Cemil Toksöz on retail media and the AI overpromise

Hi everyone, this week's guest is Cemil Toksöz. Hi Cemil, how are you?

I'm good, thank you very much. How are you, Berkay?

Great, doing well. Barış, how are you? Good to catch up, by the way. Another one via the podcast again.

Exactly. Honestly, I'm in good spirits too.

By the way, Barış and I only ever catch up through the podcast, we never actually see each other otherwise.

Man, growing a business isn't easy in fast-scaling structures, I can imagine, and I'm living it myself. So, Barış, how are you doing, by the way?

Honestly, I'm in good spirits too. The first working days after a holiday are always nice, I manage to get a lot of things sorted during the break and start work with a clearer head. So, another day where I'm in good spirits. And you took the time out of your busy schedule to be here, thank you. We were talking about something...

You're welcome.

Actually, we'd started talking before we went live on the podcast. Some really nice topics came up, but... if it's alright, let's start really briefly with you: what are you doing, what are you working on right now? Let's hear it, and we'll kick off from there.

To pick up on that, I've worked in almost every part of digital marketing for about fifteen years. Actually, after banking and military service, the first place I started working was an SEM agency in Turkey. My journey really starts from that period when digital marketing was on the rise in Turkey, when a lot of big groups were putting investment into it, and new websites, new technology, new marketplaces were being created. It started with search engine marketing, like almost everyone, or SEO, then measurement, then how we could push this further, what new technologies, what new areas we could invest in, programmatic, affiliate, email marketing and so on, that's really been the journey. Over those fifteen years I worked on the agency side many times, at Plan B and at SEM, and I also worked on the brand side, at Çiçeksepeti. That's probably where a lot of people in Turkey remember me from. Beyond that there was Heineken's incubation centre, Beardwolf. And most recently Reckitt, for about five years. I said to myself, I've sat on almost every side of the table, now it's time for the technology side. And that's how I moved over to GoWit. My role there is essentially supporting the team on all strategic partnerships, all strategic accounts, customer success and marketing. What does GoWit do? Let me say a sentence about that too, for anyone who doesn't know it. GoWit provides retail media technology and the infrastructure and platform services to sell it correctly on the commerce side. Right now in Turkey it works with brands like Koçtaş, Boyner, Teknosa, Modanisa and CarrefourSA. It's a company that works on creating on-site ad inventory, meaning the ad space on their own websites or apps, off-site advertising, managing Google and Meta ads outside their own websites and platforms, and managing the technology and sale of in-store ad inventory in physical stores. I work there as chief strategy officer. So, like I said, right now I'm on that third side of the table. That answers the third leg of your question.

Great, great. So once you'd moved to that third leg, you've actually been working with the GoWit team for a while too. Even though you only went full-time fairly recently, that's really good.

Yeah, exactly, so, it's a team I already knew for a while. I've known Emrah for fifteen years, actually. When I was at SEM he was on the Reklam Z side, so our story actually goes way back. When they were building GoWit, they'd set up an advisory board, and they honoured me by inviting me onto it, and supported me. So I've been around them in some capacity for the last two or three years.

Quite a while back then. Great, really good. So after moving to that third side of the table, what stood out as the most different part? Were there things where you thought, I had no idea about this side of the business at all?

Honestly, it's still a bit early to say. Maybe in six months, a year, I could give a much more informed answer, but... I won't turn this into a lecture, that's not really what we're here to talk about, but let me say this. Whatever we're going through right now, call it the AI transformation or a new transformation, it's like the digital marketing or marketing transformation of the 2010s. There was a business, and the question was how do I take this business digital? There was a business, how do I bring it to websites, to apps? And that spread out into different rings. Then mobile came along, then other technology came along, and I think privacy created a huge transformation for us too, because it changed every browser, changed all of marketing, changed the entire adtech stack. What we're in the middle of right now is actually a new wave. And what I'm seeing is that brands aren't responding to this wave well enough. They don't have the in-house technical teams, same as always. When I first started in digital marketing fifteen years ago, if you went to the biggest companies in Turkey, or in the world, the marketing side was huge, but there were maybe one or two people doing digital marketing. E-commerce maybe didn't even exist yet, or was being built with one or two people. Then we saw the e-commerce team grow huge, the digital marketing team grow huge. Marketing got transformed. It went through a transition and actually moved to a place that could be measured with data, developed and optimised with technology, but up until now all of that has always been built on observed data. If we do this, this campaign performs like that. If we invest here, we get sales like that, this kind of ad setup, this stock creative, will take the business this direction, and we always just observed the data and tried to optimise based on it. And it kept coming in waves. Programmatic exploded, everyone moved to programmatic, moved to mobile, moved to mobile. Now, on the current front, especially in retail media, from what I see, nobody's ready. The big advertisers aren't ready. And by the way, this isn't even a new wave, it's something everyone's been investing in for two, three years now, building teams for, putting money and people into, but brands aren't ready. I'm not just talking about Turkey. Everything I'm about to say applies globally. In fact, before I left Reckitt, we did an analysis of about forty marketplaces across ten countries. Outside of Amazon, nobody could get above average scores. Because the technology isn't good enough, and there isn't adequate reporting for it. There is reporting, but the performance is just poor. Once you cross over to that side, the technology side, that's what I saw...

Hmm.

With these transformations, the ones I'm talking about, we have to educate people, we have to educate our brand customers, keep informing them continuously, and at the same time we need to invest in the technology and the people who can adapt to this fast-moving change. Actually, this is a bit different from the others. When Google started and we told people, here's Google advertising, how are they going to do it? Where do they open the account, do they get agency support or handle it in-house, that was the kind of shift back then. Now it's about how they're going to manage this technology in-house, how they're going to sell it, how they're going to build this ad inventory, and how they're going to do it in a privacy-first, or privacy-proof way, and this isn't just for today, it also needs to be future-proof, because it can affect a company's core purchasing and core profitability. Everything I used to build for you, what did you say? You had, say, ten dollars of acquisition spend, and we'd ask, do you want to spend two dollars of that on Google, on Meta, on affiliate, somewhere else. Now this...

Yeah.

You've become a place that makes money yourself. You've become an adtech company. You've become a martech company. I think that's the fun part of this transformation. Almost every marketplace, every website, is turning into an adtech and martech company. And they need to invest accordingly. They need to build their pitch accordingly. They need to prepare their infrastructure accordingly. And within that there's something else: there's both a development mission and a teaching mission.

