Episode 168 · Season 7 · ·
The Algorithmic Brain
Jeremy Fain, CEO and Co-Founder of Cognitiv, on why deep learning can finally run advertising to real business outcomes, and why brands still cannot be built inside a chatbot.
Jeremy Fain, CEO & Co-Founder, Cognitiv. In conversation with John Horsley.
The Algorithmic Brain
0:0039:00Twenty-five years across digital media, from building the first online American Express rewards platform to the IAB standards years and a premium exchange at CBS Interactive, before co-founding Cognitiv in 2015. In this episode Fain makes the case that deep learning can finally run advertising to real business outcomes, and that no chatbot is ever going to build a brand.
Fain took his first marketing job at Digitas in 2000, helping blue-chip companies understand the internet and building early work like the first online American Express rewards platform and an HBO email newsletter. After business school he moved to the media side and spent the standards years at the IAB, from about 2005 to 2009, learning how a media company runs during what he calls the Wild West. He went into ad tech on the publisher side at CBS Interactive, which he says launched the first premium exchange, then moved to Rubicon Project, which became Magnite after its IPO. Seeing the guts of the programmatic infrastructure, and a new technology called deep learning brought to him by co-founder Aaron from a Stanford postdoc, he co-founded Cognitiv in 2015.
In this conversation with host John Horsley, Fain argues that most advertising has been held back to digital proxies like clicks and view-throughs, when the businesses paying for it run on lifetime value, incremental lift and net new customers. He explains how deep learning can finally put a marketer's data to work across the whole funnel, why 'if you're not saying you use deep learning, it could be anything' is his litmus test for AI claims, and how Cognitiv treats the right message, right person, right place and right time as one scientific problem rather than four separate products. He is sceptical that agentic selling replaces relationship-based sales any time soon, and certain that human needs for touch, experience and brand do not disappear. You can buy something online, he says, but then you have to return it.
You have built ContextGPT. What problem does it solve?
“There's no company that answers all four of those questions at the same time.”
Advertising is meant to be the right message to the right person at the right place at the right time, we have said that since before the internet. Ad tech has invested in one piece of that in every company. No company answers all four questions at once, you are an audience company, or a context company, or a DSP for timing, or a dynamic creative company. Cognitiv is one of the few that treats the auction and the impression on all four at once, scientifically. We were already doing contextual understanding to inform our algorithms, so we pushed it out as a product for the cookie apocalypse, and we were the first to use prompts to target context.
Is the journey changing the outcomes marketers want?
“These are really taking huge amounts of data, making sense out of them. Humans cannot do that.”
We can build deep learning algorithms that test fifteen to twenty creatives across segments, taglines and offers. Instead of an A/B test that names one best creative, we can say this messaging works best here, that one there, and even if a message works only three per cent of the time it may reach an important untapped demographic. These are big data tools, not LLMs, taking huge amounts of data and making sense of them. Humans cannot do that, so we are enhancing what humans can accomplish.
Five years out, what changes in the value chain?
“If brand didn't matter, On and Hoka wouldn't have been able to build huge brands against Nike and Adi- Adidas and all these other companies.”
Chatbots are becoming a large part of life, with a generation using them for everything. But malls are doing better and younger people like to touch and try clothes, so we have two competing pieces of the journey. If brand did not matter, On and Hoka could not have built huge brands against Nike and Adidas, so brand is always key, and you are not going to build it inside a chatbot. As a student of history, I do not see the human decision-making process or those genetic needs for interaction, experience and touch changing. You can buy something online, but then you have to return it, and getting them to keep it is the key.
Put yourself forward or recommend a guest for us to consider.
Request an interview →Most companies try to reach senior decision makers. We give them a reason to engage.
Partner with Us →Have your say.Add your voice to the conversation.
Voices is where senior practitioners share what they're seeing, what they believe and what they'd change.
Become a Voice →What do you do about brand safety when a place is contextually relevant but wrong?
“Brand safety is an interesting thing that we, we talk about a lot, but I think it's sort of solved, frankly”
We integrate with made for advertising definitions and similar tools. Brand safety is something we talk about a lot, but I think it is sort of solved from the advertiser's perspective. They usually have their own tools and lists of acceptable and non-acceptable sites, so we integrate with those. It is not much of a concern for us because we can read the language on every page, run sentiment analysis, inclusivity analysis, and filter everything out.
What advice would you give someone at the start of their career?
“Critical thinking skills, flexibility of thinking, problem-solving skills, those are the key to any career's future at this point.”
There is a debate that institutions should specialise more and hand out vocational degrees, and I could not disagree more. With AI, specialisation could be taken away from you tomorrow. The key in how we use AI tools, especially in coding, is seeing a problem, designing a solution and understanding its effectiveness. So critical thinking, flexibility of thinking and problem-solving are the key to any career's future. You have to stay flexible, because you never know what is going to happen to the job you have taken on.
Never miss an episode. Follow Business of Marketing.
Read the full transcript
Lightly edited for readability
Speaker 1: Welcome to the Business of Marketing. My name's John Horsley, I'm the host of today's episode. I'm joined by Jeremy Fain, who is CEO and co-founder at Cognitiv. Welcome to the show, Jeremy.
