The Operating System for Commerce Media: Jeffrey Cohen on Amazon Ads' Rise, Retail Media's Measurement Reckoning, and the Shift from Dashboards to Agents
Retail media was built on a powerful promise: connect advertising directly to transactions and finally show marketers what their media dollars produced. But as the category matures, a harder question is replacing simple attribution: Did the advertising actually cause the sale?
In this episode of Signal & Noise, Rio Longacre and Brett House sit down with Jeffrey Cohen, Chief Business Development Officer at Skai and former Principal Evangelist at Amazon Ads, for a wide-ranging conversation about the evolution of commerce media—and what happens as intelligent agents begin taking over work once performed through dashboards, spreadsheets, rules engines, and manual campaign management.
Jeff had a front-row seat during Amazon Ads’ extraordinary rise. He explains what Amazon got right, from the simplicity of “search, find, buy” to its ability to connect media exposure with actual commerce. But from outside Amazon, he now sees the larger challenge: consumers don’t live inside walled gardens. TikTok can drive an Amazon purchase. Prime Video can influence a Walmart sale. In-store activity affects digital behavior, and digital media influences what happens on the shelf.
That creates a measurement reckoning. ROAS and last-touch attribution can tell marketers what received credit for a transaction, but not necessarily what caused it. The conversation digs into incrementality, cross-channel effects, independent measurement, and why historically siloed media, shopper marketing, trade promotion, and commerce teams increasingly need to operate together.
Then the discussion moves from dashboards to agents. Jeff explains how Skai is approaching this transition through Celeste, Skai Studio, MCP connectivity, and agentic workflows. Instead of simply asking an AI assistant for an insight, marketers can build systems capable of analyzing signals, finding audiences, recommending budgets, creating campaigns, monitoring performance, and escalating decisions to humans for approval.
The goal, Jeff argues, shouldn't be another marketing black box, but a glass box: automation where marketers can understand the logic behind decisions, establish guardrails, audit actions, and retain control over critical choices—especially where budgets are concerned.
One particularly striking finding: Jeff says 45% of the questions asked through Celeste could not have been answered manually within the traditional Skai platform. Agents aren't simply making analytics faster; they're combining data and performing analysis that previously required people to jump between systems, download reports, manipulate data, and assemble the answer themselves.
The implications extend far beyond media buying. Jeff describes Skai's own company-wide agentic transformation and the emergence of a new kind of marketer: the builder—someone who may not be a traditional software engineer but understands systems, processes, data, business rules, and how to turn them into intelligent workflows.
In this episode:
How Amazon Ads became an advertising giant
Why “search, find, buy” proved so powerful
Attribution vs. true incrementality
Why platforms can end up “grading their own homework”
The convergence of retail media, CTV, social, and in-store
Celeste, Skai Studio, MCP, and agentic workflows
Why agents need guardrails, auditability, and human oversight
Why major budget decisions still require human approval
How AI changes campaign management and optimization
Why AI transformation is really operating-model transformation
The rise of the marketer as a builder
Why companies need to rethink workflows instead of simply automating them
The big question is no longer whether AI can optimize another campaign or generate another dashboard. It’s what decisions we’re prepared to delegate, what governance those systems require, and where human judgment remains essential.
Hosted by Rio Longacre and Brett House on Signal & Noise.
#RetailMedia #CommerceMedia #AmazonAds #AdTech #MarketingTechnology #AgenticAI #MarketingAI #RetailMediaNetworks #Incrementality #Advertising #MediaBuying #Skai #DigitalAdvertising #SignalAndNoise
Read the full transcript below.
Brett (00:00.878)
Hey everybody, welcome back to Signal and Noise. I am Brett House, joined by Rio Longanker, my co-host, and Jeff Cohen, the Chief Business Development Officer at Sky. Welcome to the show, Jeff. I know you go way back with Rio, and very good to meet you. And we know a lot of the same people. so media's a small world, and then you get into Chicago and it gets even smaller, right? So you're calling in from Chicago and
Rio Longacre (00:21.32)
Media is a small world, that is for sure.
Jeff Cohen (00:22.968)
It is. It is.
Brett (00:28.85)
and for those that don't know Sky, it was founded in 2006 as Kinshu, who I certainly was very familiar with in Tel Aviv. I came from Excelate, which was also a Tel Aviv-based company. and it was founded by a couple of guys named Joab Itzak, near Cohen, and Alone Schaefer. and it was really focused on search and social campaign management, if I remember correctly, right? and then you guys acquired signal analytics.
Jeff Cohen (00:53.23)
Yep. Yep.
Brett (00:57.388)
Which was more of a market intelligence and AI company and pivoted into insights versus just media execution. do correct me if I'm wrong. And then recently, or not rel relatively recently, in the last few years, rebranded to Sky, which I did read was a kind of a a mashup of the different names, right? So Signal Analytics and then the K from Kinshu and it came up came up with Sky. It's it's a good catchy name. and now you guys are sort of positioned as a full go to market intelligence platform.
That's pushed heavy into omnichannel commerce and retail media. increasingly Yeah, increasingly into AI Gin AI G and Gin AI with the launch release, I think it was in 2022, maybe 2023, of CelesteAI, which is described as an AI native, I hate to say it, operating system. there's there's a lot of operating
Rio Longacre (01:31.274)
Yeah, big retail media push, commerce media push, yep.
Jeff Cohen (01:34.085)
Yeah.
Jeff Cohen (01:48.112)
Hehehe.
Rio Longacre (01:49.782)
It's a new it's it's a new it's a new buzzword, everyone has an operating, but yeah. Yeah.
Brett (01:51.563)
Yeah, and the OS is the new term. but you guys have 1,500 users, work across 200 different organizations, so it sounds like there's some some success there. And you, your background, you came from both seller labs and importantly helped build or or or played a big role in Amazon ads as a principal evangelist, right? And that's a business that's over 70 billion dollars at this point. So that's a super interesting
Rio Longacre (02:18.358)
Yeah, and Jeff was there during that meteoric growth too. So that's something we definitely want to dig into as it went from people don't realize Amazon's but you it's it's almost seventy billion now and it it's been growing like
Jeff Cohen (02:27.332)
Yeah, think it was a baby. I think it was like it was under 20 when I started that, right? Which yeah, I mean. Yeah.
Brett (02:31.55)
R really? Yeah, so so you saw that's talk about wind in the sails. Right? That's a good time to join, right? I mean yeah, I'm sure you had something to do with it, but you wasn't just Yeah. And for those that don't know, I mean most of our audience does. Amazon ads is sponsored ads, display video, their DSP, prime video ads, and then all their measurement solutions. I think. I think that captures it all. Yes
Jeff Cohen (02:38.554)
Yeah. It was, mean, I was there and it grew. So yeah, you know.
Rio Longacre (02:57.098)
Yeah, most of our audience knows advertising well. But for those who don't, when you get recommended products, or you see any ads on Amazon.com, like that that is the Amazon ads business, which is seventy billion dollars, highly profitable. It's a big part of the kind of AWS like machine now when you look at it. And in terms of bottom line,
Brett (03:00.983)
Yeah.
Brett (03:14.55)
Yeah. Jeff, we're not du Yeah, we're not dumbing it down like you know, ridiculously here be but there are times, you know, we have listeners that aren't from the Martech out tech space that oftentimes go, you know, Rio's had a couple conversations where they're like, I had no idea what you're talking about. I don't know if it was your dentist or your your lawyer.
Rio Longacre (03:30.282)
No, that's funny. My bar my barber. We were talking about we're my barber was asking about the podcast and then I said I I texted him the link and then he happened to listen to the Adam Heimlich show, which was like very technical and containerized, bidding algorithms, and he was like, dude, I I did not understand anything. So
Brett (03:40.206)
Yeah, we're talking about like containerized DSPs.
Jeff Cohen (03:42.512)
You
Brett (03:48.141)
He just turned it off immediately. I'm like, well, apparently not our target audience. So we try to broaden the topic. But so we're here to talk about a bunch of these things, right? Like you know, what you learned at Amazon, what it looks like, how how the how the how your world looks a little different outside of that ecosystem, how AI is changing work and and retail and commerce media and what's happening in that space. So Jeff, welcome to the show.
Jeff Cohen (03:48.356)
That's alright.
Jeff Cohen (04:11.322)
Thanks, what a great intro. Appreciate that. mean, Rio and I have been talking about finding time to do this for a long time. I really wanted to wait until we had some of our operating system AI in the market as opposed to in its closed beta. So I could actually talk about it in more depth, because I know it's stuff that you guys would be very interested in and excited to dive into these topics.
Brett (04:36.333)
Yeah, yeah.
Brett (04:40.364)
And sometimes it takes a little w sometimes it takes a little while to understand what the product team is truly building, right? As they're in prototype mode, right? You're really
Rio Longacre (04:40.775)
Oko
Jeff Cohen (04:46.094)
Yeah.
Rio Longacre (04:48.597)
And to get them to prioritize the right things, right? And like like push through the right features, ship the right features at the right you know, first and in the right order. That that can take a little doing too from. So well Jeff, we're thrilled to have you on. Like like you said, we try to make this we talked about making this happen a while. In fact, I remember you were when you were with Amazon, we talked about a little bit. Amazon can be really tough to get these types of approvals. I know we're you were discussing it, then you moved a sky and then we wanted to, but yeah, I th I think it's definitely gonna be worth the wait when looking at all the stuff we really want to talk about.
Jeff Cohen (04:51.226)
Yes.
