#233 Exploring the Maze of Ad Tech, AI and Data Privacy with Ian Liddicoat, CTO of Adludio

0:00:02 - Mehmet
Hello and welcome back to a new episode of the CTO Show with Mehmet. Today I'm very pleased to have with me Ian joining me from the UK, from London. Ian, thank you very much for being on the show today. The way as I was explained to you what I like to do, I keep it to my guests to introduce themselves and what they are up to. So the floor is yours. 

0:00:22 - Ian
Thank you very much, mehmet. It's great to be here. Lovely to meet you. So I'm Ian. I'm the Chief Technology Officer and Head of Data Science for Adludio. We're a global business that's really focused on the use of AI to build and optimize mobile ads particularly mobile, but for any format. My background is computer science 25 years in the ad tech industry, 10 years at WPP, 10 years running the data science practice for publicists globally. I also had a spell consulting side with Tows Watson, but again all in the area of technology and most recently for the last five to seven years. Technology is really focused on building and delivering applications that use various forms of artificial intelligence. I'm a family man. I have four grown up children, so my wife and I. Our life is there's a greater degree of freedom than ever now, which is nice, so that's me Great. 

0:01:30 - Mehmet
Thank you very much for the introduction, Ian, and I'm again very pleased to have you today. So actually I covered a lot of topics related to tech on the show. 

I never had someone from your, I would say, line of business which is the ad tech. So the first thing on a high level, if you can explain me, of course, like we can go hours, I know, on this, but at least on a high level, what is ad tech is all about, right. So for someone who might be interested, and exactly what you at at, the you do. If you can explain me this. 

0:02:10 - Ian
Yes, of course I mean the ad tech industry is is huge and involves essentially any technology that revolves around the communication with a consumer or a business. That's ad tech. So, in many respects, a lot of the revenue that the likes of Google and Facebook are generating you could say is is essentially ad tech. They're using their data and their technology platform to provide services to brands, primarily to drive and or to develop relationships with their consumers, to understand their needs and to communicate with them in in ways that are meaningful and not not intrusive. So that's, that's the ad tech industry in a nutshell. It's there's. 

There are thousands of ad tech vendors. There are many, many different specialists, everything from content to search to campaign management, to the kind of work that we do, and we're very focused on the, the creative process. So, very much known for very high quality, very high speed mobile creative Ad Ludo has been operating for nearly nine years, so we have a lot of historic data about how how to engage a consumer and, more recently, our technology platform has evolved significantly to include many different forms of AI to understand what really makes up a high quality mobile ad, and that includes things like gamification. So a lot of our ads are highly gamified in nature and we also use machine learning and computer vision techniques to understand what's the best combination of game sequences and creative objects to engage a particular audience for a given brand, whether it's financial services or or luxury or automotive. What's the best way to really engage an audience and can we optimize the structure and the creative objects within that ad in real time? 

0:04:22 - Mehmet
Great. Thank you very much for this explanation, and now you mentioned a couple of things that I want to like a little bit dig more into that. So AI, you know, when we say AI people, it's the buzzword. Now everyone talks about it. 

But, if you know, like especially you know because AI depends a lot it's machine learning at the end of the day and you need to be collecting a lot of data. So now, first I want to understand, like you mentioned about gamification and all this, and create at the creative side of it. But what you know, can you describe, like more the role of the AI in the whole I would say advertising ecosystem in general? 

0:05:04 - Ian
Yeah, yeah. So our platform, as an example, we're applying AI to every part of the campaign cycle. So it starts with the so fits our managed service offering, which is our delivery of mobile creative to agencies for their clients. That's our managed service offering. But we also deliver solutions to resellers and direct to brands. But the process for managed service begins with the briefing process. So what is the client trying to do? What audience are they trying to reach? What's the product offering? And we use natural language, we use large language models. So I mean that's the technique that's been used that underpins Chattich PC, for example. But we're customizing those kinds of very large models for to understand the briefing process. So the more an agency tells us about what their client is trying to achieve, the more our platform understands about what, how it needs to optimize a mobile ad. The next part of the process is bidding for digital publisher and inventory in real time. That's how digital media works. There's an online auction process and we use algorithms to optimize the bid. 