Speaker 2: Thanks for having me.
Speaker 1: What would be really great as a starter for 10 is if you can just give our listeners, um, uh, an example in terms of how you decided to get into the world of marketing, media, advertising. What was it that attracted you in, in the first instance?
Speaker 2: Yeah, my journey's been, uh, a little bit of a nonlinear journey, but it's sort of covered all of marketing and advertising, which I think has helped me a lot have a sort of a full view of the space.
Speaker 1: Mm.
Speaker 2: From the beginning of the internet, really. I- the first sort of marketing job I took was at Digitas in 2000.
Speaker 1: Wow.
Speaker 2: In 2000. And that was just my second job out, third job out of school, but second real job. I was at a startup for about six months. I don't, I don't know if that counts. Um, but joined Digitas and helped all these blue-chip companies understand how to use the internet. Per- we called it the internet. I worked in-
Speaker 1: Sure
Speaker 2: ... the internet back then. And we built things like the American Express, uh, uh, rewards platform, the first one that was online. Uh, we helped HBO reach their membership with a newsletter for email. All those kinds of things were brand new to business, and I think my, my, um, career has sort of evolved from there. And after business school, I started moving towards the media side of the business, and I joined the IAB, um, doing, uh, industry standards, which was a really important time from about 2005 to 2009, where it was still the Wild West, and those standards really helped business open up and flourish across the United States and then the globe. Uh, and that was sort of my beginning of media. I really learned everything about how to run a media company and all those pieces.
Speaker 1: Mm.
Speaker 2: I ended up in ad tech because on the publisher side, I was at CBS Interactive. We'd launched the first premium exchange out there for really premium inventory-
Speaker 1: Mm
Speaker 2: ... and understanding programmatic was just beginning then. I moved over to Rubicon Project because I got to know them. They became Magnet, of course-
Speaker 1: Yeah
Speaker 2: ... after we went public. And then because I was there, I saw a lot of the programmatic guts of the, uh, of the infrastructure and the industry, and I saw an opportunity for us to use this new technology at the time called deep learning-
Speaker 1: Mm
Speaker 2: ... to really make bidding and ad placement a lot smarter.
Speaker 1: Yeah.
Speaker 2: And that's why we started, uh, Cognitiv in 2015.
Speaker 1: Wow. What an awesome journey. Um, that's making me think of pioneering and k- kind of being on the frontier of, of each step of, of the web, as it were.
Speaker 2: Yeah.
Speaker 1: Or evolution of, of tech.
Speaker 2: My, my first job was consulting-
Speaker 1: Oh
Speaker 2: ... but in technology consulting, uh, PwC Consulting, which is now sort of IBM Global Services.
Speaker 1: Mm.
Speaker 2: And I'm an engineer by training in, uh, undergrad, so I've always liked to solve problems, but I've always liked the technology of that, of that solution, and I've always wanted to work on the new interesting things.
Speaker 1: Mm.
Speaker 2: I think I've just been drawn to that my whole career.
Speaker 1: Yeah. Fantastic. Um, tell me a little bit about Cognitiv. What's the raison d'être behind the business?
Speaker 2: Yeah. So speaking from that sort of, uh, that progression, uh, when I was at Rubicon Project, we had a view into all the bids that the DSPs were making, and they were, back then and frankly still now, relatively, uh, let's call them, um, formulated, templated.
Speaker 1: Sure.
Speaker 2: You know, it was $10, $1, 10 cents. Uh, a lot of things have changed since then, but really we didn't really see a lot of intelligence from a campaign-to-campaign, brand-to-brand difference based on anything more than maybe context or the number of cookies you had. So that changed a lot of things for me. And then my partner, one of my co-founders, Aaron, he came along and told us about this new thing he had been using in his postdoc at Stanford called deep learning.
Speaker 1: Mm.
Speaker 2: And that was brand new, 2015. It really was more academic at that time. Uh, and he said, "Look, it's a really powerful predictive technology, but it needs a lot of data." And I said, "Hey, I know where there's a lot of data, programmatic advertising."
Speaker 1: Mm-hmm.
Speaker 2: And so we started the company with this idea that, uh, the data that we have in programmatic can help inform a deep learning algorithm in a way that it is much, much better at predicting consumer behavior and therefore helping from a performance perspective, and that's sort of, we started at the bottom of the funnel there with that, with that hypothesis.
Speaker 1: Wow. That's great. So 2015, 11 years. How have things progressed over that time ahead? We're obviously in the world of AI now-
Speaker 2: Yeah
Speaker 1: ... as well, so, but you were already h- thinking along very similar lines is what it sounds like.
Speaker 2: Yeah, we were, we were first movers, definitely. We were early, but I think, you know, we-- It was funny because I, I dug up some, uh, investor, uh, decks from that period, and in those decks, we really had a very clear vision that we now I think have finally been able to realize, and that was to be a full funnel algorithmic solution for a marketer.
Speaker 1: Mm.
