Privacy First Advertising with AI with Kartal Goksel

Russ Fordyce 🏆 (00:01.064) Welcome back to the winner's circle, everybody. I'm here today with Kartal. He is from Seedtag. They are the latest AI Excellence Award winner, and they are going to talk to us about advertising, programmatic, contextual, identity-based, AI, what's going on with all of this. They are at the forefront of advertising in this space. Kartal, welcome to the awards, or welcome to the winner's circle, and congratulations on the award. Kartal Goksel (00:25.848) Thank you, thank you for having me. Russ Fordyce 🏆 (00:27.838) Before we get started, give us a little bit of background on SeedTag. I know they've been around for, I think, wow, almost 10 years, or a little bit more than 10 years. Give us a little bit of history and kind of what you guys are focused on. Kartal Goksel (00:38.786) Yeah, absolutely. The company has been around a little over 10 years. And I've been with the company two plus years now. The company has always focused on, from the earliest days on, to contextual advertising. They started with in-image advertising in Europe out of our Madrid headquarters. And the first technology was how to align the ads into the image ads contextually so that the ad itself doesn't look any more different or like out of context with the image. With the privacy coming on and believing in like the privacy-first advertising, the founders shifted to a larger contextual advertising covering not only in-image advertising, but like in-text, like the regular in-article advertising. And here we are after 12 years. Keep iterating on the technology and getting it to the next level of precision and next level of cognitive alignment with the user. But the focus stays the same. Like how do we stay contextual? How do we stay true to the privacy first? And that's the mission and the vision. Russ Fordyce 🏆 (01:51.066) It's such an interesting time for advertising as a whole, because there's so many shifts going on. There's AI on either side of this equation from the user perspective is now in chat GPT and Claude and not seeing ads at the moment. This is recorded in April 2026. May 2026, we'll probably see something different. But you've got kind of the consumer side, you've got kind of the publisher side being able to just do so much. Kartal Goksel (02:12.046) 100%. Russ Fordyce 🏆 (02:19.08) with these tools now. It's almost like you've got, bad analogy, but it's an American one. It's almost, you we used to be doing shotgun approach. You'd kind of spray and pray and you try to hit, you know, a lot of targets and maybe one would pick up and you hoped your ad was contextual. But now we're really getting down to those rifle shots where I can say, this person's looking at that. I know them, whatever. know their function. And now we can really deliver that personalized ad that we've all been promised for, for so many years. Walk us through kind of that contextual model that CTag's so focused on and why it's so different than identity. Kartal Goksel (02:55.63) Yeah, mean, it's like, you know, identity-based advertising had its values as well, right? And it depends on, like, what kind of advertising you're doing, too. But in essence... I think what the contextual does for me, and this is primarily why I joined CTAG two plus years ago, is because it resonates with me. When we think of our brains and I am engaged with the content, whether that's a text in an article or I'm watching something and I'm listening to some news and I'm engaged with the content, my brain is focused on it. Anything that is out of that context creates this in my opinion, a cognitive friction in my brain. And our brains is not much different than the models we build and train, right? Over time, we learn to ignore anything that is out of that context. that's what I believe, truly I believe, and we believe as the company with our technology is that we eliminate that cognitive friction when the ad itself contextually aligns with the content in the right time in the right moment, right? So I think that then our brains interpret that ad within the content and we have the attention going to it rather than the ad that is out of context or the ad that sort of somehow some algorithm is detected something about me 15 days ago and now something's showing to me like our brains are used to like ignoring that. And that's where we think like the contextual advertising. is much much superior than anything that you can do. Russ Fordyce 🏆 (04:38.986) Well, you know, I didn't really ever think about it, but I used to do training around this for novices of advertising. And the question I always get was, and I built a whole presentation around it, was why do those shoes follow me around the internet? You know, if you look for a pair of shoes or a canoe or whatever you're looking for, you might be on New York Times, you might be on, I don't know, know, Barstool Sports or whatever. Kartal Goksel (04:53.666) Yeah. Russ Fordyce 🏆 (05:07.114) you're looking at and then all of sudden shoes show up and