance and being able to reason like that and then the
- 3:10:01cost. So a lot of people I talked to
- 3:10:03you were like, I tried this with an LLM, I tried this computer
- 3:10:05vision model. I gave up is the ROI didn't
- 3:10:08work. It was so expensive that I
- 3:10:09couldn't sell this at ACPM or I couldn't inject into my buying
- 3:10:12workflow because all the ROI broke down.
- 3:10:14So that's like the really cool thing is a lot of science
- 3:10:16projects that were treated as failed experience like now are,
- 3:10:20you know, kind of like things that we're picking up and doing
- 3:10:22with customers. Yeah.
- 3:10:23And it sounds to me like your experience has given you a
- 3:10:25really good lateral understanding of connecting the
- 3:10:28dots of these all these these ambiguous areas you're going
- 3:10:30into. But you know, that may just turn
- 3:10:33into something completely different in a year or.
- 3:10:35Yeah, I would say like the whole App Nexus thing into like ATT
- 3:10:38trying to become a media company and then Microsoft, you know,
- 3:10:41doing all the things was definitely a good place to
- 3:10:43learn. Absolutely.
- 3:10:44Well look, Tom, it's been an absolute pleasure to talk to you
- 3:10:47and make sure you enjoy the rest of possible.
- 3:10:50Try to try to grab some solo time without corporate events
- 3:10:55and then we'll see you next time.
- 3:10:56Cool. Thanks for having me.
- 3:10:57Thanks. House from possible 2026 Day 2,
- 3:11:01we're getting towards the end of the day.
- 3:11:03I have the Global Chief Revenue Officer of Nomics Group, Todd
- 3:11:06Ulis just today who has got a lot of that.
- 3:11:09Well, what I want you to tell everybody a little bit about
- 3:11:11yourself and about Nomics. You're educating me ahead of the
- 3:11:15pod. Apologize for not doing my deep
- 3:11:17homework on the topic, but a lot of critter economy conversation
- 3:11:20and I think there's going to be some interesting topics to cover
- 3:11:22here. But tell us a little bit about
- 3:11:23Nomics and about your sort of journey into the organization.
- 3:11:26Absolutely, Todd Ulis, Global Chief Revenue New Officer of
- 3:11:28Nomics Group. I've been in digital marketing
- 3:11:30for 26 years, so I started when I was 3.
- 3:11:33But funny, funny, funny man so. Yeah, I've been my 4th company
- 3:11:38in 26 years and kind of you find the pain point, solve the pain
- 3:11:41point, you build the business and then you sell the business.
- 3:11:44That's it. And you know from I started off
- 3:11:46with a company called Traffic Marketplace to be one of the
- 3:11:48biggest ad networks in the world, sold the Universal for
- 3:11:50under 60 million in 2001. Solve the problem of performance
- 3:11:53economics back in 2001. Yeah, the economy was down then.
- 3:11:57Solve the problem of perform performance economics to qualify
- 3:12:00that. So think about, think about what
- 3:12:02happened in 2000. Yeah.
- 3:12:04You know, obviously all the dot coms went like this in terms of
- 3:12:06bust. Yeah.
- 3:12:07So publishers were hurting. They were looking for nude
- 3:12:09format, new format at, not nude format, new format ad units.
- 3:12:13And we kind of brought to the marketplace this thing called
- 3:12:15the pop under ad unit, Netflix X10.
- 3:12:17So you can blame me on Sports Center.
- 3:12:19They said we're in more places than an X10 camera.
- 3:12:22So we powered the distribution pipeline behind that was really
- 3:12:25distribution that I built. Yeah.
- 3:12:26So, you know, Universal took notice, bought a company that
- 3:12:29was highly profitable. This is pop under technology.
- 3:12:31This is just how to serve ads. Yeah, it was funny that Rio and
- 3:12:34I think the Co host of the signal noise.
- 3:12:36We're part of a company called Pop Up Technologies that was was
- 3:12:39and it was the first pop up or one of the early rich media.
- 3:12:43Under more, we're more invasive. We were less invasive.
- 3:12:46Yeah, yeah, interstitials. And it was right at that period
- 3:12:48where there was the collapse and they basically had done a small
- 3:12:52raise and then just revenue went down the tube.
- 3:12:55We actually ended up trying to resuscitate the business and
- 3:12:57that's how we kind of got our start in ad tech and it was
- 3:12:59early days. This was still when they were
- 3:13:00delivering the Internet on like AOL discs through the mail with
- 3:13:04the modem dial up. You remember that I.
- 3:13:06Remember that? Hey, you know, my mother never
- 3:13:08wanted to cancel her AOL e-mail like subscription because she
- 3:13:11lose her e-mail address. Yeah, so I get it.
- 3:13:13Yeah, exactly. So yeah, so, so yeah.
- 3:13:15So you did that and then and then keep going so.
- 3:13:17We ended up bestselling that and met a friend of mine from
- 3:13:19college. She calls me up.
- 3:13:20He was scuba diving with a guy named Dane Cook in Ecuador
- 3:13:24looking for sunken treasure and he knew I knew digital
- 3:13:27marketing. Go Who's Dane Cook and what is
- 3:13:28he seeing? Ended up producing Dane Cook.