Rio Longacre (05:17.287)
And I think the timing is good for this. And it's our hypothesis for this episode that retail media, now that it's become such a big part of digital media itself, it's entering a new phase, right? The first era was defined by the meteoric growth of Amazon, right? Walmart and a few others, like giving people, connecting advertisers directly to transactions, which was something that was really kind of novel, right? Like that I think retail and commerce media did better than really did bet better than any of these other channels, right?
But it's our belief that the next era is going to be defined by really proving out incrementality, right? Showing, okay, is this is the spend worth the squeeze? Is the juice worth the squeeze? Are these ads that we're placing, are they driving incremental like transactions that that people wouldn't have made? So I think I think we're kind of entering that. And plus the fragmentation, I think this is where Sky really comes in, right? There's so many hundreds of retail and commerce media sites. Everyone's been building over the last few years to kind of the figuring out.
How do these this fragmentation like what does it mean for marketers? What does it mean for media buyers? And what does it mean for measurement and actually planning and activating ads? So I think Sky's in a really great position and we're keen to dig in here. So maybe we start this off like maybe going a couple of years back. Like we met when you were at Amazon, did a lot of work together. Like when you left Amazon, what did you see about retail media that made you think like like that made you really excited about the opportunity at Sky?
Jeff Cohen (06:37.52)
Yeah, mean, listen, I think at the end of the day, I I saw what Sky was building when we, when, when I guess at that point, when they released Celeste. So they released their, their AI tool, Celeste in 2025. Maybe.
Brett (06:53.442)
I was I was a little early. Twenty twenty five, it's very recent.
Jeff Cohen (06:54.864)
Yeah, maybe late 24, was maybe in beta 25, it went into public. And when I sat down with the product team, it was more about the vision of what they were trying to build and the problem that they were trying to solve. at the end of the day, I think we all have kind of a really simple problem that we're trying to solve that's really complex, which is what do I do today? Where do I spend my energy and my effort? And what we do is we sift
through reports and charts and graphs and try to figure out how we can move the needle, whether that's gaining market share, whether that's improving margin, whatever it is that we're trying to do. And the promise of AI, the ability to be able to do this in the type of scale that AI can do it in today gets us so much closer to being able to have that answer. And we're able to do things that we never were able to
Brett (07:51.628)
Yeah.
Jeff Cohen (07:54.69)
to do before. And when I was watching kind of the market evolve, I just kind of felt like it was the right time. Right. I said to myself that I would leave Amazon whenever I felt there was a great opportunity that was out there. And I felt that Sky was a well positioned company working with enterprise brands, working on enterprise solutions, very, you know, technical technology first. Right. A.I. driven. And it just seemed like the right right place.
to be and the next move to make.
Brett (08:27.745)
Yeah, no for sure. So so so what did you from from the Amazon experience, what did you learn about sort of commerce, advertising, simplicity, like what are you seeing differently now that you're part of Sky? or or or as I'm sorry, as a result of that experience that sort of led you to this this decision?
Jeff Cohen (08:44.184)
Yeah, it's a good question. What I see now, like...
When I was at Amazon, maybe had, I was a homer, right? Ria, you've heard me. Like I would always say I was an Amazon homer. I was a tech evangelist for Amazon. I was supposed to be a homer, but I ate my own dog food, right? Like I believed in the Amazon product. I still believe in the Amazon product, right? I still believe that Amazon has one of the best products that's out there. They're able to give you some great information back, but I maybe was a little bit more naive to the larger market around
Brett (08:58.251)
Yeah.
Jeff Cohen (09:19.442)
around retail media and all the other retail media networks that were out there. And so maybe like what was maybe simplistic, right, about being at Amazon was I was in a hedge garden who had end-to-end tracking and a maturity and built a maturity for the products that they had. And the market around it was still kind of growing up. And so I would say like they didn't have a lot of competition in those early days.
because they were so advanced in what they were doing. And maybe now that I'm outside, I maybe see, and I was maybe a little naive to people on the outside impact of things. And now on the outside, I definitely see the stronger impact that advertising on TikTok and how it drives sales to Amazon or advertising on Amazon and Prime Video and how that's driving your sales to Walmart, that maybe was a view that I didn't have as strong of an opinion on when I
had more of a myopic Amazon view.
Brett (10:22.197)
Yeah, yeah. And f and for those that you don't know, and but before you ask your your question, Rio, the the the total retail and commerce media market, which I've done a ton of research on in the past when I was at Media Radar, and relatively recently is about a hundred and eighty billion. but it's a it's a duopoly, right, more or less, with eighty percent of that total market share. I think th these numbers still hold up. Could yeah, Amazon it will no controlled by Amazon and Walmart, but you're saying Amazon adds as seventy five percent? I thought it was
Rio Longacre (10:42.631)
Seventy five about seventy five percent is Amazon ads.
Rio Longacre (10:50.461)
O of of retail media. But you add commerce media, the number gets closer to what you're quoting at, Brett. It comes becomes closer to like Yeah.
Brett (10:52.053)
Yeah. Yeah. Yeah and and commerce media is about ten billion, right? And that's that's all of the non-retail sort of thing, travel, fintech, right, and other inter intermediaries that are controlling.
Rio Longacre (11:06.569)
Like United Airlines has kinetic, which is their like commerce media network, Chase has one, a lot of the PayPal has one. So there's a lot of these ad networks that are related to different types of business different types of publishers, different types of of businesses across different industries, Brett. Totally. But you look at just core retail media, I think Amazon, last time I checked, it was could have shifted a little bit, was around seventy-five-ish percent, give or take. And it Yeah, but which which is a lot. So but when looking at that, Jeff I the one critique
Brett (11:14.124)
Yeah.
Brett (11:19.874)
Yeah.
Brett (11:25.229)
Y yeah, yeah. S some something somewhere around seventy, yep.
Rio Longacre (11:33.541)
like you that you hear about retail media is these ads are coming up, it's a tax on these brands, people would have bought the stuff anyway. Like w when you hear that, I'm sure you have a lot over the years, especially when you work at Amazon, like how do you respond to that?
Jeff Cohen (11:48.058)
Well, the first thing I always...
like to remind people of is that there's a tax and an advertisement on marketing and promotion wherever you go. You don't have a right to these things just because you're there. Right. And so even the while I'm in Chicago the local hot dog stand down the street from me if it wants to get people in the door it's got to find ways to advertise. Right. It's just it's the way that it is. So I think that when you start with simplicity right. Amazon started out as a very
Brett (12:11.776)
Yeah.
Jeff Cohen (12:20.11)
simple model of search, buy. And in that search, find, buy, a lot of brands got access to a lot of traffic that maybe helped drive and propel their business. And so there's early, there's early brands who were emerging brands who were able to really grow and build market share on Amazon because they were really just there before anybody else was. And Amazon went to go and start monetizing that traffic.
make sense that they monetize that traffic. And that allowed for new brands to who weren't maybe getting discovered through the old model to begin to get discovered. And the fact that at that point in time, we really only had one way to measure it, which was last touch attribution, because there were no other attribution models. There were no other ad types that even needed to be accounted for. We maybe over inflated the impact that those advertisers
advertisements had. I remember the early days where you had ad campaigns with 40-50 % return on, you know, return on ad spend, like crazy numbers because it was last touch attribution and you weren't, you weren't doing...
Rio Longacre (13:30.152)
Right.
Brett (13:32.448)
Yeah. And now the mar and the market's now insisting on incremental ROAS, right? I R O A S, right, versus just pure ROAS.
Jeff Cohen (13:36.984)
Right. Yeah.
Yeah, and it's just a growth. It's a growth of the growth and the maturity. There's now a lot of ad types that exist, right? Amazon went from sponsored products to sponsored brands to sponsor display to the DSP and then a whole bunch of offshoots of those. And once you started having multiple ad types, now all of a sudden you have to find a way to give each ad type its credit on its impact. And advertisers are now wanting to ask that that better question, which
is how much incrementality am I actually getting from this ad spend versus if I was in isolation, would that buyer have bought from me anyway?
Brett (14:21.579)
Real
Rio Longacre (14:21.705)
Yeah, that's it's a good point, by the way. So so like for those who don't under like who haven't worked in retail media, endemic advertisers, the people who have their products, let's say, listed in Amazon, they can run ads. Those are with through Amazon ads, and to Jeff's point, there are different types of ad inventory and creative they can push up there. And then being able to tie those ad exposures, again, it could be a sponsor brand listing, could be it could be Amazon Prime Video. There's different type the like on tying those
add exposures back to a transaction. And what but what's great about Amazon, if you're endemic, the transactions take place in Amazon, right? So it's really easy to track. What have you seen for non-endemics? I know that's been a big area of growth as well. I mean, are you can and what are you seeing? Is that is that still an area of growth and like have you seen things like conversion API or different techniques to actually have that same measurability?
Jeff Cohen (15:08.036)
Well, know, conversion API or CAPI has become kind of an industry norm that's been built out now across, you know, all of the retailers with the explicit goal of trying to provide this value. The non-endemic advertisers is a huge avenue for growth. The biggest avenue that you saw in the media, the media, the media or, wow.
Brett (15:35.754)
You got it in meteoric.
Jeff Cohen (15:37.678)
The fast rise of Amazon ads is the acquisition and growth of Prime Video, right? So as Prime Video bought MGM and started expanding their properties, as they went into Thursday Night Football and then expansion and live sports, all of a sudden, the types of advertisers that would advertise were a different mix of advertisers because these are the brand advertisers, right? The Jake from State Farm,
Brett (15:39.34)
Ha ha ha ha.
Rio Longacre (15:52.181)
That was huge.
Jeff Cohen (16:07.602)
who are used to advertising in these types of media and taking the reach and frequency model and driving it down to a customer acquisition cost. so Amazon was one of the first to kind of go into that arena, which opened them up to a lot of these non-endemic advertisers. And I think it clearly...