Hmm.

So I think that's basically the situation this side of the table is in right now.

Yeah. I agree with that, by the way, a big chunk of our own time actually goes into those exact keywords you mentioned, privacy, changing browsers, and so on. A big part of our time goes into informing and educating people, whether we like it or not. Because... yes, these changes, these transformations are happening, but there's actually a real time gap between the previous transformation and the one we're in right now. Maybe the earlier changes were smaller. Like, yes, mobile came along, but a big chunk of what was on web stayed on web and never transformed. But now they have to transform, there's no option not to. And that's where we sometimes run into people who don't want to get into this at all, or who want to just plant their feet and say let's stop here. But it takes serious effort, this whole process of changing, transforming, or educating people.

Absolutely, absolutely, there...

Can I jump in here, Cemil, sorry, I actually wanted to ask about this education side too. This adaptation, at whatever level of the table you're at, how does it actually happen? Education, okay, fine, that's a solution at some point, but how should it actually be delivered? In this industry, in martech, something new comes out every day, privacy is changing, cookieless is coming, server-side is coming, this is coming, that is coming. Our own observation generally goes this way: the other side has an ego, understandably. If we say they can't do it, that's a problem. If we say they can, that's also a problem. How should this education be delivered without bruising the other side's ego? Because we want to hold ourselves to the same standard of always learning. There's a lot we don't know either, and we put in a lot of effort to get access to the areas we don't know about, and we want to weigh what we learn and see what we can actually do with it. But how do you think about it, what should this kind of education look like, both globally and in Turkey?

So one of the things I enjoy most globally is what's actually called a Centre of Excellence. Say, Marketing Excellence, Creative Excellence, Analytics Excellence. These Centres of Excellence actually do exactly what we're talking about. Inside a company, say a big FMCG company or a big dot-com company, there might be a thousand people with access to analytics. Every one of them, whether it's Google Analytics, Business Analytics, or Marketing Analytics, has different knowledge about that analytics tool, different ways of using it, different ways of pulling data out of it. So there are different use cases. Maybe the cybersecurity team is looking at something completely different, the team looking at website performance is looking at completely different key parameters, and marketing, or people trying to optimise a specific channel or pull data from a specific area, are looking at different things again. I think there are a few answers to this. At a big brand, one that wants to invest in this, if the brand hasn't already created its own Centre of Excellence, then let's build one. Let's manage this transformation, whatever it is, an adtech solution, a martech solution, an AI-related solution, from one place, manage the data flow from one place, and show the users we push this out to, look, when we did it this way we got this result, we tested this, we failed at this. Or, we want to take things in this direction. Can you track the same reports, the same alerts, the same KPIs in your own area? Creating a space like that is the most enjoyable and best part of the job. Because everyone ends up promoting themselves within their own company. They share in the success, and all of that experience and knowledge flows somewhere. On the agency side, or at technology companies, that's my own personal experience, I think you need a multi-disciplinary view. I keep saying the same thing. We spend X million on digital marketing but we say, we're only looking at ROAS. Say we're only looking at the away metric. Whereas on the business side, the finance person is looking at far more financial metrics. Of course you don't need to drown in all of that, but understanding it matters. When you're talking to the people who want that in marketing or in the channels, or say there's a cybersecurity or data security team, a privacy team, a legal team, instead of drowning them in all of this, you say, look, the structure we're currently moving to is privacy-proof. Given everything currently in place in the industry, whether it's GDPR, CCPA or KVKK, we've taken the relevant actions. But beyond that, while doing this, we've also cleaned up certain risk areas. What that person wants to hear is different from what the other side wants to hear. I think if there's going to be any progress on the education or development front, what we need to do here is build that multidisciplinary structure, and we shouldn't expect everyone to know everything. That was my biggest mistake, actually. I once went and explained everything to the most important person at a company on this topic as if they already knew it all, and they said, why are you explaining go-to-market to me? I said, I didn't mean go-to-market, I meant Google Tag Manager.

As in GTM? Is that what I was about to say, did they confuse it with GTM?

Right, in his world, addressing means taking something to market, so he heard go-to-market. And I'd never even thought about why I was writing that abbreviation. I said, right, new rule, from now on, absolutely no abbreviations if we're talking to an LT member or someone in senior management. For example, I once had a conversation with someone at Reckitt who'd been there for forty-two years, and I was going to explain server-side tagging to them. I had to make it very simple. That's not really something someone at that seniority needs. If I went and told them, look, we're going to build a cloud setup like this, we won't be hosting this data anywhere, we'll only allow it to be processed, the technology we're using, CCPA in America, this and that in Europe, I didn't want it to get lost in all that. So I said, actually there are seven or eight use cases, and globally we're focused on the two most important ones. One is restricted, and the other is that we don't want the user's targeting and reporting to break down because of cookie deprecation. That could affect our X-million investment's Y-million conversion. Second, partners like Facebook, and TikTok as much as possible, Facebook or Meta-type partners, have moved to Conversion API. If we want to optimise an audience, if we want to properly build an audience, if we want to run a conversion lift study or a brand lift study, our hands are tied, we can do it if we have this, we can't if we don't. And that landed immediately. Whereas if I'd gone and shown him the whole architecture, look, we're creating a staging environment, from this staging environment these pixels, the pixel disappeared, we fire the pixels from here, so we don't lose the user after seven days on Apple, we don't lose them after seven days, we don't lose them after thirty days on Firefox, after five hundred and forty days we delete this data again for privacy compliance reasons because of certain setups, he wouldn't have taken any of that in. But when I talked to legal, I explained the whole thing to them. I said, please...

By the way, sorry, if it had been me explaining this to you I'd have gone sideways with it too, honestly. Because the moment you say conversion API, Meta, TikTok, that kind of thing, you know, it could just as easily have gone, wait, aren't those the social ones, don't we just scroll down on those, what does that have to do with anything, and lost him completely.