- 3:13:31We took Dane to the marketplace and I got the AOR for Comedy
- 3:13:34Central Records for a number of years and my buddy won a bunch
- 3:13:36of Grammys. I worked with all the top
- 3:13:38comedians of the world and with Myspace of social media ads back
- 3:13:41in 2004. You know, vertically integrated
- 3:13:43in the comedy space because comedy marketing, a problem, was
- 3:13:46underserved. Yeah, right.
- 3:13:47And were you into comedy? Because I think I saw that on
- 3:13:49your on your LinkedIn profile. I'm not funny.
- 3:13:51Like mediocre comedy. Or well, I'm not funny.
- 3:13:55I asked my children. Yeah, I asked my older daughter.
- 3:13:59I say, OK, what rhymes with with dad?
- 3:14:01Like dad jokes. She goes bad, I go no daughter.
- 3:14:04Rad, rad jokes rad. So you like?
- 3:14:07That those are Dad. Jokes, yeah.
- 3:14:09Exactly. You can empathize with me so.
- 3:14:12So you probably had a little bit of experience.
- 3:14:14With, yeah, you know, I. Comedy.
- 3:14:15Working with comedians, you learn about delivery, a lot of
- 3:14:19it's delivery and communication and timing.
- 3:14:21Yeah, timings. I mean, obviously what you guys
- 3:14:23are doing, obviously with Signal Noise podcast, you know, it's
- 3:14:25like timing, delivery, communication.
- 3:14:27It's essential. So you know, it was, you know,
- 3:14:30comedy markets and I can, you know, we can do an hour and a
- 3:14:33half long podcast. I can spend, you know, I can
- 3:14:35spend probably the whole time telling you stories about
- 3:14:38comedians. Yeah.
- 3:14:39And stuff that we've worked. With yeah, it was, there's
- 3:14:42probably highs and lows, Yeah, with comedians, right?
- 3:14:44You know, like there's the highs, which are extremely
- 3:14:46hilarious, but then there's there's a dark side to a lot of
- 3:14:49very funny people, right? It's almost like it, really.
- 3:14:53Is you know. What I'm talking about?
- 3:14:54Well, I mean, some of these comedians are geniuses.
- 3:14:56I mean, but they're just not able to live in a real world.
- 3:15:00Yeah. And you know, but obviously.
- 3:15:02You almost need the audience validation when they're on
- 3:15:04stage, right? And they get off the stage and
- 3:15:06you're like, now you've got to adapt to kind of a different
- 3:15:08audience environment, right? And I've, and I've hung out with
- 3:15:11a lot of comedians that, like, practice their routines on you
- 3:15:14all the time, you know? But the pat, once you get past
- 3:15:17the routine, right, then what? Who are you?
- 3:15:20Like, the naked truth comes to the surface, right?
- 3:15:24Yeah. It's interesting that we're
- 3:15:25talking about. This but we we we can do a
- 3:15:26longer podcast about about this and I could definitely go into
- 3:15:29more stories. Then I ended up selling that
- 3:15:31asset back in 2010 thousand 11 and you know I ended up working
- 3:15:34for a company called ad.net, which basically, you know, we
- 3:15:37isolated pain point Google was monopolistic so we built an
- 3:15:41incremental search marketplace, kind of ran sales and marketing
- 3:15:43there for about 14 years yeah sold, you know, portion of the
- 3:15:47company, you know and. Then an incremental search
- 3:15:48marketplace. So tell me, tell me how that was
- 3:15:50so. Yeah, so, so basically we put
- 3:15:52search results on websites. Search Yeah, you know, so and
- 3:15:55you know, brands really wanted to diversify because, you know.
- 3:15:57And you're syndicating the the Google search.
- 3:15:59Our own, you know, we built a, built a direct business.
- 3:16:01Yeah, yeah. So it was a because.
- 3:16:03It was a pain point that I had find a problem like.
- 3:16:06Research like classic. Yeah.
- 3:16:07You know, you put ads on like little listings.
- 3:16:09Yeah. So on, you know, within, you
- 3:16:11know, publisher ecosystems. Yeah.
- 3:16:12So I ended up selling that and we ended up selling a portion of
- 3:16:14that entity. And then, you know, I came
- 3:16:16aboard Nomics Group and saw what Nomics was Group was doing
- 3:16:18holistically. Yeah.
- 3:16:19You know, find a problem, solve the problem, scale the business
- 3:16:22and sell it. You know, it's been my
- 3:16:23trajectory, all three companies I've been at and, you know, kind
- 3:16:27of seeing what Nomics was doing. You know, Nomics is a holding
- 3:16:30company. Nomics started about 3 years
- 3:16:31ago. Yeah, as Shop Nomics, right and
- 3:16:34company scales significantly, Shopnomics is in the performance
- 3:16:37marketing standpoint works with some of the biggest brands in
- 3:16:39the world, you know, and kind of get some increased distribution.
- 3:16:42And over the past couple years through acquisitions, you know,
- 3:16:45Nomics, you know, built the system commerce everywhere
- 3:16:48solving the problem of fragmentation.
- 3:16:50And what that really means is that, you know, where do users
- 3:16:53interact And that leads us to some of the the acquisitions
- 3:16:56we've done. You know, we have a company
- 3:16:57called Creator Nomics where we have access to 400,000 creators
- 3:17:00and we also hard launch today at at possible Bannomics, which is
- 3:17:04our AI driven content studio. So creators can be both real and
- 3:17:07synthetic. Yeah.