Brett (16:26.536)
Yeah. And it's Prime Video, it's Fire TV, it's Amazon DSV, all of those things kind of cater to that because they're going off site now. It's sort it suddenly starts to look like kind of a I I'd hate to say a DSP, but it looks like almost like a non-pure retail play, right? Because you're sudden suddenly appearing to audiences that aren't just in a retail environment, to to drive brand awareness to your point. and and Walmart did the same thing with their acquisition of Visio.
Jeff Cohen (16:31.023)
Yes.
Brett (16:54.296)
with their partnership with Roku, which is now owned by Fox. Right. So you see a lot of retailers and re and and think of Walmart as trying to win share of wallet from their brand advertisers that that are endemic that are actually advertising in store. They're both placing product on shelves, they're running in caps and in store sales. And for them it's an opportunity to sort of say, how do we actually get more money out of our existing clients' pockets by you know, which is I think to your point, Rio
part of the tax argument, right? To say, hey, isn't this just retail? Isn't this just an expansion? If I'm an advertiser that's already had a has a huge paid relationship with Walmart for both product placement in the store, but also for all the advertising I do in store. And then I move into the digital ecosystem, aren't they just getting more of my dollars within a contained ecosystem, whether it's digital or physical? Right? And and so there's people in the in the space that argued that that's just retail.
That's the this whole notion and and how much real incremental customer acquisition are you really doing with those investments? I mean do you think that that's a a fair argument to make around the tax issue?
Jeff Cohen (18:06.84)
Well, again, I think everyone's going to make arguments and there's two sides to every argument. But I think the other at the end of the at the end of the day, the retailer has to feel comfortable about the growth of their brand. Right. Retail is a tough margin game.
Brett (18:22.281)
Yeah, yeah, that's a good point. I mean two to five like there's super small margins. Yeah. And then once you once you go into the digital ecosystem your margins might go up to thirty to fifty percent. Something like that.
Rio Longacre (18:25.749)
Two to five percent typically, yeah. Very small.
Jeff Cohen (18:32.174)
Yes, so, listen, with Vizio and on TV, the number of televisions that now Walmart has the ability to place ads into has grown significantly. They just bought Vibe, which is giving them a lot of inventory in the CTV space. And so...
Brett (18:47.422)
Yeah.
Jeff Cohen (18:51.44)
This is the new frontier of where advertising's occurring. My wife is a huge tennis fan. We've been watching the tennis channel the last couple of weeks for a couple of tournaments, mainly because they're all actually on US time, because they've been in the US or Canada. just watching the ads that come through there is very fascinating. It's a lot of D2C brands with QR codes for buying.
connected through our Roku or through our Fire TV. It's a lot of pharma, right? And that's not what those ads looked like three to five years ago. The ads that were on the fast TV channels and those smaller, I call Tennis TV a smaller channel. It's not getting the same kind of contracts that like an ESPN is getting, but they're getting big national brands that are looking to expand their
Brett (19:25.6)
Yeah.
Rio Longacre (19:33.833)
Yes, changed, yeah.
Jeff Cohen (19:51.374)
their market share.
Brett (19:52.746)
Yeah.
Rio Longacre (19:53.651)
Yeah, Brett, I thought you brought up an interesting point, Jeff. I'd love to get your take on it too. I've done some recent work with with some big retailers and grocers and specifically on retail media and monetization. And like we started a couple times we had to r start digging deep into the in-store element, right? And we actually had to work with the teams running running all the trade promotions, the end caps. I mean, it's I I I've never up until recently, I did yeah. I didn't have too much. This is my first time. I'd done personally a lot of work with.
Brett (20:10.24)
Yeah.
Brett (20:16.521)
The white glove mafia. That's what they call them in the industry.
Rio Longacre (20:22.143)
with him. I thought it's fascinating. It's a whole other business, right? So like I I'd love to hear your take on that. Like how is are some of those dollars starting to migrate into more reach more like the like the retail commerce media? How is it adapting to that, Jeff? Love your take on it.
Jeff Cohen (20:36.464)
Well, so first off, a lot of those dollars are controlled by joint business plans, right? So they're not necessarily shifting, but what's happening that's changing is that the way that either the agency or the brand is starting to look at them is to understand the impact that one has on the other. And so a year or two years ago, 10 years in dog years, is that
Brett (20:40.832)
Yeah.
Brett (20:57.514)
Yeah.
Jeff Cohen (21:06.66)
Those were very siloed organizations. And what you're seeing today is that they're now either rolling up to a similar VP or an SVP that's now asking those questions to understand what's that total cost? What's that total impact? What is my in-store impact on online and my online impact in store? And was kind of what I was mentioning before. We're now realizing that people are watching, they're in a Walmart store and they're buying from Amazon.
They're sitting on Amazon and buying from a Walmart shelf, right? And they're doing the same at Kroger and Dollar General and everywhere else. And so you can't negate the impact of in-store because no matter how big we want to think retail media has become, in-store is still significantly larger.
Brett (22:00.361)
Yeah. No no for sure. And and
Rio Longacre (22:00.604)
That's a good point.
Brett (22:04.191)
Well you w you wonder if if like it that they we've been talking about that for a long time. I meant the the combination of digital plus and store and whether you're driving some sort of incremental action by the consumer that who wouldn't otherwise purchase just because they're within that store environment. you know, so I w I wonder if that's the the pr this kinda ties to that that that notion of incrementality and how effective it is in driving incremental sales. are you guys getting to some of insights that kind of show
the the impact of of both in store and off in in in on and online kind of combining to drive
Jeff Cohen (22:39.952)
So I don't know about the in-store and online, but I do know that we are, I know like a case study that was recently shared internally where a company was trying to understand their.
their impact of their TikTok social campaigns on their retail media. And we were able to run in a regression analysis to understand by isolating the campaign dates and the glance views of their page, what the impact of that was and come back to them with a scientifically backed answer as to your TikTok campaigns are increasing your views on Amazon by X, which then allows them to understand like what the value of
their social influencer campaigns are so that they can determine where they want their investments to be. And the challenge is, is that for the brand, the social investment is being done by one team. The influencer investment is being done by one team and the retail media is being done by one team. But it goes back to what I said earlier. They collectively as a company are now looking at that holistically and asking the questions. And so when you can bring all of those data sources together into one application and then you're able to
ask it intelligent questions using an AI that can produce those kind of results, you can get a lot deeper into that information to know what the impact is.
Brett (24:03.657)
Yeah.
Rio Longacre (24:03.774)
Jeff, I was going ask you a question about is each of media continuing to grow because some of the, let's say, the in-store and other marketing dollars is starting to migrate there. But it sounds like you're saying it's growing because there's a like the the two are not being seen in isolation anymore. People are beginning to look at it more holistically. Marketers and brands are in agencies too, because you're right. They're up until recently, those are totally different teams. And I am starting to see them collapse together. So I guess, so would it be fair to say that?
They're growing because we're able to prove that effectiveness and those th th those two silos are now starting to be viewed more as one and the impact of one, how it affects the other. Fair to say.
Jeff Cohen (24:43.514)
The short answer, yes. The more complex answer is there's leaders and there's laggers. And leaders are the organizations who...
have made those organizational changes and are doing that and they're probably gaining market share advantage and being able to do that ahead of others. And then laggers are those who are maybe slow in making that adoption and need to speed it up. And so if you're looking at like best practices, ways you should be thinking about building your business in the future, new ways of looking and slicing and dicing your data, it's definitely questions you should be asking.
to see if it's right for you. But I've never been a believer that one system applies to all. And so each people at each organization kind of needs to figure out what works for them. But I am seeing more and more of it. And even a year or two years ago when I was on stage at Amazon Unboxed interviewing some companies, there were a lot fewer companies who had that organizational structure set up that way. And today we're seeing more and more companies setting that organizational
structure up and also creating kind of their own centralized way of looking at data as opposed to just using the isolated data that's coming in and asking those questions. What is the impact? And one of the hardest parts of AI, and I'm sure we'll get into more AI in this conversation, but one of the hardest parts of AI that we've seen is that people don't know what to ask it. And so you have to have that data foundation layer that's AI ready.
And that's, think, what Sky is really great at doing. And then you have to be able to ask it the right questions to get back the answers that can impact your business.
Brett (26:33.746)
Yeah, and you made some good points. I mean it's a part of its data congruence, right? So having and and then it's also organizational alignment to your point around the the offline retail folks versus the media folks, right? If they're all o if they're all kind of fighting over a dollar of an investment, and back when I was at Nielsen, the argument was that s on the buy side of Nielsen, that seventy five cents of every dollar was spent on in store shopper marketing. Twenty five cents was what the entire
media spend was being was be what was being dedicated to all media channels, right? And so the entire advertising ecosystem was arguing over the twenty five cents of every dollar spent. But if you can get all those people in the same room with an underlying data layer that that sort of helps them decide against and and aligned incentives. Yeah.
Rio Longacre (27:22.203)
And Brett, you could prove that incrementality. Like where's the money actually driving incremental value? Yeah, makes sense.
Brett (27:26.856)
Yeah, and and that that to me seems to create a lot of an alignment within these organizations that didn't previously exist, right? And everybody was
Jeff Cohen (27:32.856)
Yeah, because you have teams who had KPI metrics that were based on like reach and frequency, right? And as long as you hit enough people at enough times, you met your KPIs and your goals. And then you had other teams that needed to have, you know, ROI and conversion. Well, right.
Brett (27:50.282)
Yeah, and so they're trying to claim credit, right, for for the purchases or for the ad exposures.