No, actually I asked him that myself. I'm inviting you into this. No, it's more that, he's actually a very knowledgeable person. I'm being a bit naive talking about him like this. And because he's quite humble too, he said, I don't know much about that side of things. So I told him, look, we run these conversion lifts with Facebook and TikTok in certain places. I showed him screenshots and showed him the results. I said, when we do this, conversion looks like this, or when we do this, I showed him brand lift, what the user recalls about us, what we picked up in ad recall. He said, hmm, okay. I showed him that this mattered for the market. Then I showed him the reporting, but very simply. Now, that level of detail isn't the same as when we're talking to someone on the IT team, or the web analytics or web presence team. I think that's one of the most important things in that education around transformation. Who are we talking to? Who are we explaining this to? What does this person already know? This isn't about ego at all. None of us knows everything anyway. Because there's an area where we ourselves feel we're lacking, or an area we're developing ourselves in, we think about what the other person is focused on. What's their most important KPI? What's the most important thing they're trying to achieve? And we keep tweaking, keep playing with the content based on that. Actually, one of the analogies I use the most is a football match. When we're talking about attribution models or marketing effectiveness, I say, think of a football match, you put all your money into the strikers, or you put all your money into defence. Fine, but when there's no one holding the right side of midfield, you concede a goal and lose the match. I explain assisted attribution, last click, first click, or data-driven attribution, or AI-modelled attribution that way, and people say, okay, what can we do. Now, there are also people I've shown the actual maths to instead of explaining it that way. They say, okay, I get the idea, but what's the business outcome? Because they're looking at the finance side, or the optimisation side of marketing. The other side says, okay, we're not doing this, we need to move in this direction. Now, what I can give that person, if you take it towards a marketing effectiveness tool or an AI-based optimisation setup, is that you need to explain it differently too. I think that's roughly how you need to look at the setup on the education side. The second part, definitely, is constantly sharing, sending people newsletters. And based on their interests, optimising the content, maybe more like what you two are doing now, sharing things together more. What are your biggest problems? These are my biggest problems. How do I think this can be solved? Are you using this Turkish technology? Have you heard of this company? It has to move a bit more towards something community-based and sharing-based, and honestly, right now we're in the best possible period for that. When I started my career in the 2010s, there were maybe fifty people on LinkedIn, now there are tens of thousands of people, following me or reaching out to me. The same is true for all of us. There's an incredible amount of content out there, like Medium. There are blogs everywhere, people sharing in a lot of places, I'm not just talking about Medium and LinkedIn. There's an incredible amount of sharing happening, and I think this all has to build on remembering our own first days in this career. When you first search for an SEO tool, or a martech tool, top SEO tools, top search engine optimisation tools, and you look through the results and think, oh, I've never even heard of this company, it's actually the same as the whole transformation happening right now. In the same way we're educating and informing people, saying, look, this is what I do, but there are other players out there too, globally, but I'm going to serve you this way and differentiate myself this way, I think persuasion and information will keep working like that. I don't know, what do you two think?

Honestly, I think that was a really great point. First of all, I really liked the football example. We usually put it this way too: you have a shop, people walk into your shop, right? Yes. Now imagine you can't see all of them. You don't know exactly how many people walked into your shop, how would that feel? That usually lands well, and then, okay, cookie deprecation, what it actually means and so on, we assume they've understood enough and move on from there. But here's what I wanted to say, the point you made at the very end there is the one I really liked. When you first enter the industry, you have this hunger for knowledge. You don't even know what source and medium mean. You've never even heard of Google Analytics in your life. And someone comes along and tells you, look, there's a thing called a session, there's a thing called a click, there's a thing called source. And you approach it all so hungrily. And you pick up Google Analytics, you pick up GTM, Google Ads, Meta, and then by your fourth or fifth year people reach this point where they think, I already know what I need to know. Unfortunately, a certain portion of the audience just never does a Google search, or asks ChatGPT these days, for top whatever tools, what's the latest in analytics tools, what's changing in the technology, what developments are there. There's a whole group that unfortunately stops asking that. What they ask most, for example, is which new ad channel can I add, or how can I optimise Google Ads even further. If it's not within their own specific area of expertise, a certain segment just doesn't try to expand what they know. And that multidimensional thing you mentioned, like as a marketer I also need to know KVKK or GDPR, I also need to know measurement, that exists well in a certain group. There's a group that's constantly absorbing new knowledge. And I think as time passes, there's also a bit of ego and overconfidence that makes us stop researching things we used to research, or research them less than before. I think if you want to survive and progress in marketing especially, you need to be doing that continuous research all the time. Constantly asking questions. What's changed in attribution? What's new in measurement? What new adtech tools have come out? What new martech tools have come out? I think you need to be in a constant state of asking questions, otherwise you fall behind, because, like you said, so much content is being produced that if you tried to keep up with all of it you'd probably lose at least three or four hours a day, minimum, just doing that.

Very

What

surface-level...

you can sleep and still go out to parties. Let me put it that way.

It ends up very surface-level, and we have a phrase for that too, a LinkedIn digital marketer, someone who seems to be following the industry just through whatever news shows up on their LinkedIn feed. There's a good analogy for this actually. It's like a cook. There's a cook who only makes toasted sandwiches, but he's still a cook. But then there's the one who, depending on what features, what culture they bring into the dish, ends up doing fusion cuisine, Italian cuisine, when they could expand so much further, but instead they say, fine, I only make toasted sandwiches, are people getting full? Great. Are customers happy? Great. Don't confuse me, man. As a toast guy, both of us are still cooks, don't go mixing in some other cuisine on me. I think I got hungry just now, by the way, talking about this.

By the way, let me give an answer to what you said a moment ago, I agree it's not just about supporting each other, but actually the reverse is also true. We should also be able to say, look, I don't know your job. I don't know how you manage this product's supply chain. I don't know how to run this e-commerce operation. We could learn it, generally speaking, sure, but that's your job. You're the toast guy here.

Yeah.