- 3:17:08And so that's entirely synthetic.
- 3:17:09That's that's creating creators that fit 100 or 1000 different
- 3:17:13molds. Is that is that?
- 3:17:14Yeah, I mean, you had it's in essence, when you build a
- 3:17:16technology piece, you have, you know, you queue the system,
- 3:17:20right? So let's say you want to have
- 3:17:21take, for example, you know, we're here in Miami.
- 3:17:24Yeah, let's say a hurricane's barreling down on Miami, right.
- 3:17:30Demand immediately spikes for things like generators.
- 3:17:32Yeah, water, plywood. OK, so let's say a brand wants
- 3:17:36to capture that demand real time.
- 3:17:38Yeah, We can go from media brief to creative development to
- 3:17:43execution to optimization and scale within a matter of
- 3:17:47minutes, right through our fanomics technology.
- 3:17:51Yeah. And then we have the performance
- 3:17:52infrastructure through our Shopnomics team, you know, as
- 3:17:55well as our you know, creator team to activate cross channel
- 3:17:59creator economy as well as kind of traditional media's, you know
- 3:18:03like a New York Times or media activation.
- 3:18:05Well, so, So what is the fundamental technology behind
- 3:18:07phenomics? Because that's, that's where
- 3:18:08like, because you just, you made it like a cognitive leap there
- 3:18:11that I'm like, I thought phenomics were synthetic
- 3:18:13creators. It's our, it's our technology
- 3:18:16that we've built and we actually.
- 3:18:17And what is a synthetic creator? Let's define that.
- 3:18:20Is that. Is that an individual?
- 3:18:21Like, I've got a profile, I'm a person, I've got an appearance,
- 3:18:25I've got likes and dislikes, right?
- 3:18:26You've read a good point. Let's say, for example, you're a
- 3:18:29product. You know Home Depot wants to
- 3:18:31reach men in Miami that are home owners in a certain demographic.
- 3:18:35We can tune the engine to spit out the synth, the AI creator,
- 3:18:39within those demographic profiles.
- 3:18:41Yeah, so and so this would be someone that matches both kind
- 3:18:44of behaviorally, demographically, the audience
- 3:18:47you're trying to target imitates that audience, talks to the
- 3:18:51audience like hi, I'm Jesse, you know, you're, you're And so so
- 3:18:55because I was trying to make that connection between how
- 3:18:57would that work in a, in the case of of a hurricane, yeah,
- 3:19:01hitting, you know, and people in, in there's a demand, there's
- 3:19:03a drive in demand due to this. This you can have qualified
- 3:19:06people. Both real creators can activate
- 3:19:08which takes a little. Bit of time giving advice yeah
- 3:19:10and then how how what's the distribution mechanism behind so
- 3:19:13you got this thing it's trained it's got some sort of memory
- 3:19:16system behind it that gives it human like quality so talk to me
- 3:19:20about that we've built where you building that we've.
- 3:19:21Built the distribution platform. We're connected to every single,
- 3:19:24you know, we're connected as we were talking about earlier
- 3:19:26before the podcast, AI agents, we're connected to all media
- 3:19:29platforms. So we distribute it on our own,
- 3:19:32you know, our own kind of media execution.
- 3:19:34But how do you make how do you make the synthetic persona We
- 3:19:37right What? What's the Intel behind that?
- 3:19:39Think about a media brief. Yeah, think about when agents
- 3:19:41will do a media brief. We basically take the media
- 3:19:44brief and put it into our platform and then it spits out,
- 3:19:47you know, these are the core functions on the media brief.
- 3:19:49So think about it. People have to hire creative
- 3:19:53agency studios and then when the cost of production is
- 3:19:57compressed, let's say there is an example for any media
- 3:20:00execution take take Home Depot. Yeah.
- 3:20:02Say they have a media execution of $1,000,000.
- 3:20:05OK, they're going to have to allocate X amount to creative
- 3:20:07production, X account to media, X account to, you know, staff or
- 3:20:10whatever agency fees they have. Let's say we're able to compress
- 3:20:14that cost to pennies on the dollar for the creative
- 3:20:17execution. If there's still $100,000, where
- 3:20:20does that go? The money is going to go into
- 3:20:22performance marketing, so you're going to see more brands
- 3:20:25reinvesting dollars. Think, think about.
- 3:20:28It is at the budget you're cornering, Yeah, right.
- 3:20:30And and performance marketing. And so the asset you're
- 3:20:32delivering to Home Depot is we're saying, hey, Home Depot,
- 3:20:36we're going to give you a synthetic agent, right?
- 3:20:39That, that, that they could, could, could fit a couple
- 3:20:42different flavors. Maybe there's a, there's a host
- 3:20:43of them. They're all they've got avatars,
- 3:20:46they've got personalities. We've somehow trained a memory
- 3:20:49system so that they can deliver probably some conversational
- 3:20:52interface. You can talk and chat with them.
- 3:20:54They're giving advice to a specific audience around the
- 3:20:57specific problem to buy a certain product to meet a need.
- 3:21:01Like like an example of Home Depot, like you need to prepare
- 3:21:05for coming, you know, and it's done in a more human delivery
- 3:21:08model versus just a, a targeted ad, a targeted banner ad.