Jeff Cohen (27:53.74)
It doesn't it doesn't work. Those numbers don't work. And so you have to have it's got to start at the top and the top has to be asking the right questions and enabling them to work themselves down. At the end of the day, if you want scale, you have to have region frequency, right? It doesn't work. You will limit the size of your business if you're only trying to do things based on a cost per customer acquisition or or a two or three X row as right. You won't get to the growth level that you need, which
Brett (27:59.402)
Yeah.
Jeff Cohen (28:23.674)
is what we know about traditional brand advertising. Listen, I have a friend with a business back in my hometown. He still does billboards and he believes billboards work, right? So there is no one answer for any one business, but it's a matter of the metrics that you're applying and how you're looking at it and judging its impact across the board of everything that you're doing.
Brett (28:34.729)
Yeah.
Rio Longacre (28:48.787)
Yeah, it's a good point. So we did a bunch of incrementality studies. I remember when Jeff we were working together back when we were doing with like some early AMC work, we'd for endemic advertisers is pretty cut and dried, but for non-endemics using conversion API or at least getting hold of first party data, doing uploads, and looking at the match and seeing who actually who actually placed those orders. and and almost every time we were able to prove out that the ads were highly effective, that they were driving incremental value, and they certainly the juice is worth the squeeze. I think we'd see a drop-off sometimes after like nine or ten ad exposures.
Right, but generally speaking, like you know, the retail media spend was was was certainly worth every penny 'cause there'll always be return and out spent. You mentioned
Brett (29:18.43)
Yeah.
Brett (29:24.904)
And you had controls to that type of analysis, right? Yeah, you know, where you would have an exposed versus unexposed audience and you'd have complete Yep.
Rio Longacre (29:30.846)
Yeah, we'd have a holdout group. We'd we we'd we'd we'd ru we'd run the test, we'd look to see okay, like how many incremental like transactions is this is this actually driving. We'd look within specific groups and we compare it. And obviously like ad spend like within with like for for certain look like they're usually perform lot better, depend depending on how you model those out. So those could those could be quite effective. But Jeff, you'd mentioned AI. One criticism of AI when I've heard it recently is like is I think this applies a lot to Google, right? Is
Brett (29:46.76)
Yeah.
Rio Longacre (29:58.634)
they're rolling out these AI models, maybe Meta too, right? But all they end up doing is creating a black box that ends up favoring favoring the house, right? That ends up favoring the the walled garden or the hedged garden or the platform, right? So I don't think that's necessarily the case, but that's something I hear quite a bit. And Google recently rolled out lot of AI features in in Search Console that I think actually do exactly that. So thoughts on that? I'd and then I'd love to hear about what this guy's doing.
Jeff Cohen (30:22.448)
I like the joke, it's the AI max, the P max, everyone kind of came out with their own plus max version. We've integrated in with all of them across Microsoft, Google, Amazon, and many, many more. Listen, at the end of the day, if I put my old Amazon hat on, nobody was more intelligent.
Brett (30:30.76)
Yeah.
Jeff Cohen (30:48.464)
and had more data to properly run studies for MMM or MTA than Amazon.
Now, if I put my non-Amazon hat on, I go, yes, but they're grading their own homework and they have a bias for their market. I've never been a conspiracy theorist. I don't believe that they are inherently trying to push dollars to them because if it doesn't work for the advertiser at the end of the day, they're not going to continue to retain advertising dollars. But they may be looking at things within a siloed organization that's not giving credit to outside forces.
as strongly as an independent study would.
what we recommend, Whether you're looking to expand into like AI Max or PMAX with Google, for those that don't know that that's exactly what Rio was talking about, where you're basically giving it dollars and you're letting it kind of run the AI campaigns. We believe you run tests and those campaigns against your manually managed campaigns need to perform at or better than how they were performing before.
What we found is that for most of our customers, obviously it's not at every campaign level, but for most of our customers, they perform pretty well.
Jeff Cohen (32:12.462)
What you don't know is how much incrementality you're getting and to what degree that you're getting. And so we believe that you should be doing and continuing to do both. I wouldn't put all my eggs in one basket of AI or not. Now, one of the challenges you have in the market is that whether you're talking about Amazon and wanting to show up in Alexa shopping, whether you're talking about Walmart and wanting to show up in Sparky or we'll say Google and you want to show up in Gemini is that if you're not doing their performance plus business
plus, you know, AI, Max Plus, that's how you show up in those kind of LLM ads, right? There is no LLM ad type per se. You're showing up in them through the performance-based advertising. So you kind of need to participate in those because that's where the market and the traffic is shifting to. And if you want to be part of those and you want your ads showing up in those, you have to be running those ad types.
Rio Longacre (33:12.487)
Yes, Fuddy, you mentioned that conspiracy. I I'm not a conspiracy theorist either, but I was working with a client about a year ago and they were running this is an auto OEM and they were running some campaigns in Meta and they were targeting specifically people who had purchased a vehicle and they're they were targeting them on Meta to sign up for like oil changes, different service appointments. And then I remember at one point the client showed me the Advantage Plus box had checked magically, right? And then the Audix extensions had checked magically.
And they had gone from like whatever, like the small number. They were uploading a list, doing a match on Meta, and targeting us that specific audience. Suddenly it was like 10X that their CPMs actually went down. Their spend didn't adjust too much, but they started mark blowing their budget on complete randos. They didn't want to market to actually like weren't eligible for the offer. So there was actually some legal and compliance risk. I j so I you know, we had they shut it down immediately when they found it out. But that went on for a couple of weeks. I thought it was wild that these AI like black boxes.
Be careful, I guess it would be my my advice to marketers.
Jeff Cohen (34:12.302)
Yeah, listen, mean, you, you there, there's, there's not a, we believe it's hard to say I don't believe in AI automation because
we've all been using AI automation and advertising for long before workflows and, and, gen AI was a thing, right? Rule-based advertising is automation. You're giving it a rule and when a rule applies, you're having it taken action. That's an automation. So I don't want to blanketly say that that's a bad thing because I actually think it's a very good thing, but you have to have proper guardrails and you have to have proper, you know, review processes. And so like when,
we built our agentic workflow tool that we call data, that we call Sky Studio, it actually has like checks and balances built into it that kick it out for human review, right? And so if you can't have those kinds of systems and processes in place, then you do have more of a black box than you have a glass box. But when you, when you could put those rules in place or where you can use a third party like Sky that can be kind of looking
for those types of anomalies, it can create a glass box around that black box that ensures that your business is operating the way that your business is intending to operate.
Brett (35:36.349)
Got it. So can you walk us through like an end to end example of how Celeste AI works from an agentic workflow perspective?
Jeff Cohen (35:44.494)
Yeah, Celeste by itself is, I've been making this argument internally to stop calling it Celeste, because.
Celeste is our AI that powers our platform, right? It allows you to ask any question, get detailed analysis, take action, but then similar to like a Claude co-work, you can schedule activities and you can kind of build skills and knowledge based to them. Our agentic workflow tool, which is Studio, is our newest application.
Brett (35:56.702)
Yeah.
Jeff Cohen (36:23.438)
And that is one that is building like a full blown like call it SOP operating process and and the two somewhat work in tandem and so when you're in our platform you would just be able to ask any type of question about like building a Monday morning report building a regression analysis trying to identify where additional spend could go those are all things that are pretty standard for our Celeste AI to be working but if
you wanted something that had multi-layer processes built into it. Like, let's say you wanted to build a DSP agent, right? And the DSP agent was going to run analysis, look for new audiences, find audiences that you could potentially advertise against, make budget recommendations, send those budget recommendations to a human to review. The human then says yes, and then it goes,
back and creates the campaign. It then sends it to you for flight review. You check the flights to make sure that they're what you want. You say yes. You then go tell it to execute and then you set it up for whatever type of reporting you want. And then it's doing the bid adjustments and the management adjustments. That's how a full workflow could potentially operate versus Celeste built into our platform, which is more of like a day to day I'm in the platform.
Brett (37:45.277)
Yeah.
Jeff Cohen (37:53.138)
asking questions or scheduling reports or scheduling.
Rio Longacre (37:57.34)
So it's more like giving you an assistant that can support different functionality that or just exist natively in the platform. It's not like a headless version of your platform where to your point it's like it's orchestrating all of these things.
Jeff Cohen (38:07.576)
Yeah, so we actually, I don't want to confuse the situation, but we have both, right? And so Celeste is the assistant that's working with you. Studio is the tool that you can build with. And so if you're a builder, right, if you go into cloud cowork and you like to build workflows, that's where you're using Studio. you're just a data.
Brett (38:23.069)
Yeah.
Rio Longacre (38:31.111)
So Celeste is more like Rufus, like to look at the Amazon ads analogy, right? Just okay. Okay.
Jeff Cohen (38:34.51)
Yeah, you're looking at yes. And we have pre-built templates and we have, but we also have a third one, which is the ability to connect via MCP to your own LLM if you're building outside of Sky and you want to use Sky as a build partner.
Brett (38:50.67)
Well, can you use it as a build partner to build within your own native environment without data movement coming out of let's say it's an Azure, an AWS, a Google?
Jeff Cohen (38:59.758)
Yeah.
So that would be the studio application if you actually wanted to do the full build where you're, and we have FDEs, Forward Development Engineers, who come in and work with you to build your orchestration layer. But that's where the operating system, you joked about the term operating system, that's where the Sky operating system comes in because you're actually building it to the way your business operates. And so our data layer makes it to where you're conversing with our data.
Brett (39:09.863)
Yeah.