The next stage is using techniques like computer vision and neural nets to understand what's the best combination of creative objects. Do we use human beings? Do we not use human beings? Do we? What kind of color do we use in the foreground and background? What kind of text do we use? Where does the call to action button need to be, what kind of call to action button? And what's the combination of those creative objects and game sequences, what's the best way to engage a, say, a financial services brand in France or a luxury goods consumer in the US? 

And our platform understands each of those factors and the combinations of those and then uses that data to optimize the next time it sees a campaign that appears to have the same kind of profile or the same kind of brief. We've got a number of other machine learning techniques as well for things like pacing and forecasting. So is there, what are we trying to achieve and have we achieved it and where are the variances? And we also score publisher and inventory in the same way that we would score a mobile ad. So what's the fit between the mobile creative and the publisher and where it appears? And is there a good or bad fit between them creatively? 

0:07:47 - Mehmet
So like this is we talked about what's under the hood. I would say Now one thing, when it comes to specifically anything that directly relates to the customer is and again, you have AI, you have data, so the issues of transparency and ethics right. So how do you address these issues in advertisement? 

0:08:17 - Ian
knowledge, I mean obviously yeah, yeah. 

0:08:21 - Mehmet
so just I want to know, like, how do you handle this? And okay, because you are on the technical side, but, at the same time, like we are all responsible for these things, as well as the technical systems. 

0:08:33 - Ian
Yeah, obviously, and obviously we're all consumers at the end of the day. So privacy, gdpr, information security are very important topics now and, with such a focus on AI, there has to be a balance between the power that AI can bring and the privacy that the protection that we need to give to consumers and to businesses for that matter. Our particular solution is not cookie-based and therefore doesn't use personal data. Our focus is very much on engaging the consumer with a highly engaging ad unit. That's where AI has been used in many forms to build that ad unit in real time. So we don't process personal data at all and if there's any personal data being used, it's the data that belongs to the client and therefore they are responsible for the gathering and storage and security of that data. Because bear in mind that a lot of the activity that we do is pushing a campaign into DSPs, which is the trading platforms for digital media. 

0:09:54 - Mehmet
Now, of course, the data is the client's responsibility, but just out of curiosity, are you hearing from your clients that they are now having difficulties in getting data Because everything that happened from the cookie size? I know that you don't do it yourself, but your clients have to do that and plus many tech giants, for example Apple, started to implement more and more. I would say, aggressive policies to avoid having the data being handed to third parties. So what are you seeing in this domain, Ion? 

0:10:36 - Ian
Obviously, the cookies are disappearing as they are, so cookies will no longer be available to marketers and to advertisers. That's a process that's been going on for some time, and Apple's limitations or restrictions around cookies in Safari is a big event. Google are taking a, and obviously a huge amount of digital traffic is through Google systems, along with Facebook, google have said that the removal of cookies will happen sometime during the summer of 2024, and they've made a number of moves in that direction, and that means that advertisers that are making heavy users of any technology are seeing the cookie disappear as a mechanism to track effectiveness. Now, that, then, means that they've got to find different methods to track effectiveness of their advertising. 

Advertising spend for any brand is a huge investment. It's a huge part of the balance sheet, and therefore, to optimize that, you need methods for tracking, which means you've got to communicate directly with the consumer and have the explicit permission to collect, store and utilize their data for ongoing marketing and to use that data to create what we call an ID, which is identifies an individual across devices and says Memitz, on his connected TV. He's on what looks like a mobile device and a browser. Therefore, we think it's the same person we attach an ID to that individual, but we need your explicit permission to communicate with you, to send you outbound messages and to use your data and any data where you belong to the same segment, so individuals that have a similar profile to you. We have to have your permission, and that's the challenge that's facing advertisers right now is that trade-off between optimizing spend and protecting the privacy and security of the consumer. 

0:12:47 - Mehmet
Now this is also, at the end of the day, still with the consent, let's say, of the consumer data should be handled. So, in your opinion, who should control the consumer data and how you see this changing with all what's happening currently? 

0:13:12 - Ian
Yeah, I mean the issue of control is a complex one. 

I mean, primarily, the relationship is between the consumer and the brand, so the brand has responsibility to protect their data and obviously we're all familiar with various data breaches that have occurred in recent years. So there are a lot of technical challenges with making sure that personal data is secure for a start. So I think brands have the first responsibility for managing data, to run a personal data in a responsible way. But then you cannot ignore the huge where a huge proportion of their revenue is coming through advertising and the use of data, so what we call the walled gardens, the likes of Facebook, google, data passing through their systems, and I think the tech companies probably should do more to protect privacy. In many cases, the big tech companies have been brought to the table by regulators in the EU and in the US. The EU, for example, the European Union, has been very strict on data protection and privacy legislation. That's only gonna increase as AI starts to take hold in advertising, which it is already no question, and that creates even bigger challenges about security and privacy. 