Speaker 2: And our tagline since the beginning really has been the brain behind the brands, and we had to start somewhere because ad tech is really not based on full funnel marketing. It's based on one piece or one tactic, context, audience-
Speaker 1: Mm
Speaker 2: ... uh, DSP, uh, DMP, you know, those types of things. And I, and so we, we had to start somewhere, but really over the years, we've been able to fill out a platform, which we call a deep learning advertising platform, that's really enabling marketers to use us entirely across all their channels and across all of the funnel.
Speaker 1: Mm.
Speaker 2: And I think that is the really cool- ... powerful thing that deep learning can finally realize is all these marketers say they have all this data, and they've been collecting data in their CDPs and all these other things over the years, but activating that data has largely not been practical.
Speaker 1: Mm.
Speaker 2: Deep learning, AI, now synonymous with AI, has made that possible.
Speaker 1: Yeah. What, um, what are people, I guess, drilling into? So what is it that they're trying to understand? You know, is it normally the same thing, or is it not so atypical?
Speaker 2: Well, the beauty of what we do is that everything we do is based on the goal of the marketer and, um, and the KPI of the campaign.
Speaker 1: Okay.
Speaker 2: Generally speaking, uh, the most advanced marketers that are using the majority of our suite, uh, are, are looking for a business KPI, and I think that's the next evolution of marketing and advertising, just like advertising really, is that before we've been sort of held back to, um, digital KPIs. Okay? Like a click or a view-through or a visit to the site, things like that. Most businesses don't drive investment, don't drive revenue through just those pieces. You know, uh, insurance companies are looking for LTV, they're looking for incremental lift. Large consumer brands are always looking for net new customers.
Speaker 1: Sure.
Speaker 2: Those are the most important pieces. Those don't naturally come, uh, to the infrastructure of the internet, but those are measurable, and we, the deep learning algorithms in our platform has built those event systems that we can bring all that information in and actually run your marketing-
Speaker 1: Mm
Speaker 2: ... towards your business KPIs. I think that is the, the most fun thing that we do with our advertisers.
Speaker 1: Yeah. I mean, that, that makes total sense, and I mean, in the past-
Speaker 2: You'd think
Speaker 1: ... I look at everything else, and it's kind of vanity metrics or people being rewarded, frankly, for spending a lot of money. Um-
Speaker 2: Yeah
Speaker 1: ... and, "Oh, it's great. We got a lot of engagement, we got a lot of clicks," but perhaps-
Speaker 2: I think-
Speaker 1: ... not much of an outcome.
Speaker 2: I, I think it's just not been an option before.
Speaker 1: Yeah.
Speaker 2: You know? And I think that is, uh, that has always been the promise of digital marketing-
Speaker 1: Mm
Speaker 2: ... since the beginning of when, when I got to start in it.
Speaker 1: That is, that is so measurable.
Speaker 2: So measurable.
Speaker 1: Mm.
Speaker 2: But how do you use that data? How do you activate that data? What does it mean? Um, how do you connect the offline and the online worlds?
Speaker 1: Mm.
Speaker 2: We're at a point where we can now do that pretty effectively, and, and marketers now have that option. I think the, the question is, do they know about it? Are they, are they defeated enough that they're still using clicks and things like that? But let's, let's bring them into the next evolution.
Speaker 1: Yeah. I'm thinking, wow, so in 2015, you're selling something with deep learning-
Speaker 2: Yeah
Speaker 1: ... and so on that people are probably not so familiar with as well. Um, and perhaps you're going against the, the convention to the norm, as it were. Um, was it hard making those sales then versus is it slightly easier now that people are f- bit more educated?
Speaker 2: Yeah. Well, I don't know if you've heard, but AI is a thing, I think. Uh-
Speaker 1: Sure
Speaker 2: ... it's, it's definitely getting easier now. I think when we started, we knew, we knew there were gonna be a number of challenges, but the challenges, uh, you know, I was listening to one of your other podcasts, and they were talking about, you know, uh, the evolution of programmatic and knowledge and comfort and things like that.
Speaker 1: Sure.
Speaker 2: And even in 2015, programmatic was already pretty comfortable.
Speaker 1: Mm.
Speaker 2: So I think we had, we had to toe the line between a new technology and a better mousetrap, really, 'cause performance marketing's been around for a long, long time. That's what we were offering.
Speaker 1: Mm.
Speaker 2: And so I think w- it, we ha- you know, whenever you have a new company, you have to spread your message really widely. We were lucky enough to find some really great consumer brands that w- uh, were really looking for a solution-
Speaker 1: Mm
Speaker 2: ... to that problem, that, that business KPI problem, and we had some really great early adoption. But as we've gotten older, you know, the comfort behind the technologies and the opportunities has, has definitely made it easier.
Speaker 1: Mm.
Speaker 2: And then one last thing on that, like when we, in 2015, it was still pretty, uh, close to where people had been talking about AI a lot still, but it was really not living up to its potential back then.
Speaker 1: Sure.
Speaker 2: And so for us to say AI was not good in our pitch, so we really talked about deep learning, and I think that DNA's really stuck with us to this day. We really talk about deep learning.