it is like you're saying it's cognitive break. You're immediately in this ADHD world we live in. You're immediately taken away from what you're doing and given this thing that doesn't really play a role in what you're looking at. So it's got to be super powerful for both the advertisers themselves and the publishers especially. I would imagine they love it because now you're not leaving their site. Kartal Goksel (05:29.294) Yeah. Kartal Goksel (05:33.679) 100%, it's also not about leaving the site, right? To me personally, when I get into that, like the friction mode, I'm trying to read something, I'm concentrated, I'm engaged, and I'm seeing something that's not totally not aligned, that's kind of a turn off for me for the content as well, right? And then today's, the publisher dilemma is like, There's so many publishers out there, like as humans, we generate so much content. Like there's always the next one that I can go to and consume that content. So I think it's a good tool for publishers to keep the engagement with their own users by not distracting them from the content, by keeping them focused on the content, but also still giving the good ROI to the... to the advertisers so that at the end of the day, we know the advertising in ways bankroll all of the free content that we're consuming as humans. And part of my sort of a personal belief and the mission is like, I want to keep that content free and available for the next generation for my kids because I think when we look at our understanding of the world prior to Open Web and our understanding of the world as human race, With the enablement of the open web and the content that we can consume, we need to keep that going for the next generations for not going backward. Russ Fordyce 🏆 (06:55.274) Now, you guys have had this model for, like we said, almost 10 years. This new generative AI has been around for a year and a half, two years. If you go back further, it's a little bit there. But really, practically, last few years, you guys have built this new engine, Why don't you give us a little bit of a background on Liz? for somebody outside the advertising world, what's the simplest version that you can explain Liz to us? Kartal Goksel (07:24.6) Yeah, I mean, if I wanted to explain it in the simplest version, we have the models that detect the interest, emotion, and intent from the content. If you think of the interest, what kind of categories this may be aligning in, you can take that further down and make it more granular. What are the primary or secondary emotions that we can predict the content with? generate in the human being who is engaged with it right now. And then the third component is the intention, the intent of the content. If I'm an enthusiast, runner, I do like running. So if I'm reading a content about comparing this year's releases of running shoes that are geared towards marathon running, The intent of that content is obviously different than if I'm reading an article about New York Roadrunners, New York Marathon. So if you sort of, by Liz, what we do is we detect all of these signals and we blend them together to build custom audiences that you want to achieve as the marketer. that's at a very high level that I can explain. Obviously underneath it is complex embeddings and fine tuning of domain-specific models and et cetera, but at a high level, that's what we're trying to do with the contextual advertising. Russ Fordyce 🏆 (08:51.434) Now, we got started a little bit early and you mentioned that you were in the early days of CNET and working there. We've built this entire industry around third party cookies, right? These things that live on our machines until we remove them. And it goes even beyond that folks to your machine IDs and other things that we won't get into. But these third party cookies were really the basis of this billion dollar industry. And we know Google is now Kartal Goksel (08:57.24) Yeah. Russ Fordyce 🏆 (09:18.548) trying to kind of get rid of those cookies. They've been trying to get rid of them for years. What did that moment feel like from inside an ad tech company that had been building an alternative for over a decade? Kartal Goksel (09:29.71) I mean, obviously, what are Google? So there's what the technology companies do, right? Apple was the first one who came out and said, I'm going to eliminate these cookies. And the mobile phones and the Apple phones become a no-go for a large swat. of users that the advertisers, I would argue, actually needed to reach, right? Then the Google were going to follow it, they pivoted, but at the end of the day, in my opinion, the end user's point of view doesn't change, right? The end user does not want to be followed on, the end user does not want to be shown the shoes that they were looking at 15 to 20 days ago. The end user wants to be more privacy conscious, especially the newer generations, right? I'm in advertising, you I accept the cookies myself because I'm like, all right, show me the ads. At least I can study and sort of see who's doing what. But like