Jeff Cohen (39:30.418)
in a way that is familiar to your organization, how you do packaging, how you do case counts, how you do specific terms and terminology. And that makes everything AI ready for whichever path you want to go. Because what we found from our data, and we surveyed around 350 or so people within the industry, is that the industry is ready.
but the teams don't feel ready. Right? So the willingness to want to do things is very high. Like 80 % of teams are interested in heading in this direction, but only 35 % of teams feel that they have the skills necessary to be able to do it. So there's a lot of desire to want to do it today, but either you don't feel the data is ready, they don't feel that their team is ready, they don't feel they have the right skill set to be able to do it.
Brett (40:03.227)
Yeah.
Brett (40:23.175)
Yeah.
Jeff Cohen (40:30.012)
We're building the tools and the advisory services to be able to help enable.
Brett (40:34.608)
And that that seems like a common play, whether you're an independent agency or a vendor. I mean I've read facts some numbers out there. One, most most organizations, I think it's upwards of eighty eight to ninety percent, are have adopted AI in some flavor.
Right, across all brands in North America and the US, but like one percent of C suite is feels that they're mature in their AI implementations on a on a on a yearly basis. You know, and that's you know, state of I reports and things like that. So there's clearly a gap between the use of AI and then the sort of advanced use cases they know it's capable of and sort of the promise of the industry that they don't feel they're quite there. And and and there's a lot of concern. We actually just had Eddie Drake from from
Rio Longacre (41:02.546)
Yeah.
Brett (41:17.704)
snowflake on our show and w there's a lot of concern around data sovereignty right and zero trust data not to not to belabor those points and sovereign AI meaning does a brand and let me ask you this does a brand that you work with have control over the agents that you're sort of implementing and building for them if you're in a build mode as well as the the the patterns that that are being developed as part of this process what the agents are learning to optimize
Rio Longacre (41:17.875)
Stuff like
Brett (41:47.5)
let's say their media spin in a particular retail media network. and the context around that, are are are you giving clients sort of complete sovereignty over that? Or is some of that information being going is going going back into your own platform? Yeah.
Rio Longacre (42:00.528)
Yeah, 'cause Brett, that could include bid management, that could include like big p b media strategy. That can include a lot of things. Yeah.
Brett (42:06.533)
Yeah, and so thematically, like th that's where the competitive advantage is for a brand, right? They're building out all this logic and for an agency.
Jeff Cohen (42:10.82)
Yes, and for an agency, right? So I would say like, the simple answer is the brand has complete control over that within our system. don't take any of our, we do have like generic workflows that can be applied and you can choose which ones you want. But when you as a brand are building anything, it's not being applied to the larger good. It's the same way that we've operated for the last 20 years. If you go into our system to build rule-based
Brett (42:34.673)
Yeah.
Jeff Cohen (42:40.754)
like we're not applying your rules to others that are in using our platform.
Brett (42:47.495)
You're you're not learning yeah, using it to train other services to build build product for yeah.
Jeff Cohen (42:51.312)
Correct, and we are training our servers on.
on contextually how marketers and how retail media networks work, but we're not training them on the data from our customers. And that's a very critical distinction. you know, being an enterprise solution, we have very strict governance contracts. mean, you guys know this from your past roles, like in what both what we can do from like the retail media networks with their data, as well as what we can do with our partners and their data. Right. And so
Brett (43:01.253)
Yeah.
Brett (43:04.646)
Yeah.
Jeff Cohen (43:25.65)
It's all very siloed. I'm probably going to describe this wrong and my IT team will get upset with me. But it's like it's like almost like everyone has like their own little versions in the that things are siloed within our system. But the overall context layer is what like our data science team and our our our Celeste team is building that then gets applied. But that's more of like what is an ad, what is an ad type, how does a campaign work?
Brett (43:37.457)
Yeah.
Jeff Cohen (43:55.682)
The things that are gonna make our system more intelligent than if you were to just go ask. Yeah.
Brett (43:57.787)
Yeah.
The basic governance, but is that stuff's a adjusted and optimized in a client's environment, right? Because it's gonna adapt yeah, it'll be specifically for them and it'll be cordoned off from from your basic set of governance and instructions.
Jeff Cohen (44:05.773)
specifically for them.
Jeff Cohen (44:10.468)
Wraps.
Rio Longacre (44:12.07)
But you have your state your basic context layer that you could apply universally to everyone, but then it it's adjusted or customized for each client. That makes sense, yeah. So Jeff so Jeff, looking at like some of the things you've built, right? Whether it's MCP servers or the fact you've built the studio, which really sounds like it's almost like a headless capability where people can build their own solutions, like you pulling data like like what using your MCPs in order to connect with your databases and your processes and everything that Sky Sky provides. That's that's that's pretty cool. But curious, because we had talked to
Jeff Cohen (44:17.55)
Yep. Yeah.
Brett (44:19.045)
Yeah. It yeah, these
Rio Longacre (44:40.818)
Brett, remember that we talked to at Ken, we talked to Sarah Robertson, who's had a product at experience, and she had told us we built these MCP servers, expecting people to start to use them, and apparently they're saying there's almost no usage so far. They think there will be at some point, but apparently she was saying so far there really hasn't been a lot. It's been a little underwhelming. What's been your experience? I don't know what you're able to disclose by the way, but like are these being used a lot? Are people jumping and testing them? Like, where is the state of adoption? How's that looking, Jeff?
Jeff Cohen (44:59.887)
Yeah.
Jeff Cohen (45:06.157)
So I don't remember the exact numbers of our MCP servers off the top of my head, but since we've released them, which has just been three months or so, maybe four.
Rio Longacre (45:16.528)
Recent. Okay.
Jeff Cohen (45:17.57)
Adoption has grown significantly and the number of pools is continuing to grow on a regular basis. A lot of what we're finding is that some of our partners are setting up their own, you know, Claude instance and they want to use us as the MCP because we're a much more efficient MCP than trying to pull from all the different retail media networks. The one that we do have more data on is like just the overall use of like our Celeste AI because it's already been in market now for a little over a year.
Brett (45:35.59)
Yeah.
Brett (45:38.939)
Yeah.
Jeff Cohen (45:47.426)
The adoption rate of that has gone up significantly.
And one of the data points that was shared with our team recently that I actually thought was really great, and this is like our IT team is kind of looks at this stuff on the back end, 45 % of the questions that are asked within Celeste could not have been answered manually within the Sky platform. So if we just look at the questions that are being asked and whether the actual platform of Sky could have answered them, 45 % of the questions being asked could not
have been answered and that means that the agents are able to pull from multiple data sources, run analysis that didn't exist before, create dashboard views that didn't exist before and drive more than just faster BI but a faster and a better interface for the customer. And so like I said,
Brett (46:42.543)
And and they're and they're pulling f that information from the customer zone stack. Right? It could be their ERP, their yeah, their their C R
Rio Longacre (46:48.934)
That's very cool.
Jeff Cohen (46:50.606)
Yeah, because if you think about it, like before you had to go to here and get this data and go here and get this data and here and go this data and then you had to like download it all, put it into Microsoft, do a, you know, a macro, pull it all together. Now you just ask one question and all that's done for you, right? And you can ask like, yeah.
Brett (47:06.203)
The the topic of a federated data ecosystem, right? You know, you centralize what you can and then in other cases you need to pull in other stuff that's not you know, going
Rio Longacre (47:09.703)
Yeah.
Rio Longacre (47:14.416)
Well well, Bright, remember Kyle Chick was complaining about Copilot, right? How it could like you you ask it for your inventory, you can't do that. Sorry, you cannot do that. Infrace is not available. So that's intri that's actually really very useful, Jeff, to the fact that this can actually hit those hit those other databases and create things. And forty five percent that's a big number. I'm not surprised, but that's that's really cool.
Jeff Cohen (47:32.878)
Yeah, you know, Brett, you kind of mentioned this before, like, and I use this analogy before, but it's sky like we're eating our own dog food, right? And so.
our CSDs are connected to our MCP. So when they're running analysis on the accounts that they manage, they're pulling it from our MCP in the same ways that our customers would pull it from the MCP. We're building our internal workflows using Studio. So we're actually a customer of our own product. We reorganized our company. I did a big LinkedIn post on this, but we did a full company reorganization.
Brett (48:04.069)
Yeah.
Jeff Cohen (48:11.954)
around agentic modeling. It started from top, right? You mentioned Joav, our founder, and Nir and Elan. They started by pushing down to everybody what the agentic modeling of the business was going to look like.
Brett (48:29.475)
Yep.
Jeff Cohen (48:30.532)
But that wasn't good enough. We then stopped our business for a whole week and we did an agentic transformation where we went bottom up and we literally were training people on terminology and how to build agents and what it meant. And we gave everybody access to co-work and we had people build Slack agents and Salesforce agents and you know, and then we did like a hackathon throughout the company to figure out how do we remodel the business and like,
give you another data point that was just shared with us. Before doing this, I think we did like 30 or 40 QBRs for our clients in a year. Last quarter, we did like 30 or 40 in one quarter, right? Because we're now using agentic models and workflows to improve our processes. And so as we as a business are becoming smarter about AI and the impact that it can have, we're able to be better at advising our clients on what they should be doing. And the biggest...
Brett (49:17.882)
Yeah.
Brett (49:28.548)
Yeah, and it and it kind of democratizes that knowledge. So it's not like you're transforming your business to be just a managed service. You're still sort of a tech software oriented business with the margin with the margins associated with that business. But everybody is so much more informed and and you know, at the use case level of how to do these things on their own.
Jeff Cohen (49:37.006)
We absolutely, yeah.
Brett (49:48.593)
That I'm sure they bring a lot of value to clients who tend to be on the unsophisticated side, not to the detriment of clients, but I'm hearing this across the board. Is that is that yeah, is that the clients tend to be a few steps behind the tech vendors, maybe some of the independent agencies, and there there's an education that is needed, but most companies aren't gonna switch their business model to say we're gonna go entirely manage services, throw a bunch of FTEs at this to to solve the problem. They're gonna need their kind
Jeff Cohen (49:55.822)
Right. Well.