Speaker 1: Mm.
Speaker 2: And, and I f- I think it's a litmus test with a lot of companies that say they have AI, they have AI, they have AI. You know, I think if you're not actively using deep learning technologies, you could be lo- using machine learning and saying you have AI, which is a totally... It's sort of an umbrella, uh, term, but really if you're not saying you use deep learning, it could be anything.
Speaker 1: Yeah.
Speaker 2: 'Cause it's really a subset of machine learning.
Speaker 1: Mm. Yeah. I was, I was wondering that, and y- you were almost coming on to answer it, frankly, which is, you know, every company, every tech company says, "We're AI powered. We're AI this. We're AI, AI whatever." Um, and it's about differentiation. So, you know, and the marketplace is fairly crowded. It's very commoditized as an ad tech to a certain extent. How do you differentiate yourselves in terms of positioning and, and message?
Speaker 2: I think we are, I think there, you know, there's always challenges with first mover position, but for us, I think we've done a very good job at sticking to our message.
Speaker 1: Mm.
Speaker 2: We haven't pivoted. We've always been trying to build out this full funnel, omni-channel algorithmic-
Speaker 1: Mm
Speaker 2: ... you know, solution that delivers really great insights to marketers, et cetera. And so I think we're in a place now where we have really good brand recognition. I think we have really good trust in the marketplace. Our, you know, uh, the, the three founders, we all come from what's now called STEM backgrounds. We're all engineers or computer scientists. I really moved into the business side.
Speaker 1: Mm.
Speaker 2: But my two other co-founders, we've known each other since we were 10 years old.
Speaker 1: Wow.
Speaker 2: We went to all these elementary school and, and, uh, high schools and secondary schools together. Then we went off and did our own careers in universities-
Speaker 1: Yeah
Speaker 2: ... and things. But we've always had a position in the marketplace, and I think our position in marketing is we're the smart, trusted, honest, transparent, because we- the algorithm's doing it. We don't have anything to hide. The algo- we're gonna tell you what the algorithm is doing, and we're gonna break it open and show you the insights it's delivering to you because that's the power. We're good at programming the algorithms, building the infrastructure for the algorithms, and giving it the data it needs, the features it needs to make it a s- a, a, a, a solution.
Speaker 1: Yeah.
Speaker 2: I think that's helped us a lot.
Speaker 1: How do you see, yeah, how do you see ad tech evolving, um, as the customer journey is changing? So more people are using LLMs now-
Speaker 2: Yeah
Speaker 1: ... um, to identify a product or find additional information or rese- you know, researching a purchase. Um, we'll gradually move into a world of I, I guess business to, to agent, and then actually potentially further down the line, agent to agent. So how do you see ad tech evolving to stay relevant, I guess, in, in an ever-changing world?
Speaker 2: Well, ad tech is, uh, very adaptive.
Speaker 1: Sure.
Speaker 2: I think we can all agree on that. I think if you wanna w- if you wanna have a very, uh, slow-evolving, uh, steady, uh, career, don't go into ad tech, right?
Speaker 1: Mm-hmm.
Speaker 2: Um, I think it's a new channel. There's many, many channels.
Speaker 1: Yeah.
Speaker 2: Um, the, the brand journey may be changing slightly, but it still needs awareness. It still needs, um, consideration.
Speaker 1: Mm.
Speaker 2: And it still needs research, and it still needs loyalty, right? And I think the ChatGPTs of the world have changed a lot of the bottom of the funnel and maybe some of the middle of the funnel, maybe even discovery, right? But the messages that are being delivered through CTV or-
Speaker 1: Mm
Speaker 2: ... sight, sound, and motion, those are still extremely important. And I do see, I think we've seen an evolution of people through this crazy COVID time of remote work and being at home all the time to going back out there and really craving live human-to-human contact. And anytime you've got humans together, there's an opportunity to create awareness in a brand, I think.
Speaker 1: Yeah.
Speaker 2: So it's, it's a new channel. We need to adapt to it. Uh, OpenAI has APIs into a contextual signal, which is, you know, a big part of what we're doing.
Speaker 1: Yeah. And I'm wondering if that journey is changing the outcomes that marketers are looking for in terms of advertising for yourselves. Um, is there a shift to a certain extent to build brand, um, brand building as an example? Um, what are you seeing within the marketplace?
Speaker 2: I think that we're lucky that this is coming at a time where our company is ready to handle all the different pieces of that funnel. And so our product mix may change over the years, but it's all coming from the same place. We need to help marketers understand their consumers, how they move through the user journey, um, what starts that journey, what messaging works well. We can do all pieces of that. So for a, a few of our advertisers, they use us to really understand the best messaging, the placement of that messaging, the effectiveness of that messaging. So we can build deep learning algorithms that have, you know, just to get a quick into the nitty-gritty, we can, we can test 15 to 20 creatives. Uh, you can do 100, but really-
Speaker 1: Sure
Speaker 2: ... that takes a lot of, uh, time and effort to really do it the way we do. But let's say you have 15 to 20 creatives. They could be for different, um, consumer segments. They could be just different taglines.