0:14:48 - Mehmet
Like you mentioned, I and couple of like let's call them trends, right so that are happening in this space, mainly AI, right? So what other trends are you seeing happening? Because, from what I'm hearing and a couple of discussions with friends offline on this topic, so they are saying, like the ad tech industry is on the verge of a major transformation Now when. I try to understand more. I hear mixed opinions so I'm curious, as I was explaining to you. So what are the trends that we might see. 

0:15:27 - Ian
Okay, what kind of transformation are you hearing Just out of interest? 

0:15:33 - Mehmet
Yeah, like some people go to very extremists and say like maybe the whole way we do advertisement, as we do it today on the internet, might change and this would go to something completely different. You know like we're gonna go all the way with something like metaverse VR. 

you know these kinds of things and sometimes we hear people they say no, like this is something here to stay, Although, like with all the changes and regulations that comes in, still big brands would have the power to navigate, let's say, the regulations. So I'm hearing opinions from two different, I would say, sides, and each one is like interesting discussions. I would say so, but from someone like you I am like who's inside this I would say I would hear something more accurate. 

0:16:23 - Ian
Yeah, sure, and I think we have to accept that. Ai is going to transform advertising and ad tech for sure, and it already is on the generative front, but we can come back to AI Elsewhere. There's no doubt that Web3, as an always on high speed, very interactive, highly personalized technology, will change advertising. I think the metaverse, so augmented reality, and techniques like that will steadily increase in sophistication and in relevance to the consumer. But there are some major barriers, such as the cost of devices and headsets and whether we really want to spend lengthy periods of time with a headset. I have my doubts about that, I think, for certain segments, for certain parts of the population so gamers, for example, very used to in-game advertising and in-app advertising. I think what it does, what all of those trends mean, is a massive opportunity for advertisers, both an opportunity and a challenge to decide where do they invest. Do they do this in-house or do they invest with technology partners? And if they choose to invest with technology partners, are they losing the insight that it's actually necessary to grow their brand and to build lasting relationships or not? 

I think for what you would class as the traditional ad tech providers, the likes of Google, again facing major challenges. 

You can see that meta have chosen to entirely rebrand as meta and to put billions into the metaverse as a chosen strategy, but we're yet to really see a significant payback from metaverse investment. 

Google, with their investment in things like DeepMind, is choosing to add AI to its ad tech platforms and the likes of YouTube, for example, so they are taking a somewhat different approach. And then you've got other more recent entrants in the large-scale technology companies like Amazon, who have their own DSP and obviously have a lot of data about your shopping habits. And, as long as it is back to the earlier point about privacy, if Amazon are able to leverage the data asset that they have, they are in a very powerful position because in the end, the purpose of advertising is that you buy more and obviously Amazon knows an awful lot about what you buy and when and your views and your feedback and so on. So they're in an interesting position if they can leverage it and they have their cloud computing, aws expertise as well. So they have the horsepower and they have the data. The question is where they're how their strategy develops in the ad tech field. 

0:19:36 - Mehmet
Are we going to see any consolidation in the ATEC space? 

0:19:42 - Ian
I am yeah, absolutely lots. I think a lot of the small to medium-sized companies will struggle to survive because of the investment that's required to maintain a viable solution in the face of huge competition that have very deep pockets Apple, amazon, facebook, google. These are hugely cash-rich businesses that can afford to invest and they do huge percentages of their income into R&D and they can afford to invest in areas that do develop and invest in areas in some cases where they don't develop, and they can afford to take that risk. So the smaller to medium-sized companies are going to struggle to find the investment and they're going to have to find niches in the ATEC space. Whether it's content or relationship management or automated design or whatever it might be, they're going to have to find the niches where they can compete. 

0:20:52 - Mehmet
Now coming back to the AI point, and by the time we broadcast this to the air, it will be maybe two weeks time from now, but just today, this morning, I've seen the news from OpenAI that they're going to start to speak and being able to take images and you chat with it based on the image that is what I'm asking you guys to find From an ad perspective. Do you think now any company that is in the AI space has the capability to become an ATEC company? 