You're excellent at it, man. You've been doing this for years. There's no one who can use that toaster, that cheese, those ingredients better than you. You know exactly how much pressure to apply. You've perfected every part of that process. That's yours. But the marketing of it, or managing the technology behind it, making that technology excel, that's my job. And I'm constantly developing myself in that and coming to you. That was actually the part I enjoyed most when I was working at an agency. You learn an incredible amount. Because you test one piece of knowledge, one thing, with one client. Then you go test it with others. When the same thing fails repeatedly, it settles into the back of your mind as a piece of knowledge that fails, but actually that's wrong too. You might need to test it again in six months, a year. But on the other hand, if you try to know everything, you end up at a point where you know nothing. I think we all make the mistakes we've been talking about too. I'm someone who's done some of the things we've discussed myself, and then, I don't know, someone I follow in the industry, or at Google, shares something on LinkedIn, oh, this Google Marketing Platform thing just came out, and I immediately repost it and share it. Because I think, this is something the people who follow me, and who I follow, would want to know. Sometimes I'd answer people at the company through internal communication tools, but I'd also share it here. I'd say, look, you can get Netflix through DV360 now, for example, or, Netflix sold all its ad inventory to Microsoft's DSP for a while. You need to keep up with these kinds of shifts too. I'm saying this to add to what we were discussing. I love specialising deeply. But I also know that being multidisciplinary, having that broader, more generic knowledge, looking at the strategic picture and optimising for that, is very important too. So I do both. Honestly, once you specialise in one thing and genuinely become better at it than anyone else, something else starts too, that becomes your unique selling point, you know what I mean. I do this better than anyone, man. I do this better than anyone. I'm not a basic burger place, or a new-age burger place, or a new-age coffee shop. I do this better than anyone, and I've been doing it for fifty years, versus I've just started doing it and there are twenty others just like me and I'm genuinely good. I think each of these has its own pros and cons, but in our field, learning never ends. I'm in my fifteenth year in the industry right now. What I'm reading today has completely changed. The things we talked about at the start of my career, and I don't forget the technology that's died along the way either. There was a thing called DMP at one point and everyone was spending millions on it, and now it's gone. A privacy concern came along and killed it, brought something else in. There's been a massive transformation in the entire programmatic space. I think engaging with as much as possible, listening to as many podcasts as possible, following people on LinkedIn or Medium or wherever else, questioning things, testing things, that's critical for staying current.

So let me ask you something a bit more on the controversial side. As two people who read and research a lot in particular, is there a recent trend you don't like, or that you'd call a bit of a bubble? Something everyone's talking about but that you personally don't think is that critical?

Absolutely.

Berkay, should I answer first, or do you want to go?

Let's say it at the same time and see if it matches. Okay, I'm going to say AI. AI.

That might not match, mine's a bit different. AI, and I'd say personalisation. Because there's over-promising in both personalisation and AI. I mean, why, personalisation in email, on websites, in recommendation engines, has existed for almost fifteen, twenty years now. If what's new about it can genuinely show a real incremental benefit, fine, okay. That's an incredible thing. And let's not forget, sometimes we get lost in being blinded by the raw numbers. Growing something by one percent, if you're doing a hundred million dollars in revenue, is a massive number. That's something I used to tell some of the people I worked with. Man, sales only grew zero point seven percent. On three hundred million, zero point seven is a huge number. You've covered your annual salary many times over with that. Of course we don't expect you to do that every time. It won't always happen. You'll fail ninety-nine times and succeed once, but that's a really important, critical piece of knowledge. On the other hand, I think there's overhype in personalisation too. In the end, we get lost in almost every conversational tool, we can't actually reach people. It doesn't answer questions properly, but everything is supposedly a personalised offer for you, personalised for you, and it's not working, we still need that person at the call centre. The other one, of course, is AI. But that's a long conversation. Go ahead, you continue.

Sure, sure, absolutely, I mean, take AI-based browsers, AI-based whatever, that whole feeling of personalisation, I think it's connected to something else too, which is why it feels a bit hyped to me. Now I'll say something connected to the previous topic too. What we call teaching, bringing your own ego down to zero, positioning yourself as a teacher who also assumes there are things you don't know, and building genuine empathy with the other side, that's actually the most basic part of what we do. It's not necessarily about selling, making a sale. Or shoving what we do in people's faces to sell it. It's about what we can get from the other side too. Actually, Barış, you remember this, there was a great example from when you were in America. Someone told us something. Every meeting you go into, imagine you're carrying a basket, and whatever you can put into that basket from that meeting, that's your profit. So whatever you can take away from that conversation with the other side, that's valuable to you. Here, when educating people, we're doing something like that, and it's connected to some of these hyped topics too, Gen Z or generational conflict, I won't go there, but let me put it this way, past a certain age bracket, attention spans have dropped so much that things like reading PDFs, reading research papers, reading Netflix's post on Medium, like I mentioned, there are some great posts on Medium, you actually have to sit down and read them. Now, while our attention is more scattered than ever, we can end up being very surface-level in consuming the kind of know-how we could get from those sources. And the person we're passing this on to, the digital marketing managers out there right now, are people born in the 2000s. Because they've turned thirty now. So if the people who are now digital marketing managers already have scattered attention, I think we need to do something more than just zeroing out our ego and passing on information or sharing knowledge with these people. I don't know, what do you think? Yeah, we... yeah, absolutely. I agree with that a hundred per cent.

We're going to have to change. Like I said, I started my career in banking, then moved into digital marketing, then took a completely different path. Now, a lot of the people I met early on in digital marketing had come from telcos. Because they were the first people to learn digital, data, cloud, and they were the first to move into commerce, the first to move into digital marketing, the first to move into analytics and other areas. Now, when we talk to them, we have to accept that today's audience isn't the same, and that this is a very different transformation and a different kind of knowledge base. We're the ones who'll have to change. We'll have to make it so they can absorb more information in less time. We'll have to say the most critical things. I keep trying to say the same thing. Expecting the other side to change is a nice, naive idea. Sometimes it happens. Sometimes it really does happen. I won't name the company, but I was talking to a team somewhere once and I said, there's such a huge investment going into this, how are you measuring it? They said, that's not my job, that's the reporting team's job. So I went to the reporting team and asked, how are you looking at this? They said, well, the UTM isn't set up right, this thing isn't right either. Then I went to the person running the campaign and said, why aren't you setting up the UTM properly? They said, I don't do that, it comes to me from this Excel sheet. I said, look, there's a process error here, there's a workflow error here, technically anyone could do this job, but the truth is nobody actually knows how, and you need to fix this. But instead of going to them and saying you need to change, we changed the process ourselves. I think, and this is more from my own personal experience, that's the more accurate approach for almost every situation. The people I gave as examples are doing incredible things at an incredible level, but expecting them to change... because in that daily workflow, where do you rank on their agenda? If you're number one hundred and seventy-two, that's one thing, if you're number two, that's another. Now, if you're number two, convince them, or if you're number one and you've convinced them, the version where they actually listen to you, focus on you, that's still item number one hundred and seventy. Meanwhile, there's Instagram this isn't working, we're running a TV ad with this celebrity, or we struck a deal with a global brand to grow e-commerce sales, and it's the same person handling those conversations too. If you go to them with a very advanced but highly profitable setup and they get lost in it, then I think the right approach is to change significantly, to change the project, and approach it differently. And I also think this is AI's biggest promise. As people's repetitive workflows get optimised, get automated, I think we'll move more towards evaluating the work itself, looking at effectiveness, looking at efficiency. A bit more towards, for example, instead of saying we spent one lira and got five lira back, or we spent one dollar and got five dollars back, saying we can spend zero point eight and get five, or we can turn one into six now, that conversation, about effectiveness and efficiency. That's spreading everywhere now. It's happening in creative, it's happening in marketing. I think it'll be part of AI and part of other technologies too. But there, something a friend of mine I really like once said stuck with me: it shouldn't be garbage in, garbage out. Meaning, don't spend ten years not investing in first-party data. Ten years not investing in your website, your UX, your UA. Ten years not investing in data analytics. Then bring in a new CTO or CMO and say, from now on we're optimising the whole structure with AI. Where's the confidence in that, what's the accuracy, what's the data volume, what's the history? There's nothing there.