Brett (50:18.566)
Of lean teams to be able to educate. And if they're not using the solutions themselves, how can they educate the client? Right?
Jeff Cohen (50:26.66)
Well, that's the biggest learning, right? The biggest learning is that if you want to implement an AI transformation in your business, it has to be top down and you have to be given the right tools to be able to do it and you have to set realistic expectations. Right? And so that's the biggest learning that we've had as an organization. And now we're advising a lot of our clients in how to do this too. And we're not a managed service. We don't want to be a managed service. We are a tech service.
Brett (50:36.612)
Yeah.
Brett (50:42.779)
Yeah.
Brett (50:55.396)
Yeah.
Jeff Cohen (50:56.594)
but we also understand that at the end of the day it's critical and it's important for us to have services to help enable the companies that we work with.
Brett (51:04.41)
Yeah. Yeah.
Rio Longacre (51:05.584)
Yeah, it's a good point you made about AI transformation really being org transformation, operating model transformation more than anything else. I mean, I think that's been a key takeaway from a lot of my clients. And I even look at, Brett, you remember the article we published a couple of weeks ago in Sigil and Noise, the rise of the marketing engineer becoming a new kind of role. And so cu looking at roles themselves, Jeff, as you've rolled out these AI features, whether it's Celeste, which is automating certain features, or even Studio where people are building kind of whole new kind of front ends and experiences, like
Are you seeing how your platforms being used change? Because you look at media buying and planning and execution, right? And the whole trafficking process has always been very manual over the years. Lots of people in platform, buttons, knobs, controls, for everything from pacing to bid management, it's it's very complex and very specialized. Are you seeing the types of users change? Are you seeing fewer users, more users, different levels, different roles? Like, what are you seeing in terms of how like platforms being used and adopted out in the field? Maybe it's too early, but just curious what you're what you can comment on that.
Jeff Cohen (52:05.646)
Yeah, so first off, thanks for that article. I shared that with my daughter, who's a communications major. And actually, she's a junior at The Ohio State University. She interned this summer with Havas. And I gave her that article and I said, hey, this is what if I believed everything that you wrote. And I said, if this is the industry you want to go into.
Brett (52:10.47)
The junior.
Brett (52:20.336)
Nice.
Jeff Cohen (52:31.876)
This is the skill you need to have if you wanna be successful. So I wholeheartedly agreed with your premise in that article.
Brett (52:35.301)
Yeah.
Did you pay her twenty bucks if she could make it to the end of that article? 'Cause it's like seven thousand words. No, it's it
Jeff Cohen (52:43.47)
No, I did was I turned it into a Google LLM and then I told it to rewrite. No, I had it rewrite the article and explain it to a college student who doesn't understand e-commerce.
Rio Longacre (52:50.214)
Yeah.
Brett (52:50.938)
He summarized it.
Brett (52:58.597)
that's really interesting.
Jeff Cohen (52:59.884)
It came up with great analogies and it was a 20 minute podcast for her to listen to. And the biggest learning from that, I'll get back to your answer here in a minute. The biggest learning from that was that she was like, wow, I should do more of this Google LLM for my classrooms. If I gave it all of my notes, I could literally create podcasts and I could go on my Saturday morning walk and I could digest all the information from this marketing class in a way that maybe I wasn't understanding it before.
Brett (53:05.339)
that's fantastic.
Rio Longacre (53:07.73)
Very cool.
Jeff Cohen (53:29.898)
She kind of had a little bit of an epiphany, which is great. We'll see how much of that applies during the semester. But to go answer your question, Rio, I think the truth of the matter is it's too early. That's just the honest truth. It's too early.
For our organization who started this in January, but really got into this in June, we're now in our 2027, just starting our 2027 planning. And we're asking really big questions based off of the fact that AI has taught our team to think beyond the bounds of how you had thought before. that is, so what I would say is that most people are still thinking of how do I have AI
Brett (54:08.41)
Yeah.
Jeff Cohen (54:15.41)
help make my job easier? How do I automate things I didn't like doing? How do I get reports that I need to run regularly? Like the real kind of basics of AI, very few have broken the bounds and said, if I can completely rewrite this, but I want to keep these governance rules in place and I want to keep these outputs in place, but I want to change the whole workflow, that I think is not happening just yet.
But that is the breakthrough that I think most brands are going to see that are going to take them to the next level in what they can do and where they'll find honestly the most value from a tool like Sky because those are the kind of breakthroughs where you're pulling a lot of data and a lot of org. And to give you an example, my team built a whole, my team runs all the partnerships throughout the company. So Google, Amazon, Walmart, Microsoft, and we
built a whole partner dashboard that can give us a inside view of how we're doing with all of our partners. And it's effectively a live QBR, right? So it's all the information we typically were pulling for a QBR, but I can pull it up at any point in time. It even does the like analysis. And when I asked my team, Hey, how much time is this saving you on your QBR? Cause they're now working on their Q2 QBR. said, well, it hasn't saved me any time, but the two hours that I would have normally spent
Brett (55:26.628)
Yeah.
Jeff Cohen (55:45.322)
all the data, I'm actually spend looking at the data and figuring out what's going on in the business and the direction that I want to go. So it's actually allowing them to be doing the stuff I want them to be doing as opposed to the stuff they were doing before. Yes. Yeah.
Brett (55:58.434)
Yep. It's raising the value of their their ability to deliver some yeah, the the value that they could deliver because they're going get getting quicker to insight because all that stuff is already captured. 'Cause yeah, QBRs take a ton of time to pull together to gather all this information, to put it into presentable format. And then you've done all that, now you gotta do the analysis. Now you've got to come out with the insights. Now you've got to present it to the board.
Jeff Cohen (56:16.388)
Yes. So what?
Rio Longacre (56:18.886)
Yeah. I I working I that's why I'm working more now than ever, right? Like my wife said, Why every night, like you why are you working at ten or eleven? Well, I'm actually I'm enjoying what I'm doing. The stuff I didn't like, AI's taking care of most of that, right? And I'm able to focus on like I'm writing more, I'm creating more really good content, like I'm I'm I'm much more productive at work, it's funny. Like so AI is doing a lot for us, but we're working more. I think that's the paradox of this whole thing, or the irony.
Jeff Cohen (56:42.928)
So our senior engineer on studio is a like 15 or 16 year veteran of Sky who hadn't programmed in 10 years and got so into AI that he's like, have a new love for programming because I don't have to write code. And the same thing, like he's way into it because he's solving problems and.
Brett (56:43.29)
Yep.
Brett (57:02.821)
Yeah.
Jeff Cohen (57:09.572)
He's just kind of doing code reviews, which he was doing anyway in his role of managing.
Brett (57:14.659)
Yeah. I have a new love for QBRs. 'Cause I don't actually have to pull all the information and say
Jeff Cohen (57:17.423)
Yes.
Rio Longacre (57:20.38)
Well well be but being able to problem solve, do code review, which which he's probably great at 'cause he's been doing it for fifteen years at a high level, right? He's probably can do it like a hundred times better than certainly junior person could, right? So be but being able to apply those just what he likes doing and he's best at, that's a huge unlock, right? And you can see why people are enjoying this so much. So yeah, I totally get that.
Jeff Cohen (57:34.052)
Yeah, it is. Yeah.
Brett (57:39.236)
Yeah, this has been yeah, i what what I think is is super fascinating how you guys have taken that top down as well as bottom up approach to transform how the organization thinks about holistically using like you're transforming yourself into an AI first organization. And not just not just words, but actually implementing it at a very deep level. 'Cause that's really the only way that you're gonna help bring your clients along on this journey, right? Who are often
Jeff Cohen (57:53.53)
We are.
Jeff Cohen (58:02.8)
Yeah, and at the end of the day, we're gonna build better products, right? Because we're thinking that way and we're gonna be able to help customers. Like I was meeting with a customer, they were talking about it and I said, no, you're telling me how your process works. go, what would you, could you just tell me like, at the end of the day, what would be your ideal situation? And now they're thinking and they're like, well, I would just love to be able to, you know.
Brett (58:06.755)
Yeah.
Yep.
Jeff Cohen (58:31.524)
have this output occur and you're like, okay, great. And now you can think in 30 different ways and you're not confined by an SOP. you know, I heard this great analogy that like a lot of what tech was built on was basically part of the industrial revolution, right? And the widget, the widgetizing of the world. And we as humans kind of became part of those widgets where we created the repeatable processes. We were just using human process to be able to do that.
Brett (58:39.812)
Yeah.
Jeff Cohen (59:01.578)
And AI is taking us to the AI revolution where that widget process is now being done by computers, which means we need to kind of reinvent ourselves and reinvent what we can do, which is scary, right? I'm not going to deny that it's not scary. maybe it's even, you know, what's funny is, like...
us older people are probably okay, because we're towards the end of all this and the younger people are probably okay because they're going to come in and understand this. It's the kind of the people in the middle who have to kind of learn it and unlearn what they already know, which is why you see some of that AI backlash that you're seeing of AI being scary. But I think once you learn to use AI in a very meaningful, responsible, and
Brett (59:26.693)
Yeah.
Brett (59:38.02)
Yeah.
Brett (59:41.977)
Yeah.
Jeff Cohen (59:52.29)
efficient and effective way. That's where you see just exponential growth in both your personal, professional and business development.
Brett (01:00:01.536)
Yeah, you you make some really sorry, go ahead.