- 3:21:13Greater velocity is a competitive advantage, yeah.
- 3:21:15Because you think about the problem that brands are facing
- 3:21:17right now, right? Fragmentation.
- 3:21:20You know, people can start the consumer journey on TikTok, They
- 3:21:23can start it on search, They can start in a conversational agent
- 3:21:27Fragmentation, right? The media is more fragmented
- 3:21:29than ever. One click does no longer tell
- 3:21:31the story anymore 2. Speed, if you don't capture that
- 3:21:34demand, your competitors are going to yeah yeah 3 you know
- 3:21:38you need to be able to perform we've moved to an outcome based
- 3:21:40economy everything new has a performance aspect behind it
- 3:21:43yeah you know vanity's gone it's all about performance yeah and
- 3:21:46when the dollars. And you're humanizing that, that
- 3:21:48performance, right, as opposed to the old world of sort of
- 3:21:50direct response marketing where it's like buy now, get the Super
- 3:21:53shammy and let the time offer, right?
- 3:21:55A lot of that was meant to drive action, whether it was late
- 3:21:58night television or it was an ad or a retargeted, you know, or an
- 3:22:01e-mail that you're getting that's that's trying to get you
- 3:22:03to commit some sort of action. This is saying, no, we're going
- 3:22:06to actually, because to me, that's the unique value.
- 3:22:08So, so like, like all this other stuff, to me, it's like, hey,
- 3:22:10you're having a human or a simulated human, right?
- 3:22:14Which is either a creator that's part of your creator network or
- 3:22:17simulated human delivering A humanistic message tied to a
- 3:22:21very specific problem that they're facing at a, at a point
- 3:22:24in time. And you're you're zeroing in on
- 3:22:26budget that is ordinarily just dedicated to performance
- 3:22:30marketing, old performance marketing.
- 3:22:32So it's almost like the next generation of performance
- 3:22:34marketing. And so to me that's really
- 3:22:36interesting. And what are the like I want to
- 3:22:37know what the distribution mechanisms are for like agent a
- 3:22:42attacking the hurricane problem in Miami, right?
- 3:22:46How it for on behalf of Home Depot.
- 3:22:48That's the use case here. How are you is that going
- 3:22:50through TikTok video? Is it going through short form
- 3:22:53videos? How is it being?
- 3:22:54Delayed all of the above. So I'm glad you kind of use a
- 3:22:56late night TV example. You know, we can bring back
- 3:22:58Billy Mays, we can bring back to Sham wow guy, we can bring back,
- 3:23:02you know, the Clapper. So what gave rise to performance
- 3:23:07marketing? I mean, what cave rise to late
- 3:23:09night television? OK.
- 3:23:11Cost of creative was cheap. Yeah.
- 3:23:12Yeah. Late night TV spots were cheap.
- 3:23:14OK. So wall traditional, you know,
- 3:23:17Super Bowl ads or, you know, ads during, you know, with the next
- 3:23:21game and they lose to a team that's trying to lose the Hawks.
- 3:23:24You know, don't take a breath. You want a Laker fan like myself
- 3:23:27so we can talk about that. It'll be fun.
- 3:23:30You have LA, New York, you have New York and Boston.
- 3:23:31It'd be fun, you know, So you. Want the better Better.
- 3:23:34You want the better ratings, you know nobody wants.
- 3:23:36Nobody wants. No offense to any of our
- 3:23:38listeners in Atlanta, Nobody wants.
- 3:23:39To see the hawk. You said it, not me, Brett.
- 3:23:42So, so you know, think about the rise of it.
- 3:23:45There's two cost savings, right? All right, so let's say there's
- 3:23:47always a fine amount of dollars. Dollars just move one way or the
- 3:23:50other. You know, literally Billy Mays
- 3:23:52or the sham Wow guy showing a product.
- 3:23:54You'd have to do big creative production, too.
- 3:23:57TV spots are cheap late at night.
- 3:23:58Yeah, you know. OK, so think about if we can
- 3:24:01minimize those costs like QVC, same thing.
- 3:24:04Think about minimize those costs by creative with our AI creative
- 3:24:09studio as well as you know, we built the distribution
- 3:24:11infrastructure with every platform, you know, almost every
- 3:24:14platform online. So we have so much buying power.
- 3:24:17So we help these brands save costs.
- 3:24:18So we're kind of taking the performance model to the I, I
- 3:24:21could, most people are looking at, can I reach these creators?
- 3:24:24Do they want to promote my product?
- 3:24:26Can I put this product in? It's very read and react, not
- 3:24:30looking at what the future of the creator economy can be.
- 3:24:34And ultimately too, then you're probably going to ask me, are
- 3:24:36the creators concerned that there's potentially synthetic
- 3:24:40creators that could take their jobs?
- 3:24:41Yeah. Is that disintermediating the
- 3:24:42actual human creators? Right.
- 3:24:44And the answer is actually going to be No, the answer is actually
- 3:24:46going to be it's going to help creators.
- 3:24:48And I'll tell you why. Let's say you're a creator.
- 3:24:50You can only. What's your most valuable?
- 3:24:51What do you want to do? You want to create?
- 3:24:53Yeah. There's only 24 hours in the
- 3:24:54day. Yeah.
- 3:24:55Your most valuable commodity is time.