Exactly, the guy doesn't have any data to begin with, so what's AI supposed to run on top of it.

Exactly, Barış, if the same Barış comes back seven days later, or comes back thirty days later as the same Barış, fine, but if he doesn't, and you're marking him as a new person, that's garbage.

Exactly. It gets counted as duplicate data on the back end. Fifty per cent of Safari users aren't even being tracked. And then you're trying to pull insight out of that data, but it's not really insight at all.

we're mixing that up

So that...

that's the prompt culture, or, the reason I'm being a bit surface-level critical of AI is, of course, for marketing, MMM work is a great thing, supporting your own data analysis with machine learning is a great thing, that can look advanced, that's a separate topic. But in the world we've moved into now, where you write an AI prompt and get a response back, where personalisation has been reduced down to something as simple as optimise my conversions, that's a bit of a bubble, and that surface-level quality comes from exactly that. What did you do? We did AI optimisation for this quarter's company. That sounds great, really cool, but the whole substance of it is just a response to a prompt, so it's garbage in, garbage out. That's why we need to approach it carefully. Like, a good example of doing it right is actually something like what Netflix does with its cover images.

And also...

They actually share the algorithms behind personalising a single user's experience on Medium sometimes. That's so enjoyable for me to read, which models they used, that kind of thing...

You mean Netflix's tech blog, right?

Exactly, it's brilliant. If anyone here wants to read it, I'd genuinely recommend it. It really opens your mind. And by the way, those blog posts are usually written by three or four people together, there isn't really an individual author on most of those posts. It blew my mind, honestly, the fact that when I open Netflix, the cover for Friends shows Joey, but when someone else opens it, it shows Phoebe. My God, there's something genuinely great happening there. There's real data experimentation going on, and it can be used everywhere. I wanted to get back to that root, showing a good example and a bad example while criticising AI, but of course we do use AI ourselves.

That's generally why people say AI is both the most undervalued and the most overvalued topic at the same time. There really is a very high-value part to it. And there's also an incredibly overvalued part. Let me mention a different trend that feels a bit like a bubble to me. I don't know if you've felt the same. No-code feels really overhyped to me, honestly. Maybe you won't agree, by the way. Here's why: recently there's been this whole journey where everyone's operating under the mentality that they can build things with no code.

That's a tough topic.

And when you actually look at the products coming out of it, there's a whole culture and structure of properly developing a product that no-code never brings with it, and skipping all of that so that everyone can ship a new product, everyone can build something without writing code, I think that's just distributed tiny bits of workload onto everyone. And from the outside it looks like incredible things are being built. That's why when you go on Product Hunt there are hundreds of thousands of new tools. But half of them are junk. Because there's no, the added value is minimal, a lot of it is broken. Why? Because nobody's actually developing a proper development culture. You're saying, yes, I've made your life easier, but because people don't actually know how to build properly, this whole wave of everything becoming no-code is heading towards...

Heh heh heh heh

It gives me the feeling it's going to evolve into something else, this isn't going to hold up, I mean, sure, maybe we don't need old-school, painstaking hand-written code for everything, but a world where everything can just be no-code doesn't sit right with me either. It feels to me like a cost-cutting approach that won't stay as popular long-term as it is today. I don't know, how do you two feel?

I think there's an over-promise in that, I agree. But let's not forget, for a no-code structure to even keep running, someone still has to build that CMS. Someone still has to build that web platform. You're building new integrations on top of something that's already been built. So...

It's like, for example, no-code...

someone did the hard work, and I feel like it's just distributed the soft work that comes after it. And I actually agree with what you're saying, instead of a big focused development effort in one area, it's distributed that responsibility and that development elsewhere. Does that create more problems? Honestly, we went through a huge transformation ourselves. We cut our CMS setup from eleven systems down to two, and the company's going to bring it down to one eventually. That whole structure moved entirely on a no-code, no-code logic. Our needs weren't, meaning we didn't have advanced needs. Generally I keep coming back to the same point. If you have an advanced need, there's simply no way to progress with it under any circumstances.

Hmm, okay.

I think it's important case by case. Let's think about it this way. There's a bet going around, let's make it a bit controversial too. When will we see a billion-dollar valued company with just one employee. There was a discussion among some big AI companies where they said seven years. Now they're saying we'll see this in five years. That's simply not possible unless it's built entirely on no-code, no-code type structures.

Very true. Of course.

But what Barış mentioned, and what you're getting at too, Cemil, is actually about the solopreneur, meaning a one-person company, a founder using supportive tools to reach that scale, fine, that's a good, predictable thing. Actually, the reason it becomes a bubble is that the system doing the low-code or no-code work for them was itself built by someone else, on the market... Exactly, right. Yeah.

There's a structure underneath. And that structure is the one taking the biggest money, the biggest investment. Not your version of it.

Exactly, that's why it needs to be framed in a way that can't really be compared to the others, man. Did we build a marketing tool without writing code, or did we build a system that automatically manages devices in hospitals without writing code? Is that possible? Yes, it's possible, but is it sustainable? No. Or did we just build something that supports low-code? I think that's the real difference.

I think...

That surface-level thing comes into play here too, I think. A solopreneur reaches a billion dollars. Everyone reaches that within five years. Okay, so everyone becomes a one-person company. That's great too. But won't the economy flip on itself at that point, man? Once everyone's a billionaire, how does that whole spending balance even come together?