Rio Longacre (01:00:01.892)
You mentioned you mentioned Yeah, go ahead. Yeah. Well I was just gonna say yeah, I'll I'll do real fast. Like so I was just gonna say like the judgment comment you just made, like being being able to have the judgment, because you think about your your colleague you just mentioned, he's able to do code review already. He's great at it, right? But this is helping him do code review more because he's able to create code and and create code for himself and other people he can then review and then he can think about how to solve problems and how to at a high level apply this.
Jeff Cohen (01:00:18.255)
Yeah.
Rio Longacre (01:00:28.124)
But you have to have that context in order to get you have to already be good at it in order to use AI to do it. Right. So so I think the people in the middle who hadn't graduated that level is probably a little more challenging to them because they need it takes years to learn the skills to do that, right? In order so I I so I think that I sort of think that'll that probably more challenging for people in the middle and bread. I'm sorry, cut you off. Go ahead.
Jeff Cohen (01:00:40.451)
It does. It does.
Brett (01:00:46.198)
No, no, no, yeah, th the well we do that. I told you, John I told you, Jeff, we do this. we we've known each other for a long time. But no, you made a point point that I thought was interesting about the d the Industrial Revolution and the processes that we've just sort of assumed within organizations, within daily workflows or work streams, that oftentimes I found with clients internally with my own companies that you could have a ton of processes yet still be really process poor.
right as an organization and you almost get caught in this sort of loop where you're just running through the steps of these kind of processes that don't make you any more effective as an organization or as a team. And so and Rio, you've mentioned that before too, where you go into organizations that are that have a ton of processes but they're incredibly paradoxically, yeah, or or it's not documented, it's in people's heads. And when you can automate some of those kind of fundamental like this is how these are jobs to be done.
Rio Longacre (01:01:31.844)
Nothing's documented sometimes, yeah.
Jeff Cohen (01:01:42.02)
Yep. And that's the beauty of AI, right? The beauty of a product like Studio is that you're capturing the knowledge, you're capturing the process, you're adding the governance, you're adding the skills, right? And now it can be replicated.
Brett (01:01:42.054)
around parti
Brett (01:01:56.994)
Yeah.
Jeff Cohen (01:02:00.234)
and it can be audited and it can be adjusted. you are, it's kind of like a new version of like ISO regulate of, know, ISO certification, right? Cause you're, if you know that process, it's about a documentation process. And then you're using intelligence to kind of sit on top of it and then to tell it when to execute and how to check and how to double check and such. And you're right, like the context of it, yes.
Brett (01:02:11.234)
Yeah.
Brett (01:02:26.434)
And it creates re repeatability, right? Yeah, w without all the breakage along that path, which is often what happens within teams, is they're operating under the same set of rules and processes potentially, but there's all this breakage along because of human error, et cetera, along the way.
Jeff Cohen (01:02:41.327)
What?
But you also have a problem, a problem, a challenge, which is that like we as humans, we stop working at a certain point in time. And when you build AI to be running processes, it doesn't stop working. And so if something changes at 10 o'clock at night that you wouldn't have normally caught until nine o'clock in the morning when you run a, you know, an air report and then maybe you're sick on Monday and you don't run that until Tuesday, you've now, you know, ran three or four days of something happening. And when you build AI workflows,
that allows you to catch these things and to surface them, whether you want to surface them into an action that you take, whether you want to surface it into a Slack message that gives you a heads up or one that you want to send it to an email or whether you want to look on the action board, right? We have a board where you can see all the actions that you need to take. You can then have a conversation with it. You can understand what it's doing. You can understand the logic that it had. Our whole logic, our whole thinking, you would use the word black.
box earlier, our whole thinking is that it has to be a glass box. And so everything that we do, we show you exactly what logic we used, how we got to the decision that we got to. And then you can go in and change your knowledge and your skills to have it act or react the way that you want if it's not giving you the output that you desire. And that, I think, from a governance perspective is really seen as a high value because at the end of the day, we want to know what actions were taken.
We want to know who took what actions. We want to know what the impact of those actions were so that we can then repeat them. And again, like going back to kind of a circular moment here, these aren't anything new. We've been doing a lot of this since 2016 when Amazon ads came out with their first API and we, and we built rule-based automations. The difference is, that we never really knew which rules conflicted with other rules because to look at your 35
Jeff Cohen (01:04:41.714)
or 350 or 3500 rules and understand how they all apply never could be done. Well AI can now actually do all those things for you and can help you kind of figure that out. So it's like we're in this like mind blown world where things are almost anything you can think of. I almost get annoyed sometimes with my team when I'm like, can we do this? And you're like, yeah, we can do that. I'm like, we can do everything. The question is what should we do?
Brett (01:04:57.475)
Yeah.
Rio Longacre (01:05:10.246)
Should we be doing it right?
Jeff Cohen (01:05:11.955)
How can we make sure that what we're doing is right? And so that's what we focus in on for how we make our incremental changes as opposed to, you know, it's an evolution revolution.
Brett (01:05:15.715)
Yeah.
Brett (01:05:23.672)
I'm a s I'm assuming with the with the retail media use cases it it incr increases your sort of optimization capacity, right? Because it is a a as you're saying a glass box where you can you can f say, hey, there's thirty five hundred decisions that were made along the way. How do we actually figure out where certain things are breaking? And you can c you can conversationally interface with it to to audit almost to audit what's going on. Whereas before you're saying at Amazon ads, that it was just happening.
Jeff Cohen (01:05:37.294)
Yeah.
Brett (01:05:49.75)
And you couldn't break down in that complexity what was actually driving whatever outcome or result.
Jeff Cohen (01:05:54.234)
Well, and the thing that I'll remind people of is that ads, Rio's heard this line from me before, right? Ads need time to breathe, right? You can over-optimize ads if you don't give them enough, but that goes back to SOPs and saying like, hey, it's gotta be out there for seven days. It needs to get to a certain dollar spend. It needs to get to a certain number of impressions. Those are all part of the rules that let the ads breathe. I've talked a lot about like that.
Brett (01:06:10.52)
Yep.
Brett (01:06:18.114)
Yeah.
Rio Longacre (01:06:20.113)
Yeah.
That was always the worst when clients would like after a day or two, like, it's not working, pull it. It's like that's not even a test where like they would like, Well, wanna do a ten thousand dollar test. Like, you're not gonna learn anything, right?
Brett (01:06:31.628)
Yeah. Yeah. And and and each and each manager of that type of workflow could make a different decision based on their own inclinations and their own intuition. Which which creates a ton of human error, right?
Jeff Cohen (01:06:31.748)
Right. Yeah, you gotta have a repeal.
Jeff Cohen (01:06:40.432)
correct. Which could be based on
Well, that could be based on your products because your products might have different margins. It could be based on the ad types or the retail media networks. You can have different rules for dollar general than you have for Best Buy or that you have for Walmart, right? And so you can create rules down to a retail media network or to a product category level. It's almost, it's this.
Brett (01:07:11.628)
Infinite. Yeah. And
Jeff Cohen (01:07:12.662)
It's impenet, which creates new issues, not issues, but creates new, it becomes overwhelming for people, which is why you do the crawl, walk, run of how you build yourself to get there so that you build the confidence in what you're doing. You build the confidence in the system and how it's doing it. And you make sure that you have the right government governance and you make sure you have the right checkpoints along the way, because that's critical and important too.
Brett (01:07:30.585)
Yeah.
Rio Longacre (01:07:40.433)
It's funny you mentioned about the budget though. I had one client. I I won't name any names, but they they only approved a very small budget for this test. And I remember when we the results were actually pretty good, but there was not a huge test. And then when they came back, the final assessment I got back from my client was, it didn't it failed because the spend wasn't big enough to learn anything. Yeah. Can't make this stuff up sometimes. But
Jeff Cohen (01:08:04.143)
Yes.
Rio Longacre (01:08:05.276)
But you did mention something interesting, I'd like to dig into a little bit, because I think this is an interesting point that I'm hearing kind of again and again. Brett, I think you'll like this too. Was you mentioned it wasn't forward-deployed engineer. I know you had another term for it, Jeff, but the concept of sending really consultants who understand AI to help work with clients to understand and really apply and and and execute their AI use cases. It sounds like you're doing some of that too. Love if you could dig into that a little bit and how you're handling that.
Jeff Cohen (01:08:31.408)
Yeah, so it is an FDE and we see it in two ways. We have an advisory service so that if your organization is going through a transformational period, but you as an organization don't know how to apply it, maybe you want to run your own agentic week. Maybe you.
don't know what questions to be asking or how to even develop one of these plans. We have a whole advisory service where we can help walk you through those processes. Then when you get into using a tool like Studio, that's where we have FDEs who are part of the onboarding and management process. So you're not just assigned an account manager, but you also have an FDE who's helping you in the tool and build. And that's a model, like we're seeing that with,
Brett (01:09:18.638)
Yeah.
Jeff Cohen (01:09:21.034)
and and chat GVT that that's really what they need to scale and at the end of the day we're finding that a lot of companies have desires but they don't have strategies and they don't have the the knowledge or they don't have the skill set to be able to do it and so we are going to need to help them along that way.
Brett (01:09:37.124)
It's
And w and how do you assess the talent for for a potential F D E, not an FTE, forward deployed engineer? Is it someone that is both consultative, understands sort of in and consultative and heavily technical and s you know, strategic? I mean they have to be technically sound to be able to go in and actually build. I mean are they are they full stack developers?
Jeff Cohen (01:10:03.661)
No, they're just understanding, in our case, they're understanding how studio works and they're understanding the different components of studio and how to apply those to, I'll call it retail media or commerce media type of campaigns. But even our CSDs are very highly skilled in how to use the Sky platform and how...