- 3:24:56Exactly so. We can help creators actually do
- 3:24:59more. So it's like, OK, we're going to
- 3:25:00your likeness, your partners, you're engaged with us.
- 3:25:02We can help the creators even deploy.
- 3:25:04Think about that 3 or 4 dimensions down the line where
- 3:25:07this is an asset to ultimately creators as well as brands for
- 3:25:10cost savings. Is that in production right now?
- 3:25:13Do you have are you actually taking that 400,000 person
- 3:25:15critter We can networks you have right so and and giving them
- 3:25:18synthetic extensions. Buildings, it's not it.
- 3:25:21It's not far off for us to do. Yeah, we can.
- 3:25:23We have the core technology to. Do it said like is said like
- 3:25:26Chief revenue officer, right? No, it's not Live Today, but but
- 3:25:31it could be. No, no.
- 3:25:32No, I mean, so, so for us nomics itself, it's really nomics is
- 3:25:35the economics of shopping. Yeah, right.
- 3:25:37And think about what's happening right now in America starting to
- 3:25:40happen. It's kind of more like what's
- 3:25:41been happening happening in China, you know, Bing, you know,
- 3:25:43TikTok is live shopping. Think about if you're able to
- 3:25:46interact with a creator right then and there and purchase, you
- 3:25:49know, and then ultimately, like, you know, creators are the, I
- 3:25:52mean, we're talking. Drop ship.
- 3:25:54They will drop ship this shit direct directly from the
- 3:25:56warehouse like the DO model. Right.
- 3:25:58Do your children trust creators more than they trust the New
- 3:26:00York? Times, Yeah, yeah.
- 3:26:01And we did, we did talk about this.
- 3:26:02Yeah, they're, well, they're on. They're certainly on TikTok and
- 3:26:05when I do hear an occasional Candace Owen quote, I fall off
- 3:26:09my my chair and I'm like, do you realize what you're listening
- 3:26:12to? And please interrogate those
- 3:26:15messages because it may not be the truth.
- 3:26:17But but that's the type of stuff that's seeping into the minds of
- 3:26:21non political young kids and teenagers and college kids,
- 3:26:25because that's those are the creators.
- 3:26:26They're the people that are, you know, creating sound bites,
- 3:26:29reading click bait, right? You know, magnifying messages in
- 3:26:33these kind of quick form, short form.
- 3:26:35Is CNN not clickbait? Yeah, it's it's certainly not.
- 3:26:38I don't, you know, I mean, CNN is that clickbait in the sense
- 3:26:41of of, you know, the always on Trump controversy or the always
- 3:26:45on war footage, right. There's a click fetishness to
- 3:26:49that because it's all like, hey, we we're attracted to the
- 3:26:52negative news. But it's, I don't think from a
- 3:26:54younger generation perspective, it's nearly as engaging as, to
- 3:26:56your point, the short form video.
- 3:26:58What we'll think about, I mean like any, any media ecosystem
- 3:27:03survives and thrives when you build a community.
- 3:27:06Yeah. Right.
- 3:27:07And it's really all we're solving for.
- 3:27:08There's really, you know, there's really no new business
- 3:27:12models. Brands want to reach consumers
- 3:27:14the most effective manner. Publishers want to maximize
- 3:27:16their eyeballs in revenue. Yeah, there's different mouse
- 3:27:18traps to get there. That's what being in the space
- 3:27:20for close to 30 years has taught me.
- 3:27:22You know, right now think about who the trusted sources.
- 3:27:26How do you build community? Right?
- 3:27:27OK, maybe a Generation Z is building community through
- 3:27:30TikTok. Maybe, you know, people like,
- 3:27:32you know, old like us will build community.
- 3:27:34Going to watch the sports game, Talking about the Knicks,
- 3:27:36talking about basketball, Talking about 1973.
- 3:27:38Holding on to that Myspace. Account holding on to that
- 3:27:41Myspace. I don't know.
- 3:27:42Did Tom call you back? He didn't call me back.
- 3:27:44So, you know, you think about that and you know, it's not,
- 3:27:47I've learned to kind of put judgement aside and just be
- 3:27:50objective on these things. You know, if the children relate
- 3:27:54to it, great. And that's really the dollars
- 3:27:55are shifting that. I mean, think about what the
- 3:27:57creator economy is doing. You know, if by the whole live
- 3:28:00shopping experience, what are you taking dollars away from?
- 3:28:03It's moving away from traditional going to the mall
- 3:28:06because you can literally try things on.
- 3:28:08You can have stuff like right then and there and you know,
- 3:28:11creators of their trusted source.
- 3:28:13This creator is going to create this.
- 3:28:14And then ultimately, you know, having that platform
- 3:28:17holistically to help amplify creators, help scale creators
- 3:28:20and just literally take care of the creators partners that you
- 3:28:23have because they are the publishers.
- 3:28:24Yeah, Yeah. My last organization, we did
- 3:28:26some analysis was a company called Media Radar that did
- 3:28:28media intelligence. We did some analysis on time
- 3:28:30spent in various media channels and we, and we looked at sort of
- 3:28:34direct live sports appearances, right?
- 3:28:36A Hollywood economy, meaning production produced content, not
- 3:28:41always long form, but more often than not long form.