Speaker 1: Mm.
Speaker 2: They could be different offers, all those things. And instead of doing like an A/B test and saying, "This is the best creative," we can say, "Okay, this messaging is working best over here. This messaging is working best over here. And even though this message only works well 3% of the time, it's an important untapped demographic that maybe you need more messaging for, and let's focus brand awareness in that area."
Speaker 1: Sure.
Speaker 2: So I think from an AI perspective, you know, you've talked on your show about, um, are, is this, are, are humans being replaced? Like, how's that gonna work? I think in marketing, you know, I've never heard a marketer say, uh, "I'm doing everything." You know, they always wanna do more.
Speaker 1: Sure.
Speaker 2: They always wanna find more.
Speaker 1: Mm.
Speaker 2: They always wanna sell more. And, and these types of tools, which are really big data tools, not LLMs, these are really taking huge amounts of data, making sense out of them. Humans cannot do that.
Speaker 1: Sure.
Speaker 2: I think that's an important part and a cool thing that we're doing, is that we're really enhancing what humans can do and accomplish because they can't take all that data and really, and really assess it correctly. Deep learning AI in this, in this case can, and that sort of helps us-
Speaker 1: Yeah
Speaker 2: ... all over the place on a marketing perspective.
Speaker 1: I'm also thinking about, um, premi- slight, slightly different topic, but premium content. I'm wondering if it holds up nowadays. Um, clearly over the years, I, I'm just gonna be brutally honest, frankly, that we've seen a race to the bottom in terms of publishers, um, using programmatic to sell inventory, you know, shift inventory, remnant, et cetera-
Speaker 2: Yeah
Speaker 1: ... um, and extremely low CPMs. Do you think there's gonna be maybe a shift back? Um, those guys are being hurt obviously by the LLMs as well at the same time. Um, but do you think that there's gonna be kind of a race to the top as well in terms of premium, better quality content, premium content, and premium ads served, you know, a much higher CPM against those domains?
Speaker 2: I think we I think the interesting thing about our approach to marketing and advertising is that we say to our clients, "Don't have a hypothesis about what, what works, what doesn't work-
Speaker 1: Mm.
Speaker 2: ... who to, who to target, who not to target." It's, you don't wanna come to us if you have to have males 18 to 49, right? Like, you can get that other places.
Speaker 1: Of course.
Speaker 2: We really don't like to have, uh, positions or hypothesis on those things, like MFA and all these types of things. Maybe it works-
Speaker 1: Mm.
Speaker 2: ... maybe it doesn't.
Speaker 1: Mm.
Speaker 2: And, and maybe it's not right for your brand, but if you're talking about selling more and reaching a certain new type of consumer, it might work, it might not work. Premium might work, it might not work. It might be too expensive. It might, it might, uh, not, not add enough value. You know, those are the cool insights that we deliver back to the marketers-
Speaker 1: Mm
Speaker 2: ... if they allow us to really explore the space.
Speaker 1: Yeah. People also talk about curation and, you know, perhaps that's overused. Um, is that really a change or is it just a repackage of something else?
Speaker 2: It's definitely, it's, it's definitely a switch, you know, from where the intelligence is, is, is living or was supposed to live. There's a, there's always gonna be, I think, a push and a pull, uh, between a buy-side tool and what's sort of called a sell-side tool-
Speaker 1: Mm
Speaker 2: ... but has evolved. Uh, we are active-- We, we have, we have a philosophy of you can get your brain wherever you need it to be. When we began the company, we had to build a managed service DSP because nobody could handle, uh, deep learning algorithms, and the feature set that we were building was just not hostable in DSPs. Um, on the SSP side, we host our algorithms in a co-location capability. There's a lot of talk about containers. Right now for us, containers are not large enough to handle what we need to do, the full models and all the features and all the, all the sort of data that we need to store. So there's an infrastructure issue there, but ultimately, um, I think it's a reaction to underperforming at the DSP level, uh, a feeling that it's a little bit of a commoditized tool set right now-
Speaker 1: Mm
Speaker 2: ... and all the inventory is the same everywhere, so why not move a little bit more to the SSP?
Speaker 1: Yeah. Well, that makes sense. Um, you've developed a product called ContextGPT. So what problems are you looking to solve with that particular product?
Speaker 2: Well, I think context, we t- we hear a lot more about context these days, and there's, I think, a number of reasons for that, and, and it's the same reason we developed this product.
Speaker 1: Mm.
Speaker 2: Uh, number one, the cookie apocalypse was coming, and we were all looking for, especially as a performance advertising technology, we were looking for something that would perform, uh, without cookies. And the reality of the situation in ad tech is that generally speaking, advertising, we all agree, is meant to be the right message to the right person at the right place at the right time.
Speaker 1: Sure.
Speaker 2: We've talked about that for a million years, before the internet existed. Ad tech has invested in one piece of that in every company, right? There's no company that answers all four of those questions at the same time. You're either an aver- you're an audience company or a context company-
Speaker 1: Mm
Speaker 2: ... or you're a DSP, which is the timing, or you're a creative, dynamic creative company. I think we're one of the few companies that looks at it scientifically and treats the auction and the impression on all four of those pieces at the same time.