when I look at my nephew who is in his late 20s in his generation, they outright reject. They're much more conscious of their privacy and they're much more technologically savvy than my mom's generation who wouldn't even know how to turn off the cookies. Where we're going is foregone conclusion and it's determined whether Google keeps the cookies or not. End users want privacy first, privacy compliant advertising. They don't want to be followed and they don't want to feel like somebody's watching them. Russ Fordyce 🏆 (11:09.864) So you got. Kartal Goksel (11:10.028) So I think from that perspective, we're at the right place at the right time, whether Google deletes the cookies or not, right? Because that's where we're going to end up anyway. Russ Fordyce 🏆 (11:19.486) Well, and it's going to be interesting because the traffic's rapidly moving away from kind of search and into these tools themselves. I don't know, the last time I did a Google search and actually clicked on a result. Usually you get an AI prompt in before it. But the premise behind neuro-contextual advertising is that what you're reading right now tells the advertisers more than they could ever know about you in behavioral history. Now, how do you make that case to a marketer that spent their entire career optimizing for audience targeting? Kartal Goksel (11:44.056) Yeah. Kartal Goksel (11:50.211) Yeah, well, first of all, mean, it's like depend on the age of the marketer, I guess. If you're at my age, there was a time that you didn't have to do that, right? That's prior to, I would say like prior to programmatic advertising, now that you mentioned early days with CNET, people were buying context anyway, right? They would just go. publisher by publisher and figure out where they buy and how they run things. And that was like there wasn't a contextual technology, but the human brain was trying to execute contextual advertising. That was the same thing with the TV advertising before it, right? The programmatic advertising for the last, 12 years or like a little bit more than that, a decade and a half is the one that got us addicted on the third party cookies. So I mean, the way the way you want to always things is like the proof is in the pudding right we do we have benchmarks around like the as simple as the CTR the VTR to for every campaign we run like a four campaigns we run measurement studies where you can kind of you know do the measurement studies look at brand recognition product recall brand recall brand lift favorability And then for the neuro contextual, we've already also run a study with a scientist out of Columbia University where they actually sort of measured what's going on in the human brain with the electro recognition. where the people are seeing neuro-contextually aligned ads versus contextually aligned ads versus non-contextually aligned ads. And you clearly see in the study that the proof is the study, the approach is the right, because there is definitely a more favorability from the end user perspective when they see something really aligns with the content. So it's the sort of, again, going back to eliminating that. Kartal Goksel (13:48.356) the cognitive friction that you have to live with where you have an ad that's stuck in the middle of the content that you're engaged with, but it's not really aligned with it. Russ Fordyce 🏆 (13:59.378) Yeah. Now this privacy of kind of first advertising is, as you said, it's, it's, you know, it's become a constraint for a lot of ad tech companies. You know, they built their reputation around that and see tags made its core kind of design principle. This privacy first contextual advertising. How does starting from privacy first kind of change the architecture of what you build? Kartal Goksel (14:21.967) Yeah, I mean, think the good news for us is we care about the content. We care about predicting things based on the content, not about the user. If I'm reading something right now, it's all about what I'm reading right now and what kind of interests. and emotions that it could predict, like we can predict that it would bring in me, right? It's not about my demographic. It's not about like matching my identity to something to figure out like what kind of household I'm living in, my household income, and et cetera, right? If you just take the content as your centerpiece, your privacy first, implicitly because you don't care about who the user is and you care more about the content and interactivity between the content and the user. Russ Fordyce 🏆 (15:23.466) Now, the content and context are pretty tricky things. They're very human kind of capabilities of understanding the context from the content, right? It's not a keyword matching system, is it? So kind of give us a little bit of a description of how they're different and kind of, you know, where does that gap kind of exist in performance and how you approach this stuff? Kartal Goksel (15:37.412) Yeah. Kartal Goksel (15:48.794) Yeah, I mean, if you roll back to how the technology evolved, right, and sort of that kind of