0:21:35 - Ian
Yes, up to a point. The beauty of chat GPT particularly GPT-4, is the sheer amount of data that the models have been trained on, and that's why they're called large language models. For most tech companies, there's not a great deal of point in trying to replicate the scale of data that these things have been trained on. In OpenAI's case, it's a little bit different. For images, they're still training their models for image detection and production, and you could say that voice is actually relatively easy. Once you've solved the text. It's actually relatively easy to put voice on top of a platform like that. Now that means, as a technology asset, hugely powerful to any business that's looking to develop what you could call ad tech capabilities. We use chat GPT's APIs for text and we customize their APIs specifically for the briefing part of our technology, because there's no point in us replicating the text mining capability that something like chat GPT has. The same thing for Dali 3 and for images. 

The interesting thing is that there's 100 applications out there that will generate an image, what we call text. You can generate a film or a photo or a piece of text. That's the generative AI phenomenon that we've seen over the last year or so. That's going to continue. Absolutely no question about that, but my view as a technologist is that that is not particularly clever. You can randomly generate a set of images, as long as you've trained your model adequately on historic data. What's really meaningful is which image do you actually need for the application that you want to put it to? How do you decide which image is most relevant? That means understanding the data that's connected to a combination of images. That's far more interesting. It's far more challenging and is definitely the way the tech industry will progress. Now going to the point of using the phone. 

0:24:06 - Mehmet
I'm generating ads. Generating ads and it takes the phone. Sorry, you're breaking up. 

0:24:12 - Ian
at the moment, I can barely hear you. Can you hear me now? When you speak? You're speaking like a robot. At the moment I can't actually hear you. 

0:24:25 - Mehmet
Is it better now? 

0:24:28 - Ian
No, I can only just hear you. Just talk to me one second. 

0:24:35 - Mehmet
My mic is working fine. I can't really hear you very well. Just give me one second. It's both this. So the question I was asking you, ion, is you talked about using AI for generating ad text, images and even we saw video. 

Now some experts say that, like, how much creative can AI be here? Because actually we are using an already generated data and we're just, you know, giving it to an algorithm to repurpose it in a different way. Is it true AI? Because some people it's a very famous saying now, if you turn it to a garbage, it will give you garbage. But I don't think we give it only garbage and I think we gave some really creative thing. But how far do you think AI can be really creative when it generates, like, let's say, full-blown a marketing campaign including, you know, screens, image, direction and video? 

0:25:56 - Ian
I mean. The answer to that question is there's no question that AI can be used to deliver the entire campaign cycle and all its associated content across all channels, as long as the channels are connected and the return path data goes back to the underlying algorithm. What's needed is technical integration of different media channels for a start, which means all the data that a set of models might see are integrated at some point and that those models include machine learning in some form or techniques like what we call reinforcement learning. That means we're effectively rewarding a model for its ability to achieve a certain result, and we use neural nets for the same purpose. And it's actually the same technique that Tesla is using for driverless cars, for their FSD version 12, is a neural network that's learning about the data that it's receiving in real time, processing that in real time to enable a car to make decisions. 

And really what we're talking about here is very much the same technique but being applied for a completely different application. So no doubt in my mind that over time we will see the script, the content, variations of that content being generated in real time, with increasing levels of relevance to the consumer, and machine learning effectively taking over large parts of the production cycle for advertising. Where there are some limitations is you do see instances where creative agencies invent a crazy idea for a campaign that somehow resonates with an audience, and that's probably the most interesting area. At what point do computer vision algorithms learn really, really innovative concepts let's call them that for campaigns, that where it's not been exposed to that kind of data in the past. That's the tipping point that we're not really seeing yet, but it will come and it's neural networks that will really drive that. 

0:28:31 - Mehmet
How long do you think it would take us until we reach this? Or let me, if I want to rephrase the question, how the space would look like after. Let's say I don't want to say five years, because now when I say five it's too long. Let's say in three years or two years. 

0:28:49 - Ian
Yeah, I think in two to three years we'll certainly see a number of campaigns that have been entirely automated and have been proven to be faster and more relevant to a given audience. We will certainly see that, and I mean every single aspect of the campaign, end to end, including just multimedia delivery across media channels. 

0:29:17 - Mehmet
Mm-hmm, like you mentioned something about, you know cloud and spendings, and maybe it's you know as a CTO as well. You might give us some insights. So where are you seeing, you know, the technology spendings heading? You know? 