- 3:28:44And then we looked at like creator economy, which was short
- 3:28:46form video. And there's been in the last
- 3:28:48five to 10 years, there's been a, a huge shift from, from
- 3:28:52production to create or in terms of total time spent, total
- 3:28:56audience engagement, the, the, the live sports environments
- 3:28:59and, and live event environments have pretty much stayed the
- 3:29:02same. So that seems to be, it's like
- 3:29:03about 20%. But I think there was a shift
- 3:29:06that, that I think about 70%. Sixty to 70% of total time spent
- 3:29:09is now being spent in creator oriented, whether it's YouTube,
- 3:29:14Tick Tock, etc. Right.
- 3:29:16So it's content, right? They're even taking longer form
- 3:29:19content and breaking it down into what do they call those?
- 3:29:22The vertical soap opera? Yeah, media, you know, I'm
- 3:29:25talking about where they take like a long form media.
- 3:29:28It's an hour long episode and they break it down to the three
- 3:29:30minute increments and you have to, it's in a gamer fight.
- 3:29:34So you've got to, you've got to, you get points for going on to
- 3:29:37the next three minute episode, right?
- 3:29:39And it's always a cliffhanger. And they end up changing the
- 3:29:42monetization model for long form content because that one hour
- 3:29:47episode that you would normally pay, let's say 399 to download,
- 3:29:50they're now getting $20 because they've been able to monetize
- 3:29:54along the entire path and get people sort of addicted to every
- 3:29:57next chapter of this basically hour long episode, right?
- 3:30:00So it seems like everything's moving towards that short form.
- 3:30:03Do you think that that, like, do you think that is a positive
- 3:30:07sign from a social and societal perspective?
- 3:30:10Do you think we're losing the ability to, do you think it's
- 3:30:13like mass attention deficit disorder?
- 3:30:15I think it's kind of my question like are we losing the ability
- 3:30:18to focus on longer form deeper narrative content cuz we're so
- 3:30:22addicted to short form instant gratification type of content?
- 3:30:27We're getting philosophical on that.
- 3:30:29Yeah, I figured we'd go to that. So it's neither good nor bad?
- 3:30:34Yeah, we don't answer that. Question.
- 3:30:36Don't. Judge Yeah.
- 3:30:37So it's just the way people are living in the world that we
- 3:30:40exist in today, you know, and, you know, us as people can make
- 3:30:44the decision. OK do we have a safe space where
- 3:30:46we. OK, you know, maybe you have a
- 3:30:47family dinner. Yeah.
- 3:30:48You know, things like that. So.
- 3:30:50So you compartmentalize. Yeah, exactly.
- 3:30:52You have to compartmentalize everything.
- 3:30:54I mean, you know, look, we have and a lot of this as a culture,
- 3:30:57this does our own doing. You know, I mean, think about it
- 3:31:00when you know, the three channel, I mean, I grew up, you
- 3:31:01know, it was like you have three channels.
- 3:31:03Maybe Fox got introduced at some point, maybe there's 4.
- 3:31:06But you know, it's, it's an aspect of culture.
- 3:31:08You know, when you have 1000 options, you know, it does
- 3:31:11become overwhelming. You know, people say like, you
- 3:31:14know, hey, like the DVRS, the big, I mean, I think the remote
- 3:31:16controls the biggest like invention because it's like you
- 3:31:18have to get up and physically go switch the channel and turn the
- 3:31:20rabbit here. So it's like, I can, if I don't
- 3:31:22like it, I can move it right now.
- 3:31:24And, you know, usually all this stuff is like, you know, work,
- 3:31:28you know, it's just, it's not good nor bad.
- 3:31:30It's just the way our world commits.
- 3:31:31Yeah. But I think from a communication
- 3:31:33standpoint, it forces us to be clear and direct.
- 3:31:36And you think about it, you have, you know, people, the long
- 3:31:38winded stories like just give me the results, give me the
- 3:31:40numbers. So while we're saying we're in
- 3:31:42attention, like if you want to call it attention deficit, this,
- 3:31:45I'm not going to disagree with you.
- 3:31:47But what I'm also saying is that.
- 3:31:49You know, quickly. Well, yeah, it forces us to be
- 3:31:51concise. Yeah, right.
- 3:31:52It forces concision, but they're the signal and the noise.
- 3:31:55See, I bring it back to you guys.
- 3:31:57So, so it forces us a culture to do that.
- 3:32:01And then I'm a Leo, so I'm the internal optimist.
- 3:32:04I look at it, what does this allow us to do if we're not
- 3:32:07spending time on things that don't matter?
- 3:32:09It frees it up to do things that do matter.
- 3:32:11And that's why I look at what's happening with AI, You know, it
- 3:32:15frees us up to do the things that that will matter, have
- 3:32:18conversations with good people, brainstorm, spend more time with
- 3:32:21our family. So like, you know, think about
- 3:32:23it, you know, if you were to kind of look at.
- 3:32:25Like get the grunt work. Yeah, get the busy work.
- 3:32:28Done. If you were to think about,
- 3:32:29like, industrialization, you know, the coal miners, you know,
- 3:32:33kind of like went in and said, OK, you know, hey, this
- 3:32:36industrialization is going to take my jobs.
- 3:32:38I'm going to lose my job in the coal mine.
- 3:32:39Yes, but it will save you emphysema.
- 3:32:41Yeah, but I'll have my job. You know, you don't see the
- 3:32:43horses complaining when cars came.