predates me at SeaTag, but I'm just talking about industry and sort of the contextual technology in general, obviously everything started with keywords, but then... you do like a keyword frequency-based models back in the day, and you're trying to figure out what the interest may be, right? But as the technology evolved, as the natural language processing evolved, and we went into like the transformer-based models, now we have like embeddings. Now, And with the new capabilities of AI, we have a very good understanding of the full context of the content, rather than just going into more things around, like keyword frequency and whatnot, that may sort of deviate, or that may reduce your accuracy on what you predict. So I think that's where it started getting more interesting for everybody. And then you sort of do the advertising-based fine-tuning. on top of it. So you can look at the context, in the context of the general of the world, right? So you can sort of say, what would be an example? You can sort of say, like, we want to try to figure out, like, a luxury high-end auto enthusiast, right? And then you can sort of look at articles ranging from, like, the newest luxury cars releases and descriptions of that to Bahrain GP getting canceled this year to Grand Theft Auto and you're in this embedding space and then you have to sort of say like, maybe if you were to do it in a way that you're doing more keywords or like a looking subset of the content, maybe Grand Theft Auto could sort of like have enough in it that could align it more with the luxury enthusiast by definition than Bahrain GP. Kartal Goksel (17:55.47) But if you sort of put your advertising hat on to the model and find you in that model, you're going to sort of need to change that embedding space in a way that, yeah, somebody who's into Bugatti's and Ferrari's probably is a better candidate to read something about Bahrain GP getting canceled, then reading something about Grand Theft Auto. So I think that's where the whole sophistication comes in. And that's where the newer tech. let us do what we do in a higher precision than we were able to do 10 years ago. Russ Fordyce 🏆 (18:33.426) Now you mentioned that you went off to Columbia University and had a study run. I think you guys worked with the neuroscientists there to kind of figure out how that brain does respond to a contextual versus non-contextual ad. What made you go that far to kind of validate that hypothesis and what did you find that you didn't expect there? Kartal Goksel (18:54.307) think we're a quality obsessed company. From the founders of the company to all of us trying to build the tech, I've been in this industry for a long time. I've always taken pride at wherever I have been to build the most accurate, as best as I can to bring the technology. build the technology and create the marketer's outcome, right? So we're in ways, we're in ways quality obsessed. So that's number one, right? We needed to validate what we believe in, right? We believe in it, as you can tell probably from this conversation, I believe in it wholeheartedly, right? So I think we needed to validate that, but you also need like in our... in our industry and in the world, you also need to have somebody else always grade your homework, right? For the purpose of the third party who's engaging with you and buying the product that you're selling to, even for your own sanity, that somebody comes in and sort of validates what you're building and grades your homework. Russ Fordyce 🏆 (20:08.436) So from the way I understand this is Liz is actually looking at the text, the images, any video that's there, and plus all the metadata that we surround all of our content with these days to get all the AIs and Googles to find us. you're now from what I read, you're doing this across languages and it's got to be sub-second latency between reading all of this and then executing an ad or else that ad is going to be there or not firing, not firing. Kartal Goksel (20:32.271) Of course. Russ Fordyce 🏆 (20:38.6) What's the hardest thing from an engineering perspective inside of that system to kind of complicate, you know, that complicates all this? How did you get there from an engineering perspective? Kartal Goksel (20:48.089) Yeah, yeah, I mean, there's a lot of complications. Obviously, the technology is making certain things easier for us, right? Five, 10 years ago, building. multi-language models were a lot harder than we're able to do today. We're in 17 different countries and different regions and I think like 10 plus languages across and different dialects of the languages as well. So some of that complication is getting eliminated by the technology. with AdTech, the AdTech scale is always your biggest complication. So if you think about it at a... at a higher peak loads, we could be getting anywhere from like 700,000 queries per second, right? That is billions and billions of requests a day. We need to work in a way that are like the content owners, the publishers, the broadcasters on CTV, they only give you