Brett (01:10:09.529)
Yeah.
Jeff Cohen (01:10:29.22)
the ad campaigns work that can come in and kind of help manage that as well. And so you do have a good skill set of people who.
Brett (01:10:37.654)
Yeah. So is the but the FTE's more than just a solutions engineer, right?
Jeff Cohen (01:10:41.744)
Yes, they're almost like, we have a team here called expert services. So we don't do any managed service here, but we do have expert services where we'll do like a managed onboarding or we'll do a, might have somebody leaving on maternity leave and you need somebody to run your DSP campaigns for three months while they're out. And so our expert services team will come in and kind of help you maintain.
Rio Longacre (01:11:10.258)
Right. So you're not selling professional services, but this is something you do offer to clients that that you know, to help them to help them get better use of your technology.
Jeff Cohen (01:11:13.978)
Correct, yeah. Yeah, and so our FDEs are kind of an extension of that. We just believe that we don't know whether it's six weeks, six months, or six years before the industry will catch up with builders.
Brett (01:11:21.762)
Yeah.
Brett (01:11:30.978)
Yeah. Yeah. Or the particular client, 'cause they're in different areas of that maturity curve.
Jeff Cohen (01:11:36.736)
Exactly. you know, Rio, your article and why it was so, you know.
Well, besides the fact that my daughter is going into, know, potentially going into this space is that, you know, that is what our industry needs. We need builders. We need people who come in and, and, and I've seen what happens when you have a builder on your team. We had somebody come into our SDR BDR process and, and bring in all these tools to build these automated workflows. And when you have a builder who understands how to do all of that, the amount of unlock that they can create is exponential.
Brett (01:11:52.174)
Yeah.
Jeff Cohen (01:12:13.138)
And, you know, we at Sky have a process where we try to identify, train and create builders because we know as an organization that we need more builders throughout the organization. So who will the future FDEs be? They may be other people that are currently working at Sky today doing other roles who just find that they're really good at building workflows inside Claude, but also know how retail media works, right?
Brett (01:12:13.783)
Yep.
Brett (01:12:35.767)
Yeah.
Brett (01:12:42.242)
Yeah, b building is no longer the purview of simply the IT team, you know, the people behind the scenes that are doing coding. It's now been democratized. Maybe not at the same it's not equal across the board, but it gives people Yeah.
Jeff Cohen (01:12:43.068)
set.
Jeff Cohen (01:12:47.023)
Yeah.
Jeff Cohen (01:12:52.474)
Cause these, well, they're not actually like writing code, right? Like you don't have to necessarily write code to be able to do this, but you do need to be process oriented. Yeah.
Brett (01:13:00.312)
Yeah. Yeah, yeah, it g it puts you into more of a product role. You don't have to necessarily be an expert engineer, but it but everybody in a way is kind of like what we traditionally define as like a product leader. You can spec out a solution. Yeah.
Rio Longacre (01:13:00.858)
Need to be pretty technical, yeah.
Jeff Cohen (01:13:13.167)
Yeah.
Rio Longacre (01:13:13.466)
Yeah, you have to understand product, pro product management. Yeah, also it's product strategy. You have to be pretty good at like systems thinking, I w I as well. And then funny, we had you like what you just mentioned about people being more product people, right? We had a a guy here. He built this tool where he looked at a bunch of RFPs that were coming in, built this tool that it rips through the RFP, like under like it summarizes it, creates the potential team to staff it.
Brett (01:13:19.289)
Yep.
Rio Longacre (01:13:38.182)
finds a artifacts on our internal systems here that actually we would use for case studies and for like sample templates. It's it's it's it's outstanding what it does. It's such a time saver and even starts putting together the response. So just a guy did it in his it just
Brett (01:13:47.725)
Yeah.
Jeff Cohen (01:13:50.864)
In six months from now, it'll probably build 80 % of the RFP for you as it continues to evolve. Yeah.
Brett (01:13:58.306)
Yep. Yep.
Rio Longacre (01:13:59.719)
Yeah, these tools get better and better as they learn more. Yeah, that's really cool. I I and this guy was you know, he was fairly technical, but he's not he's not he didn't write any code to do this. It was all just done vibe coding with with clawed code and different tools. So yeah, it's pretty amazing.
Brett (01:14:11.201)
Yeah. It is it is the way forward for sure. So and you guys are are sort of living and breathing and it's certain it's very interesting. So should we go to quick hits? Yeah. So
Rio Longacre (01:14:19.44)
Do it. Yeah.
Brett (01:14:22.441)
Maybe we'll go back to the topic we've been on AI for quite a while, Jeff, is is what's the let's go back to r let's go back yeah, I know it does. Ever every conversation we won't say devolves, we'll say evolves into an AI conversation. what is the most overused phrase in retail or commerce media right now? Going back to that topic.
Jeff Cohen (01:14:27.214)
always happens.
Jeff Cohen (01:14:42.916)
Yeah, to not pull one from the AI, because there's plenty within that arena. I would say ROAS is probably one, or maybe even iROAS, or maybe the term incrementality. I like to joke incrementality is a tough word. It still has the squiggly red line under it, because nobody knows how to define it.
Brett (01:14:51.01)
Yeah.
Brett (01:15:06.133)
Microsoft OS or Mac OS doesn't know how to define it. They don't even see it as a
Jeff Cohen (01:15:08.6)
Right, we're gonna have like an industry crusade to get the word incrementality to finally be a recognized word. But I think ROAS is a term that it had a very, it has, and it had a very specific purpose in our industry. And it can be very limiting if everything you're doing is ROAS based. And if you are looking for growth and expansion, you need to be able to expand your vocabulary.
Brett (01:15:36.439)
Yeah, good point.
Rio Longacre (01:15:38.578)
Yeah, I like that. What is one marketing decision that AI should not ever make, at least right now, without human approval?
Jeff Cohen (01:15:48.784)
Probably where budget should be spent.
I think it can make recommendations, but I think it definitely requires approval to go in and look at it, make sure that it makes sense for your business. And so I think any type of budgeting spend, think it's clearly, we know that AI is smart enough to make bid changes. It's been making bid changes for a long time. But when we're talking about like new spend, would definitely want human in the loop.
Brett (01:16:20.855)
Yeah. So from all of your learnings internally at Sky, you've been there for about a year now, right? around sort of AI transformation both internally and then what you're trying to do with your clients. What what are the first c what are the most important, say top three steps that a that a that a brand should think about in terms of getting their feet wet and starting to to move down that path?
Jeff Cohen (01:16:45.72)
So I think the first one, I tell this one to my kids, right? So this is like, would you tell your kids? Which is, you've gotta be playing with this on a personal basis to understand what it is, right? So if you don't have co-work and you're not spending 20 bucks a month and you're not writing even little codes to clean up your Gmail, that's the first thing that you need to be doing from a personal perspective. From a professional perspective, it's gotta be a top-down drive.
Brett (01:16:56.248)
Yeah.
Jeff Cohen (01:17:15.474)
driver, right? Like ours is top down chairman of the board all the way down through management that AI is an important initiative. We've invested in it and we support it. I think that that is and then I think the third one is you got to look for your breakthrough moment, which is the breakthrough moment is when you truly realize that AI can do things that you were bound by in the past. And when you kind of have that epiphany,
It's really eye-opening to what you can be doing in the future.
Rio Longacre (01:17:53.597)
I like that. Turning back to retail media for a second, what's one thing that smart people in the industry seem to believe or always say about retail media that you just think is flat out wrong?
Brett (01:17:53.773)
That is great.
Jeff Cohen (01:18:04.823)
Ooh.
Jeff Cohen (01:18:12.292)
That's, I'm trying to think like, what are my pet peeves is kind of what you're asking. It's kind of what you're asking, like what are my peeves? Yeah. Well, maybe it even goes back to what we were talking about earlier about the mistrust with the hedge gardens.
Rio Longacre (01:18:16.722)
Pretty much, yeah.
Jeff Cohen (01:18:31.608)
And and that you know, there is a lot of knowledge that they have And it's a matter of how you contest their knowledge to your own knowledge It's it's not black and white as to whether they're good or whether they're bad it goes back to me being a homer and I always like to say that because I I like to joke when I was at Amazon I didn't drink the ju the Kool-Aid because I was at Amazon I went to Amazon because I was drinking the
Kool-Aid, right? And like, we have to sometimes put ourselves in perspective of like where we're at in this industry and what type of opportunities that Walmart and Amazon and TikTok shops and all these other places are creating for us that didn't exist five years ago. And to me, that's like amazing and fascinating. The idea that you can launch a whole brand now on TikTok shops and grow a multi-million
million dollar business is crazy, right? And I think that's awesome.
Brett (01:19:38.679)
That is awesome. Well, hey, Jeff, thanks for for joining us today. We we we did a nice long episode, nice and substantive. and everybody that joined us, thank you for joining us. visit us at www.signalandnoise.ai.
Rio Longacre (01:19:41.33)
It's been fun.
Brett (01:19:53.269)
we're going to be doing a huge activation at Advertising Week, Signal and Noise Live, with a number of sponsors. So if you're interested in joining that event, let us know. Visit our website. We'll have banners, etc., as well on LinkedIn, talking about the event and what we're doing. Or just sign up for our newsletter.
Rio Longacre (01:20:09.402)
Or just sign up for our newsletter. There's a sign up there too. That'll keep you in a loop.
Brett (01:20:12.609)
Yep, and that'll that'll hook you up with all of our content, from articles to podcasts, which are across every major platform you can think of. So thanks everybody for joining us, and thanks Jeff. It was an awesome conversation. And we'll see everybody next time.
Jeff Cohen (01:20:24.73)
Thanks guys.
header.all-comments