- 3:32:45You know, the people that fight innovation are usually, you
- 3:32:49know, fear, love. You know, you can already say
- 3:32:51that fear is more powerful than love.
- 3:32:53I'm afraid of this. We're saying, OK, what is AI
- 3:32:56going to do for me? It can help amplify my message.
- 3:32:59It can get the right message to the right person.
- 3:33:01It can even free agents can help me free up my time so I can
- 3:33:04either A, be more productive or B, spend more time doing things
- 3:33:08that matter and not doing the rote tasks.
- 3:33:11You know, like how often if you look at your flights to
- 3:33:14potentially change seats. What if an agent does that for
- 3:33:17you? Yeah, you don't have to spend
- 3:33:18time literally clicking on your exactly.
- 3:33:19And it's sort of that abundance mindset mindset versus the
- 3:33:22scarcity mindset. Exactly.
- 3:33:24It's, it's this like, how do you take this as an opportunity to,
- 3:33:27to create sort of a better world as opposed to pushing back to
- 3:33:29see this is wrong? Because we've, and it's hard
- 3:33:31when you're in it, when, when we're living in it, it's hard to
- 3:33:34kind of pull out and see what's actually happening at a macro
- 3:33:37level. But back to the creator economy.
- 3:33:39So I want to understand too and get your thoughts on on sort of
- 3:33:42how creator networks have grown what what percentage of media
- 3:33:47investment because you're saying it's sort of the new publisher
- 3:33:49work. The the amount of time spent on
- 3:33:51it's an under invest friend of mine and kind of restoring this
- 3:33:54like it's a under investment based on like the amount of time
- 3:33:59spent there is, you know, it's ultimately there is a dollars
- 3:34:03from an agency holdco and brand perspective.
- 3:34:06Take a while to shift. Yeah, because I mean, I used to
- 3:34:08say nobody is fired for buy more Google.
- 3:34:10Yeah, I mean, that's really it, right?
- 3:34:12Yeah, I'm doing what I did and it's more safe.
- 3:34:14You know I'm choosing an example.
- 3:34:15I don't want creator specific agencies.
- 3:34:16Yeah, you guys played that role. Are you giving a marketplace?
- 3:34:19Yeah, more marketplace. You know, I like the the the,
- 3:34:23you know, you think about any new innovations always quote UN
- 3:34:26quote, the bastard stepchild of any organization.
- 3:34:28Yeah. So anytime you bring in a new
- 3:34:30piece, what happens, oh, if there's a finite dollars against
- 3:34:33something, say $1,000,000, if where, if new dollars come in,
- 3:34:37where's it going to pull? So people are motivated by fear.
- 3:34:40And that's part of the problem that we, you know, obviously,
- 3:34:42you know, where channels, you know, people focused on channels
- 3:34:45and saying my channel did bet did it better than this because
- 3:34:48they're afraid of showing things that not work for just looking
- 3:34:50at a holistic perspective. First, if you have ACMO of, you
- 3:34:53know, Coca-Cola, they just care about Coca-Cola brand, you know,
- 3:34:56they don't care about does this channel do better than that?
- 3:34:58They don't they don't care, you know, they care about their
- 3:35:01brand. And so the, the issue is not
- 3:35:04the, the fact that critters here, the issue is the execution
- 3:35:08and fear to kind of follow the money.
- 3:35:11These are short term fears. You know, we know that the
- 3:35:14eyeballs are here. The dollars will follow.
- 3:35:16You know, I, I, I learned life. You can't time when things are
- 3:35:18going to happen. It's not about being right.
- 3:35:19It's about being right at the right time.
- 3:35:20Yeah. And and so do we think, do we
- 3:35:21think that creative creator advertising.
- 3:35:23So it's it seems like you guys are are.
- 3:35:25It will you're. Skimming off the performance
- 3:35:27marketing. Box, yeah, currently, you know,
- 3:35:29but there's, you know, ultimately creators engage more
- 3:35:32mid funnel, upper funnel, yeah, you know, the lower funnel is
- 3:35:34you know, it's, it's a place to start.
- 3:35:36But as as the business goes up, you know more brands, you know
- 3:35:40more agencies get more comfortable with what creators
- 3:35:43are doing, the dollars will continue to follow.
- 3:35:45Yeah. And you have some sort of Rev
- 3:35:46share agreement where you say, hey, we're going to we're going
- 3:35:48to put you on our network, you're going to be promoted, you
- 3:35:50know, like how does how does your creator?
- 3:35:52Yeah, I mean, I mean. Like AG2 crowd for creators,
- 3:35:55right, where there's rankings and reviews and, and, and so you
- 3:35:59know, topics that you can pick like it, you know, created,
- 3:36:02there's a subject matter expert in this or this or this, you
- 3:36:04know, so you can start to align them.
- 3:36:06That's where. The infrastructure is going to
- 3:36:07have to go. You're going to have to kind of
- 3:36:08mirror the existing models. I mean, like it comes back to
- 3:36:11kind of what I said before, you know, there's just be
- 3:36:13mousetraps, the same business models, you know, you're going
- 3:36:15to have to mirror. To your point, these are some of
- 3:36:17them. And what we've done at Nomics is
- 3:36:18build the distribution infrastructure and the layer
- 3:36:21behind it to be able to track performance.