milliseconds, right? Anywhere from 100 to 500 milliseconds to do what you need to do. And that includes the network latency, right? From my phone. I'm reading something, the ad is making a request, and it comes to my data center. And I lost that time, right? Because the network time is nothing in my control. And then now I need to get in and run a of predictions in real time. And each of these things, we give about 10, 15 milliseconds to provide an answer. Or we just got to have to move on, because the publisher is not going to give us that time, right? To be able to do that, at that scale, at that latency, and also control the costs without running millions and millions of dollars of cloud costs every month is where the challenge comes in. You have to really make sure. Kartal Goksel (22:46.138) how you build your models, how fast you build your models, what is your latency, all of your caching layers that you need to introduce, all of the high-scale deployment that can scale up and down with the traffic of the internet and the traffic of the CTV. Advertising works is a huge engineering challenge. that's actually my pitch to every single time we're trying to hire an engineer that you usually sort of do this whole pitch because like some of the engineers may be like, like, what is the challenge of that? Like, you're just putting an ad on the page. And I'm like, no, no, no, hold on. There is a huge engineering challenge that has been keeping my brain active for like 25, 26 years. Let me explain that to you. And usually, know, anybody, like, once you break it down the scale and the latencies and all the barriers, it's sort of everybody buys into it. Russ Fordyce 🏆 (23:43.688) Well, it's super impressive. I was in voice for a long time and you know, when we were doing voice over the internet and you knew that if that latency existed on your internet connection, you hear it. And when you hear the latency, there's only one thing that happens, which is you hang up the phone and it's no different from what you're experiencing. If those ads aren't popping and you got big blank holes in your content, first of all, the publishers aren't going to come around and the people won't stay either. You know, you think something's wrong with the site or something. But so huge, huge kudos to you guys for doing that. It seems quite unbelievable that you're doing that much contextual awareness. But then again, we have data centers and CLAWD and all these LLMs now to help us, which is pretty amazing. Now, you... Kartal Goksel (24:13.293) Exactly. Kartal Goksel (24:28.292) Yeah, yeah. I mean, the data centers are like, was a, like you and I, were just talking about this right before we went into the recording. think like the earlier in the days, like you had to run out of one data center. You had to rack hardware in it. costs a lot of money. think like, you know, like I said, some of the things that we do becomes a slightly easier. Like I think like the having a hardware, the cloud based hardware is one of those things with like the, we're big partners with like Google. They're deployed all around. world so I can have many endpoints and be as close as the end user as I need to be and elastically expand and contract and control my costs. So those things, some of the things got easier. in our industry over the last two decades, some of the things keep getting harder and harder because the scale goes up. We were maybe handling, I don't know, a couple thousand QPSs back in the day. Now we're at like 600, 700,000. So it's like the scale changes. It always creates new challenges, but there are some things that just makes life easier as well. Russ Fordyce 🏆 (25:40.136) Now you guys have done this amazing kind of back end of building the right ad for the right content on the right site and getting the results. But you've also now turned this back around and flipped it to the client side with this Liz agent. Walk us through kind of the insights behind that decision and tell us a little bit about what clients are actually doing with that that they couldn't do before. Kartal Goksel (26:01.86) Yeah, so. We had always capabilities of building custom audiences. So we had a web-based way of interacting with our NLP models and tried to build custom audiences. But the world is moving more and more to the agentic workflows, from our personal lives to what we do at work. The advertising that we do is not different to it. So with the Liz agent, what we're trying to do is to really go from planning to activation as fast as possible and with the lowest touch as possible. So we build an entire agentic workflow system that sits on top of our data, on top of our models. And if you want to build a custom audience, you can come in, interact with it in natural language, give your brief, and let it run. And it will come back to you with an audience. You can sort of interact with it, tweak it. If you want, go back and forth, and then once it's ready, you can sort of go for an activation. So that's our launch to it. But in my belief, the industry is going