Speaker 1: Mm.
Speaker 2: And so for us, we were already doing a lot of contextual understanding in order to inform our algorithms. We then pushed that out to, as its own product because of the cookie apocalypse, and we found a huge opportunity there-
Speaker 1: Mm
Speaker 2: ... which we really were surprised about. We were the first to really use prompts to understand how to target context.
Speaker 1: Yeah. And the flip side of that, where does it still fall short?
Speaker 2: Well, I think one of the things that we have worked really well on, I think we win on, is that context, true contextual, contextuality of somebody-
Speaker 1: Mm
Speaker 2: ... does not scale. If you're doing it honestly and transparently and you want exactly content about your shoe being used to break a world record, okay, you can spend $10 on that. How do you get the message across and i- in a s- at a scale that ma- makes a difference to your brand-
Speaker 1: Mm
Speaker 2: ... but is still relevant to that subject matter? We integrate with LLMs on top of our ContextGPT product, where I think we do a really good job of solving that tr- that scale thing transparently-
Speaker 1: Mm
Speaker 2: ... and allowing the advertiser to make a decision on the trade-offs between the exact thing they're looking for and the relevancy of, of scale.
Speaker 1: Mm. Yeah. I mean, that makes... Relevance I love as a, as a word.
Speaker 2: Yeah. I mean, look, y- you buy-- We hear over and over again that there's non-transparency in the contextual segments that they're asking for.
Speaker 1: Mm.
Speaker 2: And there's a little practicality there. I mean, it could be a million URLs in there.
Speaker 1: Sure.
Speaker 2: But we, our tool offers, and it's a slick demo, our tool offers full transparency into every single URL. You don't have to look at them, but you can look at the least relevant ones and understand, is that, is that social running clubs now and that's too far away from competitive-
Speaker 1: Mm
Speaker 2: ... running, which is where you wanna be? Or that's okay, you know. And I think that understanding really helps the agency users and, um, and the brands themselves be comfortable.
Speaker 1: Yeah. I'm also thinking about mismatches because, uh, something could be contextually relevant, but it's not a great place to, to place the ad.
Speaker 2: Yes.
Speaker 1: What do you guys do in terms of brand safety in that scenario?
Speaker 2: That's where we integrate with a number of MFA-type def- definitions-
Speaker 1: Mm
Speaker 2: ... like Jalons and things like that, and we enable our a- I would say that, you know, brand safety is an interesting thing that we, we talk about a lot, but I think it's sort of solved, frankly-
Speaker 1: Mm
Speaker 2: ... um, from the advertiser's perspective. They usually have their own tools that they've already set up, and they're like, "Here, here's our list of acceptable or non-acceptable sites," um, or, "We're using Jalons," or, "We're using something like that." So we, we integrate with all those things, but it's not really as much of a concern for us because, because we can get a detailed view of the language on every page.
Speaker 1: Mm-hmm.
Speaker 2: You can have sentiment analysis, you can have inclusivity analysis, you can have all these pieces that you can filter out everything.
Speaker 1: Yeah. And to an extent, we're touching on standards as well if we're talking about brand safety and so on. Um, I believe you spent five years at the IAB helping shape, um, standards and measurement and, and so on. Um, what do those standards do well, or are there some that actually prohibit and hold the industry back?
Speaker 2: I think that the IAB and the standard, uh, making organizations across at least Europe and the United States, because I'm most familiar with them, have done a great job of, of bringing standards together when there's a need. Um, my experience has been there was chaos, and then we needed standards, and I think it did a great job of that. Like-
Speaker 1: Mm
Speaker 2: ... when I started at the IAB, we had a, I don't know, 300-plus ad sizes that people were selling. AOL and Yahoo and MSN had their own in-
Speaker 1: Yeah
Speaker 2: ... ad sizes.
Speaker 1: It was crazy.
Speaker 2: So the advertiser was like, "I can't, I can't function on this. I can't cr- create 300 ads, and I'm not gonna spend any money."
Speaker 1: Mm-hmm.
Speaker 2: All right. Well, we fixed that. We got 10, 20 ad sizes. So the 3x600 was created, the 3x250 was created. None of those little tiny buttons were, were in there.
Speaker 1: Yeah.
Speaker 2: It was very... It was a great time to be at the IAB because it really was the Wild West, and we really helped open up the floodgates. So one of the most successful standards I worked on with Ari Paparo actually was VAST, which is this, this technical video, you know, communication. It's like open RTB basically. Like, it enabled everybody to standardize the communication between the player and the ad server.
Speaker 1: Mm.
Speaker 2: Didn't matter which player you had, didn't matter if you had your own or somebody else's. All of a sudden, you could buy and scale video advertising on the internet, and it just was adopted like wildfire. So all those things just help. I think the IAB's very good at helping open up the floodgates for-
Speaker 1: Yeah
Speaker 2: ... for, for advertising.
Speaker 1: So you've been building the business since 2015, as, as we've talked about. So that's, what, a good 16 years now. Um, you're still independent, which is great. What, uh, yeah, what has that choice in terms of staying independent cost you versus what it's brought you?