- 3:36:22Because think about it, you know, the, the, the value is
- 3:36:25really an infrastructure. And then when the infrastructure
- 3:36:28is set, because the same amount of work to Coca, let's go back
- 3:36:30to Coca-Cola to onboard something for, you know, $5000
- 3:36:34or 5 million, It's really the same infrastructure.
- 3:36:36You have to be able to build scalable infrastructure in order
- 3:36:39to be able to plug in. So will the creator models
- 3:36:42evolve to where folks are used to buying?
- 3:36:44Of course, that becomes more easy to buy in the long run.
- 3:36:47And this is part of the reason why it's, you know, kind of what
- 3:36:49you said or one of us said, kind of the dollars will follow at
- 3:36:53some point when the infrastructure is fully scaled
- 3:36:55out for the creator economy. But, you know, if people aren't
- 3:36:57focused on their creator economy, I mean, that's really
- 3:36:59where the money is going to go. And it's really part of how we
- 3:37:02and we wanted to tackle it. Obviously the performance
- 3:37:04infrastructure, the creator relationships, as well as, oh,
- 3:37:07by the way, we have synthetic creators which can save cost
- 3:37:10savings to Coca-Cola millions of dollars to, you know, a few $100
- 3:37:15and then that money can be reinvested to actually driving
- 3:37:18performance as the Glutter performance economy goes up.
- 3:37:21Yeah, and. And so it's outcome based
- 3:37:23economies. Yeah.
- 3:37:23And are you, are you partnering with agencies that, that, that
- 3:37:27connect creators to brands or and just providing the
- 3:37:30infrastructure underneath it or are you actually playing that
- 3:37:33role of, of the middleman between brands and then creators
- 3:37:37they want to align themselves to?
- 3:37:39Yeah, we're and we're a platform, you know, we're common
- 3:37:41server platform. We're not an agency, you know,
- 3:37:43so in essence, you know, an Omnicom or a, you know, a
- 3:37:46publicist would be a partner of ours.
- 3:37:49Hey, how are you doing? We can help you reach this.
- 3:37:50What's your strategic adaptive with the Phanomics piece?
- 3:37:53What's your media brief? We're going to put our media
- 3:37:55brief in our technology creatives for you.
- 3:37:57Yeah, yeah. And then you're charging a
- 3:37:59subscription type of. Yeah.
- 3:38:00I mean there's, there's multiple pricing models.
- 3:38:02It depends. I mean we have, you know, since
- 3:38:03we have obviously the technology piece for the fanomics, yes, we
- 3:38:06have different models. But if you combine it with
- 3:38:08immediate, with an execution against our platform, you know,
- 3:38:11with flexible pricing models, you know, we have an SMB
- 3:38:13strategy, we have a, you know, premium strategy.
- 3:38:16Yeah. And and are you finding that a
- 3:38:17lot of the investment is coming from SM BS in mid markets or do
- 3:38:20you see that enterprises are really moving heavily into this
- 3:38:22space the the creator space? I think it's moving a lot
- 3:38:26faster. If you asked me three months
- 3:38:27ago, things are never as fast as you want them to be or never as
- 3:38:31slow as you know as they really are.
- 3:38:33So it can always move faster, but the pace is escalating
- 3:38:37significantly and it comes to having conversations.
- 3:38:39I mean, someone as I presented a possible today and people are
- 3:38:41like, wow, wow. So a lot of folks, you know
- 3:38:44what, you'll see a lot of these ad tech companies have a salt
- 3:38:46lick. They're trying to look at one
- 3:38:48piece of a pie versus understanding how the whole
- 3:38:51ecosystem works. That's why we started, you know,
- 3:38:53the Nomics group. It's the economics of shopping,
- 3:38:56you know, and we look at economics.
- 3:38:57It's the outcome base. If I were to walk into a brand
- 3:38:59and say, you know what, we're going to ensure we're going to
- 3:39:01guarantee performance for you in a while, how you going to do it?
- 3:39:04You know, what problem are you solving?
- 3:39:06Fragmentation, how we get commerce everywhere, all
- 3:39:08surfaces, AI agents, right? We're able to solve that problem
- 3:39:12through our portfolio of products and services.
- 3:39:14And that's why we've we've had significant growth, literally
- 3:39:16one of the fastest growing private companies based on kind
- 3:39:19of what we've done. And it's been, you know,
- 3:39:21stitching it all together, you know, be able to kind of like
- 3:39:25put the tools in the hands of the partners, you know, and just
- 3:39:27saying here, you want to be, where are the people to go to
- 3:39:29when you want to activate at scale Coca-Cola, you know, for
- 3:39:32your entire catalog of every product, Do it right away.
- 3:39:35Yeah. Well, that's great.
- 3:39:36Hey, well, thanks for joining us.
- 3:39:37This was, this was a good conversation.
- 3:39:39And you know, so we haven't talked to anybody about sort of
- 3:39:41the creator kind of specifically and how and how it's operating
- 3:39:43in today's sort of media and marketing.
- 3:39:45Happy to come back anytime I can give you stories about comedy
- 3:39:48and comedians anytime you want. And you know, we can have we can
- 3:39:50have a good laugh. Yeah, well, enjoy the rest of
- 3:39:52signal and Noise and and thanks man.
- 3:39:53It was great meeting you. Thank you, Brett.