to move more and more agentic ways of buying things. And we will have the CTAGS agent interacting with the... the marketer's agent or the agency's agent, the world is moving there and we need to be ready for it. So this agent is getting us ready for that world to come. Russ Fordyce 🏆 (27:37.106) Now I read in one of the nominations that you had an airline, we don't have to say which one, but you actually were able to really kind of get down into the grid of it and show the lift that they were looking to achieve. It was pretty astounding. Walk me through that campaign, how it worked, and kind of how did you attribute the results of your approach versus just the media bot? Kartal Goksel (27:50.479) Yeah. Kartal Goksel (28:02.02) Yeah, but I think like, you know, there's a couple of things. One is, you know, the agent itself. And the models are like two sides of the same coin in a way, right? They need to work together. need to you can't have one without the other right you have to at the end of the day you have to have like a really good models and a really good like accuracy in what you need to do and then what the agent does for you is like the agent can go into different evaluation paths while building an audience figure out like what how that airline or another airline resonates in the articles that we see, what is the difference between that, what kind of like spaces that would be the most best to align within this product that they're launching. is much, much better than the human can do. And also, one thing is it compresses the time to activation. But set that aside, as humans, when we're activating something and we're activating it manually, we cannot go into as many paths. as the machine or the agent can go in and figure out and get the right sort of custom audience with the right tailoring in a way, right? So this is where I think we're going to become a very customizable as an industry and we're going to create outcomes. So in that case, the metrics were what was built for, I think, like I'm trying to recall the metrics. I don't want to. Kartal Goksel (29:43.346) recall anything out of line, we looked at, think there was like a study made, the brand favorability was like a 20 to 30 % lift, somewhere at 20 to 25%. I think there was about a 50 % lift on the recognition of the product that the airline was trying to push. So it just basically outperformed anything that the combination of the neuro-contextual models. And the agent, an agentic-based activation on top of it is basically outperformed anything that's without the agent or anything that's without the neurocontextual. Russ Fordyce 🏆 (30:25.81) The interesting thing to me in all of this conversation is that as consumers we've always had this kind of distrust of digital advertising, of most advertising. They're trying to convince us of something. They're reading my emails. They know where I live. They're on my phone. Do you think this contextual advertising can genuinely change that relationship? Kartal Goksel (30:48.689) I think it does. It does it for me at least. It does it for the people that I talk to. And it does it for the people that we run measurement studies with. So I think when I... When I'm talking at my home and all of sudden I see an ad that is relevant to it in one of the channels where I'm in, I automatically distrust that, right? Because I'm like, know, like statistically speaking, is like, this is not a coincidence at all. There's something going on in there, right? I need to go into my phone and see how I can clean up whatever. I don't want to be in that position as the consumer, right? But if I'm... reading an article and I have actually affinity to certain type of content and a certain type of publishers. If I see something that is like very relevant to the moment that I'm in and it's not intrusive to my content consumption, I actually, it resonates with me. I have bought things in the past. I sort of gravitated towards brands myself in the past where it caught me in that moment. And I think to me it's like, working the same way at the human consumption at large. Russ Fordyce 🏆 (32:04.286) Now you've described this kind of neuro contextual becoming kind of core planning layer of campaigns and we're not just going to use it for targeting anymore. This is kind of the baseline. What does that advertising planning process look like in five years if your vision plays out? Kartal Goksel (32:20.399) Yeah, mean, in ways like the planning and the targeting, they can't deviate from each other, right? So I think like the you need to plan and then afterwards you need to execute what you plan. And that becomes like the actual sort of a campaign portion of targeting and optimization, whatever we want to call it. But the planning is going to be, you know, when you're planning a contextual advertising, you have to access the same models in a way. that it's going to execute for you. So I think to us, and making that agentic, and making that fast, and you are able to iterate much more on it, or the machine is able to fine tune it