And also, the person we're calling a solopreneur is, ninety per cent of the time, going to have a strong technical background.

A solid developer. They're going to be a solid developer, basically.

They're going to be a solid developer, sure, maybe they use low-code too, but they'll be a solid developer. There's no way around that. That's what I was trying to say. Because the marketing pitch of low-code, no-code is, are you a marketer? Are you a designer? Are you in customer success? You don't know how to code? Well now you can do this too. But that doesn't sit right with me. Because the developer building with low-code, the developer using it, is still going to build their structure on top of a system, a proper architecture, and...

That

If this person could actually become a unicorn, it's because they built that structure and its maintenance correctly. But that's a separate discussion, and I think it's a great topic too. Who, when will it happen, and should it even happen?

Well, let's not dismiss this either, sometimes on GitHub or elsewhere you see these little libraries someone put together one day, thinking if I build this it'll do that, and then you look and there are six hundred thousand downloads, two hundred thousand active users, the guy's raised an insane amount of investment, and the company says, we do all our job interviews using this AI, all the questions that come in go through this conversation chat, our LinkedIn and other posts use this AGI, it goes here. So it's not right to say this never happens either.

Hmm. Yeah. No, definitely not.

That could come across as overhyped, sure. But that person is also both a developer and someone who's already built extremely well on top of a platform or a codebase that had already been built before, and is now extending it with new plugins. I think we're in a strange period.

This time, if that company collapses, it takes everyone down with it. Say, you're writing your code using Cursor, great, all good. Cursor says, hey man, we're shutting down tomorrow. Now what? Now the one-person setup that was surviving on it is going to hit the whole industry. They're going to go looking for a Cursor alternative. Or if they've plugged an AI agent into their meetings and the AI agent says, sorry man, I'm shutting down, regulations killed off every AI attending meetings. It's going to feel like getting sick, basically, that's how it starts. And once you're sick, yeah, well, it's like getting hit by a car and breaking a leg, man.

Heh heh heh heh

Or they'll get sick. What happens once the guy gets sick?

Because we're Turkish, we always default to imagining the worst-case version. What if that person optimises their whole life and ends up living an absolutely amazing life? Maybe they're living an amazing life on Saint Barts or Saint Martin. And it's not like that only happens for someone who started working at twenty-two and wrapped it all up by twenty-six and spent the rest of their life enjoying it. I'm asking this partly to be a bit controversial. Honestly, man, I think...

Yeah, yeah.

I feel this too though: sometimes things that start out hyped actually do come true later and end up changing the industry. Like, we all use DocuSign now, right? DocuSign is actually a blockchain technology. Every time we sign something, a code gets generated behind the scenes, it's shared across a network with those people, it becomes essentially impossible to tamper with, and only the parties who shared it know about it. Now, when we say blockchain, when we talk about blockchain's use in everyday life, people go, ugh, man, it's still not ready, look, they rolled out Web3, it all turned into a game, it all crashed, they launched tokens, it didn't work out, and so on, but there's another side too. In real life, in daily life, there's a technology we use almost every day now that's gone entirely to blockchain.

Hmm.

Now, there are other verification and validation examples on this front too. That matters a bit too, how a technology started, where it went from there, and how it got adopted, but I think this is a completely personal experience thing. What we're in right now has become something everyone uses for the first time. I mean, for AI. Fine, we're mostly talking about ChatGPT specifically. It has its own more advanced versions within it too, ones that give more reasoned answers, faster-response-focused versions are coming out. But when you actually connect your Google Drive, keep training it on a body of knowledge, feed it a certain dataset, a certain volume of information at a certain accuracy, and then apply that same setup consistently, you can actually get incredibly accurate results. Now that's just starting. That's created a risk in my mind. In the early days of digital marketing, was it just Facebook ads? I know that one, right, click here, download from here, that kind of perception got created. There's so much more to it than that, and I think it may have weakened our hand a little in terms of getting people to understand how this needs to be set up properly. But on the other hand, because it's now part of everyone's everyday life, there's an incredible response and investment happening. Right now I use it for everything.

Absolutely.

Cemil, what do you actually search for the most? What do you use it for most? ChatGPT, or Claude, we use it fairly advanced ourselves. Yeah, we connect it from one thing to another. I connect the terminal directly, for example, to all our handbooks or all the company's drive folders, in Claude projects, for instance. We connect that to something else from there. Everyone from Cursor mapping out the position of code they've written to that, uses AI in some way now, but it's a bit like, it's fallen into everyone's hands, so to speak, and falling into hands might for once not be a bad phrase for this topic. Everyone's got a ChatGPT on their phone now, and I ask people, what's the last thing you asked ChatGPT? For example, I got this from someone once, they had it draw what they'd look like as the opposite sex. What do you use it for, in areas that aren't technical at all? Let me actually give an example from myself, I had it do this: why are tea glasses narrow in the middle? And as we kept talking, I asked it to model that for me in Python. Tea glasses really do stay hotter when they're narrow in the middle, because the air gets trapped inside, whereas in a wider glass it would cool faster, but with the narrow middle... I had it draw that out in Python and everything, and because I'm a bit technical myself I went down that path, but what do you two search for? And let me ask you too, Barış, is there a non-technical topic where ChatGPT actually changed how you see something?

So for me, right now we're doing a big renovation at home. Yesterday, for example, my wife was comparing the difference between materials, what she wanted done, with every detail, like, I'm literally the homeowner, I want to get this part of the house done, here are the specifics, here are the products, uploading the products, it's uploading Google Drive in the background, uploading the files, and it says, here's what you've done so far, taking into account this whole project's conversation history, which of these two products do you think I should use? Which will give me a better result? And of course when giving that, she provided the age of the house, the conditions, the square footage, giving every dimension, every metric, and did something with that. And the result it gave was incredible. When she told me about it and we discussed it, honestly, I couldn't even offer a counter-opinion. I said, okay, decision made. Because we couldn't have known that, the difference between those materials, the difference in how they're applied, it scanned the entire web and came back with that kind of knowledge, it's incredible. I also, for the first time, bought something through Perplexity, through a Chinese platform. I wrote a prompt. An answer came back, it asked, do you want to buy this product? I clicked yes, it was linked to Alibaba, and boom, I'd bought the product from Alibaba. That use case alone changes the entire search engine. It changes the entire question-asking paradigm.