Speaker 2: Well, we're, we s- You know, it's interesting. If you look at the, let's call it the, uh, class of 2015, 2016, 2014 in ad tech-
Speaker 1: Mm
Speaker 2: ... there's almost nobody in it. Uh, we started the company where AI was starting to be a buzzword with the venture capitalists, but advertising was not good.
Speaker 1: Mm.
Speaker 2: Um, and we don't have any VC backing, so there's independent and then there's independent. I think we're really independent in that way.
Speaker 1: Mm.
Speaker 2: We, we were lucky enough to find really great angel investors, but we had to be very capital efficient, and I think it, it definitely slowed our growth, but I think it allowed us to really build a company at the right, um, time, in the right timing.
Speaker 1: Yeah.
Speaker 2: Because again, like we've said before, 2015 was, um, the beginning of this deep learning AI thing, and so we had to wait for a lot of things to catch up to us. Um, and that allowed us to take the time to do it.
Speaker 1: Mm.
Speaker 2: So it was really, we started off growing slowly, and then after COVID, it really all took off.
Speaker 1: Yeah. Cool. Yeah, that's a great story. Um-
Speaker 2: I think there's always luck involved. You know, we, we had a s-
Speaker 1: I think you make your luck, don't you?
Speaker 2: We, we did, but-
Speaker 1: Yeah
Speaker 2: ... you know, it could take you years to find the right first clients, and it didn't take us-
Speaker 1: That's true
Speaker 2: ... that long. So we, we had a great, you know, head of sales CRO back then.
Speaker 1: Mm.
Speaker 2: And, and he, and it, it, it all worked out. It all worked out.
Speaker 1: Yeah. I mean, that makes a huge difference. It's so hard to find great salespeople.
Speaker 2: Yeah. Absolutely.
Speaker 1: They're, they're unicorns, frankly. Um, let's do a little bit of future casing. So, you know, if, if we're moving into a world where agents start doing the buying, what happens to the people whose job is media buying?
Speaker 2: Yeah, I'm interested to see how that evolves. I'm, uh, maybe I'm too old, but I think that sales in our industry is still very relationship-based.
Speaker 1: Mm.
Speaker 2: Delivering the difference in value proposition takes time and effort.
Speaker 1: Mm.
Speaker 2: It's not an instant thing. I think if that, if you want to automate a long-standing relationship instead of, you know, sending an IO and having it approved and then, you know, and this is an ongoing relationship-
Speaker 1: Mm
Speaker 2: ... okay. I think it's gonna be interesting to see if, if this, the agentic selling can really replace fully a sales team. We'll see.
Speaker 1: Mm. So you still think there's the value in the human relationship?
Speaker 2: I think that... Well, let's, let's take a step back from my sort of buyer hat and-
Speaker 1: Sure
Speaker 2: ... and go back to the publisher side, which I started my career in.
Speaker 1: Mm.
Speaker 2: I think that all premium publishers Have very strong brands, uh, values and, um, and offer something special to an advertiser.
Speaker 1: Mm-hmm.
Speaker 2: And it's very custom. Uh, so I don't know. I, I can't see how that will, will be misplaced, uh, or, uh-
Speaker 1: By AI.
Speaker 2: Yeah.
Speaker 1: In the near future. We'll see.
Speaker 2: Yeah.
Speaker 1: And then five years out, what parts of the, I'm gonna say the trading value chain disappear or, or change dramatically?
Speaker 2: Well, obviously chatbots are gonna become a large part of our lives, and sort of already are.
Speaker 1: Mm-hmm.
Speaker 2: Um, we have a whole generation of people growing up that are using them for everything from planning vacations to buying everything.
Speaker 1: Sure.
Speaker 2: But on the flip side, malls are doing better. Younger people are realizing that it's nice to touch and try on the clothes-
Speaker 1: Mm
Speaker 2: ... you know, when you go there and actually see, see things. So I think we have two competing, um, pieces of the user journey and how that is affected by outcomes and out of home and-
Speaker 1: Mm
Speaker 2: ... all these other pieces that don't get talked about a lot, I think is, is still up in the air.
Speaker 1: Yeah. I see the purpose of stores changing, frankly. Um, being more of a showroom is probably a nice way of describing it. Uh, and certainly where you can get comfort and make comparisons and, and see things. Uh, and where experiences are formed and created and bec- I can see stores becoming a lot more experiential. Um-
Speaker 2: Again, it's a brand
Speaker 1: ... for instance. Yeah, exactly.
Speaker 2: Okay. If brand didn't matter, On and Hoka wouldn't have been able to build huge brands against Nike and Adi- Adidas and all these other companies. So brand is always gonna be key, and you're not gonna build the brand inside a chatbot.
Speaker 1: No.
Speaker 2: So it's always gonna be... They're always gonna be pushing, uh, the newest way of interaction.
Speaker 1: Yeah. I mean, you obviously need to brand everything. Without that you have nothing, right? But, um, if you don't have brand, you can never be front of mind and be on that consider- you know, that, that point of consideration.