much better than a human can, and customize it in ways that we don't have time as the planners. So that is going to change the game, in my opinion. And that whole planning is going to be agentic. That's my prediction, right? I mean, it's like, yeah. I mean, I think it's a... Russ Fordyce 🏆 (33:19.4) Yeah, I think you're right. Well, that's the other big thing, right? Kartal Goksel (33:31.282) Those portions, those, like we have processes in our industry that is very much human dependent. It has been human dependent since long time. One thing that I observed is like in any turn of the advertising, digital advertising until now, we always talked about how to make it more optimized and how to make it less human dependent. And that was like the premise of programmatic advertising at its inception. But. What I observed is we made it even more human dependent when we had the buying machine, we had sophisticated buying machines, selling machines, DSPs, SSPs connected together. But at the end, that sort of blown up the dependencies of human-based processes across the whole ecosystem. think now we're at a point that I think we're going to be able to. turn that around a little bit and optimize those processes, make it less human dependent. And I think it's gonna generate better outcomes for marketers, it's gonna generate better outcomes for the content owners, publishers, broadcasters, name it. And it's gonna make the whole thing much more smoother and less friction than it had before. Russ Fordyce 🏆 (34:47.21) Well, and I would imagine the error rates go down. remember, you know, we were running a campaign against a competitor and the competitor's name was, was, unique to us and very unique in the space. wasn't something like a, you know, a, a, a name. was a number and a letter. And so we put that number and letter into our, you know, into our contextual advertising keywords, right into our keyword engine. And we immediately started getting car ads and car people from car. companies and we got tons of traffic wasted tons of money to people who weren't looking for a phone system they were looking for a car and and that type of you know error has to be just eliminated now with with the context Kartal Goksel (35:27.621) Yeah, obviously. mean, I think that kind of error will be eliminated with the context, right? Contextual models that can look past that keyword or a keyword frequency. there's a human error there that would be eliminated based on the machines doing a lot of the buying as well, right? I mean, we do see this all the time in the programmatic advertising. The marketer wants to buy something. One system is trafficking it one way and another system is trafficking in another way. The algorithms are like fighting with each other at the end of the day. The money gets wasted. So I think that when you turn when we turn into an agentic based advertising where the machines are talking to each other and they have the protocols in place and you're removing that human error, it's gonna it's gonna be much better. Russ Fordyce 🏆 (36:24.67) So I think that's music to brand marketers and media buyers' ears. And if they're listening and they're still running those kind of primary audience targeted campaigns, what's the conversation they need to have with their agency or their CEOs that they probably haven't had yet? Kartal Goksel (36:39.749) Yeah, I think they, to me, they know, like, there is, there's a couple of things, right? There is the science behind it. There is the how you wanna be interacting or resonating with the generations, right? The brand marketers, they wanna be resonate with the next generation, right? We know the next generation is put off by, identity-based targeting where they feel like somebody is watching them. And then you also, for a swat of devices where there's like a cookies or identities are not available, but they're very well. your target audience, you're not getting to them as well, right? So I think they should definitely go talk with their agencies and figure out how to either combine the tactics together or, in my opinion, gravitate more towards or spend more towards the contextual advertising because that's what I feel like it works. That's where I see the data over and over, decade plus, that it works much better than the otherwise. So that's... That's my humble advice to them. Russ Fordyce 🏆 (37:50.45) No, I applaud you for doing all of this. If those shoes will stop following me around the internet and places I don't want to see them. I literally hit it this morning. I was searching for something to buy, not for me, for somebody else. And it was brought into my advertising flow in very disruptive places. And it's just not how we want to live. So I applaud you guys and Seed Tag for all this hard work. And it was clear to the judges that you were doing some amazing things out there. So congratulations again on all the hard work. Kartal Goksel (38:20.028) Thank you, thank you, appreciate it, and thank you for having me on your show and giving me the opportunity to speak my mind.