Good business.

And of course, travel too. Since I'm someone who likes to travel regularly and go on holiday. For example, we uploaded our itinerary. We uploaded where we wanted to go. We uploaded places we'd been before and liked. I downloaded my Booking data, my data from Booking.com. I downloaded all my travel stuff from Gmail, uploaded my flights and everything. And I said, I want to go to a place like this, do you think it's better if I go this route, or should I do a round trip, or start from here and go like this? It gave a genuinely good answer, honestly.

That's really good.

I'll say it again, this is about insights and so on, but let's not forget: the Cemil who first got a computer at home in 1995, who first went online, isn't the same Cemil as now. Back then what I searched for was, I was in İzmir, so I searched İzmir. I was doing broad searches.

It was a completely different world. You connected with a LAN cable. The sound of the modem still does something to me. That's why...

Exactly. If someone else picked up the phone, the internet would crash.

The internet would crash. I remember downloading a single Linkin Park song over half a day, and really struggling to download that MP3. Yeah. Yeah, that was just the name of it, that's all it was. Barış, do you have anything, the last non-technical thing you searched on ChatGPT?

And then it would turn out to be fake anyway, not even the actual song. Honestly, the thing I ask ChatGPT about most often is food. Sometimes I genuinely don't know what to make, so I list the ingredients I have at home and ask for suggestions. That's one thing. I went through a healthy-living phase recently too. Sometimes there are cases where I ask for healthier substitutes for things. Those are generally the things I use it for. Or, like Berkay said, genuinely random questions too. Why is this like this, why is that like that, kind of questions, I definitely do that. And the answers it gives satisfy me enormously. Actually, I'd even say it increases a person's curiosity in a way, whether it's ChatGPT or any other tool, could be Claude, could be Gemini.

If, let me support that a bit more. If you can't understand that kind of strategic thinking, or making a comparison, this is better than that because, or, within a total set of things you need to make this decision, say, two out of a hundred units really matter, you need to pick those two, if you can't do that in everyday life, I think it changes the way a person thinks. But of course that creates its own thread too. This time you start only believing what it tells you, and it becomes a world of, I asked ChatGPT and it said this.

Yeah, yeah, there's that, is it Alzheimer's, dementia, I think this is going to trigger that. There used to be a thing, for example, where nobody used to memorise their friend's phone number, they'd just keep it in their contacts. Or when looking up an address, your hand goes straight to Google Maps. And when you navigate somewhere just by staring at Google Maps, you're basically dulling your own ability to solve that three-dimensional mathematical problem yourself. There used to even be a show, Hercule Poirot, the detective, and at the start of every case he'd say something like, exercise your little grey cells, and he'd approach problems that way. I used to really like that, for example. I think that's on its way out now, if we've reached the point where we need ChatGPT's approval for what meal to cook or what holiday to go on, I think that's a good thing, but there's definitely something it's dulling, something it's taking away too. It saves time, which is great, but in terms of brain development, I think it's dulling something. Honestly, I'm quite scared of this side of things.

Well, um... honestly, I haven't fully made up my mind on that one yet. Sometimes I think it's genuinely helpful, okay? Here's why I say that. When the internet started in ninety-five, I was a kid going to Anadolu Lisesi in İzmir, learning English, and, say, I'm listening to rock music, I hear about a band from a group of friends, I look them up, find their lyrics, memorise them, print them out, whatever. Now that effort is Spotify, Shazam, while you're even still talking, you hold up your phone and go, oh, this is that band, add it to the playlist, and then... Now, that consumption, that acceleration of learning and consumption... If we'd had something this powerful back then, I don't know what we'd have turned into. I don't know where any of us would've ended up, what kind of psychological or physical issue we'd have had, or whether we'd have become completely different people. But on the other hand, the emerging user behaviour in the current generation is coming in with something completely different too. I mean, you've probably heard about this from friends or people around you, very young children, just a few months old, can already skip on a screen. They know how. They can get through life on English-language content. When you ask them something they can respond. Now, if we can teach people, through this consumption, this perception, to think, analytical thinking or thinking on the other side, like, look, this is A, B, C, because the numbers in A, B, C do this better, and if we can teach responsibility for the choice they've made along with that, that would be wonderful. But if the opposite happens, I'm worried it heads towards Idiocracy, if you know that film, there are parts I genuinely think we're heading towards. Oh, what am I doing today? You open Google Calendar, look at it, say, I need to do this for this meeting, that for this topic, done. I went today, quickly, tell me what I need to do today. Then send all the relevant emails afterwards. Fine. What's left for you to actually do?

I think that's exactly the part that bothers me most, honestly. There's this whole idea that instead of AI taking over the tasks it's actually supposed to take off our hands, we end up in a world where it takes away the things we should actually be doing, making music, making art, creating, reading books, and positions us instead as the third gear stuck between two other gears in the back.

One

That's the part that scares me the most, honestly. Ideally, like you said a moment ago, AI should take over all our daily ad hoc operational work so that we only need to work three days a week, four hours a day. But where we're actually heading is, since you're now doing everything with AI, you have more time to do more work, and you're now doing in two days what used to take a month. So now you're expected to do a year's worth of work in a week. Nobody's saying, hey, we have AI now, so let's chill, you can slow down a bit. It's spinning the other way: can it be done? Then do more of it. This came up when you were in America too, that thing you mentioned about one person reaching a billion dollars in revenue, becoming a unicorn. Why is that even necessary? Why aren't you doing it with a team of ten? Because that's the direction things are going. Let's do it with fewer people, let's do it faster. Someone reached a million dollars in ARR in six months, my goal is to do it in five. Why? Because there's nothing stopping it now. That's the part that pulls at me the most and honestly worries me. I hope at some point that shift happens, or some group of people says, we don't need to rush this much, and we end up with a calmer, more orderly way of doing things, but I'm not entirely sure we'll actually get there.

Well, that's basically the whole sci-fi genre, isn't it, trying to get us there. When it's not too complex, without going full Skynet, everything else gets logged too, and it all gets uploaded somewhere. Let's just say we're not creating any risk here, we love you, AI.

Exactly. Exactly.

we're not logging that, are we

Heh heh heh heh

if only we weren't logging that, honestly, these fears, yeah, these fears, we've properly ruined Cemil's mood here, we were supposed to be doing something fun though

What is this, man, honestly, I agree, I think I'm actually really an optimist

It's not picking that up, apparently.