Speaker 2: Right. I mean, the question is, is the human decision-making process going to change? Let's, let's call my... Let's call me a student of history. I think that, uh, uh, you know, things like work, right? Take work. We thought, "Oh, remote work. We have Zoom now. It works. We're gonna just be remote forever." I think we've realized that there's a reason why we've gone to offices for a million years-
Speaker 1: Yeah
Speaker 2: ... you know? And I don't see those fundamental human and, uh, uh, genetic needs changing.
Speaker 1: Mm.
Speaker 2: So interaction, experience, understanding, touch and feel, it's all gonna make a difference. You can buy something online, but then you have to return it, right?
Speaker 1: Yeah.
Speaker 2: And, and, and, and getting them to keep it is the key.
Speaker 1: Mm. So we've, we've obviously touched on brand there. I'm just gonna ask a few quick fire questions.
Speaker 2: Yep.
Speaker 1: So is there a brand or business that you kind of admire at the moment and you like the way that they're kind of showing out and, yeah, and, and h- and, and how they're putting their message out into the world?
Speaker 2: Well, there's a lot of them these days. I, I do have to say, even though I'm in ad tech and I'm in the sort of the nitty-gritty of it, I still really, really appreciate great advertising. Um, uh, I'm having to watch a lot more ads these days on the streamers, so I get to, I get exposed to a lot of different things. I actually like what Walmart's been doing. I think they position themselves really well. I think Walter Goggins is a great spokesperson that really represents and, and identifies with a huge portion-
Speaker 1: Mm
Speaker 2: ... uh, across ages, across backgrounds. So I, I like what they've been doing. For some reason, I'm not saying their message is really sticking something different. It's just the look and the feel of Walmart ads I think has, has really done well this year.
Speaker 1: Yeah. Yeah. I'm just thinking about their, um, commerce media business is enormous for them as well, isn't it? But yeah, their advertising is, is exceptional.
Speaker 2: Look, Walmart is a, an incredible product.
Speaker 1: Mm.
Speaker 2: Uh, if you haven't, uh, gone to Walmart, um, it's, it's the only game in town for most people-
Speaker 1: Mm
Speaker 2: ... in the United States, and you can get at least one of everything. If you need it, Walmart has it.
Speaker 1: Yeah.
Speaker 2: And it has it in stock. You don't have to order it. So it's one of those things, again, where you can go, you can have a one-stop shop, and you can get everything you need. There's, there's very few c- companies that can say that.
Speaker 1: True. Um, a go-to resource, so, you know, in terms of keeping yourself sharp, uh, helping you with your thinking or helping you perhaps to think slightly differently, do you have any sort of methods or, or go-to resources, or it could just be a place that you, you visit that just does something for you?
Speaker 2: I've always, I've always found inspiration, um, to be a, a cross-sport thing. Like you have to, you have to expose yourself to a lot of things outside of what you do.
Speaker 1: Mm.
Speaker 2: And I'll just say, I'll just say this. Every year I go to South by Southwest. Um, I don't go with a team. I go by myself, and I go to non-advertising panels, and there's always something that I can connect to whatever it is, whether there was a panel on how cat ladies become catrepreneurs or the, the Mars lander, you know, I... Or, or, or a new show that gets, gets, I get to preview.
Speaker 1: Mm.
Speaker 2: There's always something that gets me going, and I think that's been really important in my career is to really get cross-exposure to just the world-
Speaker 1: Yeah
Speaker 2: ... in general. And, and frankly, understanding- Humanity. Understanding all walks of life and, and really, that to me as a marketer, advertiser, has been really important.
Speaker 1: Yeah. I, I agree. I think that is so important.
Speaker 2: Being a person of the world versus, you know, in an ivory tower or something like that.
Speaker 1: Yeah. Uh, and then lastly, a piece of advice that you'd provide to a younger individual, perhaps they're at, they're at the start of their career or looking at the, the world of marketing, advertising, media. Um, what piece of advice would you provide to them?
Speaker 2: We talk about this a lot now because of AI tools and the types of, uh, work that they're taking over, and my advice, you know, it's interesting. There's this, there's this debate going on about specialization and how a lot of these, uh, educational institutions need to specialize more, and you go into university and you get a more of a vocational degree, like in marketing or economics or this and that.
Speaker 1: Mm-hmm.
Speaker 2: I can't disagree with that more. With AI, specialization, the AI could take that specialization away from you tomorrow. The key in how we're using the AI tools, especially in coding for instance, is that you have to be able to see a problem, design a solution to the problem, and really understand the effectiveness of that solution.
Speaker 1: Mm-hmm.
Speaker 2: So critical thinking skills, flexibility of thinking, problem-solving skills, those are the key to any career's future at this point. You have to stay flexible because you never know what's gonna happen to the, the job that you've decided to take on.
Speaker 1: Mm-hmm. Well, fantastic. Great words of advice. Thank you so much for being a guest on the show today.
Speaker 2: Thank you. Thanks for having me. This has been awesome.
Speaker 1: Yeah. I've enjoyed it. Thank you.
Speaker 2: Thanks.