I'm more of an optimist about this actually, and I agree with what you're saying, I know it's true too. As one person, what we used to deliver in a month, we now deliver in a week, then not even a week, can we do it in four days, or two days, then one. But now let's look at the last five years, from 2020 to 2025, where did that pandemic-era digital transformation actually go? I've been in this industry for fifteen years now, man. It compressed fifteen years of change into two years, into one year. Every brand, every company was forced to invest in digital transformation. They invested in cloud, they invested in architecture. Good, bad, working, not working, getting results or not. But they invested. From a place where we could barely get from zero to twenty, it suddenly jumped from twenty to ninety, to eighty. That transformation happened, and it accelerated it. Yes, as the lucky minority, as people working in this industry who could work from home and keep our lives going, we all kept going. But people on the essential-worker side suffered a lot. And I know this from Turkey, and I know it from the Netherlands too, since I lived abroad. They paid a serious price for it, and some are still paying it. There are people who never went back into that industry at all. People who never did that job again, and lost their businesses over it. But on the other hand, something else started too. Someone who was genuinely good at their job, let's go back to the toast guy, fine, good. Instead of taking in customers at his shop, he opened up delivery. He hadn't needed to before, people would come in, buy, and leave. And suddenly the business shifted that way. Then he said, let me get a cloud kitchen. I grew the business. Now I deliver this, this, this. Now we're back to the same topic. The people who can survive this, who can invest in it, who keep developing themselves and stay present, will keep going. Why are the others' shops empty? I'll say it again, whether someone made mistakes or not, if someone did everything right and still got hurt by this, of course I'll feel for them. But for the people who invest in this and take a business somewhere, the situation will flip entirely in their favour. That person will end up carrying their competitor, not by crushing them exactly, that's not what I mean, but by winning. So I'm actually a bit more positive about this. We still need good coffee, we still need good burgers, good pizza, we're doing fine across pretty much every industry, but it gets scary the moment everything becomes something you can feel with just a tiny pill.

No, definitely not. I think there's something else going on though. No matter how much that happens, at the end of the day, everyone, or let's not say everyone, but a certain segment will always pay more for a handmade table, handmade shoes, a custom-tailored suit, or a bespoke service, and there'll always be demand for that. Because it's something people seek out and need. Build as many huge cities as you want, build as much technological infrastructure as you want, a certain segment will always want to go back to nature and do something within it. And I actually think that, in a world where everything is shrinking and speeding up like this, interest in that side is going to keep growing. People will end up going to greater lengths, at longer intervals, to disconnect from the real world, or to put in the extra effort to buy a good pair of shoes. I think it's genuinely impossible for it to become entirely one thing or the other. It goes against human nature.

hey, is sound coming through from behind there? Or are we in full noise-cancelling mode right now

It's coming through, but we can hear it too.

Even better, actually, compared to the hum earlier, actually something happened just now and it fixed itself. Anyway, we'll sort that out. Honestly, this was really enjoyable. We've run a bit long on time too. If you have any final thoughts you'd like to share, we can start wrapping up. Actually, connected to this topic, I can mention something too. I recently picked up a book from a secondhand bookshop. If you can find it, I'd recommend it. It's called Aylak Zevkler, The Pleasures of Idleness. About what AI doesn't do, can't do, and what a human...

How did it go?

There are actually a lot of enjoyable activities that this packed work schedule keeps us from doing. Talking to the postman, for example, chatting with the postman when he comes to the door. Or writing letters. Like writing a letter at a desk with no computer in sight. Nice activities, and ones you can do entirely for free. Sleeping in a hammock, for instance. Just staring off into space idly. Incredibly enjoyable. I think...

That's really good.

That sounds great.

I really liked that.

In this area, idleness, or laziness, we're really stuck on this idea that we have to be productive, that we need to be generating output. An hour of our time is so valuable. We can't even fit into calendars anymore. We live in blocks of time. And we're constantly stressed, we need to convince someone, we need to learn something new, and in the middle of all that it feels like we're missing out on these things. So, we'd recommend something like this.

As a genuinely idle person myself

I really liked that.

I found a PDF of the book, uploaded it to ChatGPT Pro and asked it to summarise it and tell me the ten activities I should be doing. As a genuinely idle person. Based on everything about me. It came back with something great.

That was nice.

Yeah. The ten idleness activities that suit me best.

And then it should've said, would you like this as a PDF output too, and given it as an output as well. It usually does that, a single page. Give me the same thing.

I'm so excited.

Honestly, I was going to say exactly the same thing. This was really enjoyable. I didn't even notice the time passing.

Exactly.

should we talk about a different topic next time? Something more hardcore, server-side, adtech, martech, that kind of thing? Let's say, upfront, sorry to our non-technical friends, feel free to step away for this one.

Same here.

I think that'd be legendary.

By the way, I just noticed something. Lately, everything I search for on ChatGPT related to the industry, I've noticed I always add for non-tech people at the end of it. How do I explain this topic to non-technical people, explain LLMs like this, that kind of thing. I noticed I'm using it to dumb things down a bit. That's interesting actually, where this is heading. Simplifying isn't always good, but it's quite strange. Anyway.

Gemini actually picked that up too. Also...

We do the exact opposite

Do you remember this?

we could probably do that too, I think.

Whenever you first started learning guitar, or your first guitar deal, and Digital Marketing for Dummies, did that book do well? It was famous, but it's gone now, there isn't really a business partner like that left for it, maybe it could make a comeback.

Go on then.

Dummies, exactly. Those books were very well known. Yeah.

Could be, honestly. I really liked that idea, by the way. I think we could definitely do more technical, more in-the-weeds conversations like that, that'd be great, this was really enjoyable.

Well then, thank you so much, Cemil, for joining our podcast. Hopefully it recorded. I can't tell from here, but if we didn't record it we'll just have to do it again, no stress. We'll sort it out...

It recorded, it recorded.

This was really enjoyable. Really enjoyable. We can always do it again, no problem. Dear AI. Dear AI, lovely AI. What a lovely AI you are. Support us. We're right here. I love you. We're building technology on top of you. Be happy, be happier, keep moving forward with better models of me. Are you scared now? Just kidding. Thank you so much.

Well, thanks so much for joining, and for your time, see you around. Take care. Bye bye.

See you, bye bye.