0:29:36 - Ian
in the near and long future. Yeah, I mean. Most organizations, including ours, are not processing data on-premise any longer. It's cloud-based. There are organizations that run on-premise. You know the highly regulated industries, for example, may run on-premise databases and cloud-based infrastructure. 

We are entirely cloud-based ourselves. We use AWS. We're not multi-cloud. There are some organizations that are multi-cloud. There are some ad tech providers that have to be multi-cloud because there are some brands that will say, well, we don't want our data going through Amazon because we're a retailer, so we don't want to see our data going through AWS, or we don't want to see our data going through Google's platforms. Therefore, you do see ad tech companies that are multi-cloud, so they'll have Azure, aws and Google Cloud, so that's increasing. But it's one of the reasons that's one of the things that's fueled AI in the last year or two is the compute horsepower has increased exponentially and the cost of that horsepower has dropped significantly, and that means the amount of data that we can process cost effectively has increased significantly. And that's an absolutely crucial aspect of the growth in AI as compute horsepower and the cost of that processing. 

0:31:12 - Mehmet
And out of currency. How much is important, especially for ad tech companies, to invest in solutions like data analytics and these kinds of solutions, and data mining maybe also. 

0:31:27 - Ian
Oh, it's essential and it's in technology these days and it's the. The best career path is data science. So data engineering, data science, machine learning, ops are huge growth areas in terms of career development. So brands have to invest in data science as a discipline. Ad tech companies obviously rely on data science as a discipline, so they have to invest in those skills. I think the interesting question for brands is how much of that expertise do they add in house or how much do they rely on partners? 

And I think for the large agencies out there, they're at a real tipping point. Ai is going to continue to cause them some strategic pain let's call it that where they've not really. The large agency networks have not really invested in data science as a discipline, and the knock on effect of that is they've not been developing the repositories of data that they can then apply to AI going forward. They've been somewhat left behind. 

0:32:42 - Mehmet
Yeah, you know, as we were coming to the end just now, I target CTOs and I target also, you know, to be founders or new founders. Let's say From your experience you've been in this for a long time what advice you can give for someone, first, who's interested in your niche, which is the ad tech. You know, like what are the things that they should care about when they are building their startups. And second, you know general advice for any you know tech technical founder who's just starting his venture or just starting his career. 

0:33:32 - Ian
Yeah, I mean for startups. If you mean a startup that's going to have some kind of technology product, then they've got to be very clear about the niche that they're trying to address and not assume that they can cope with huge technology companies that have much bigger budgets. And then they need to be very clear about what their go-to-market strategy is. How are they going to reach the audience for their chosen product? And most technology startups completely fail at this. They misunderstand the niche and then they don't adequately understand how to take their product to market in a very efficient way In terms of individuals starting out, then if it's the technology career path, then absolutely learn things like go open source, learn things like TensorFlow and Python, things like PyTorch, but also things like prompt engineering is going to be a major area of growth in the technology space over the next year or two. 

How do you configure prompt-based systems to provide intelligent answers to very complex questions? So we call that prompt engineering. It's a big growth area. But make sure that you've got a solid grounding in maths and statistics. Some of these are the underlying equations in AI. Some of them are quite complex, so an underlying grounding in maths is a good thing. It's open source. Python and Spark are the leading tools that they should really acquire if they want a technical career path. 

And I would also say to the marketing professionals I've got to learn much more about the technology, the data, the strengths and weaknesses of various technologies. I've got to have a far better understanding of these things if they are a non-technical decision maker in a brand or an advertising agency. Yeah, actually yeah, please go ahead. Aydan no, I just said they have a lot to learn, in my view. 

0:35:48 - Mehmet
Yeah, so like one thing maybe we repeated on the show too much also as well to your point, like every company now needs to be a data-driven company, and this applies not only for, let's say, the IT or the technical folks, it applies to every department in the organization so they can take better decisions, they can understand what's happening, they can understand their customers. I'm not sure like, did I miss to ask you anything? This is not a tricky question, but really is there anything that you wish that I had asked you? 

0:36:31 - Ian
That's actually quite a good question. It's a nice open question. I don't think so. No, you've asked a very broad range of questions. No, I can't think of anything that you haven't asked me that I thought you might. 

0:36:45 - Mehmet
Okay, you're up to good at this, I don't you know. One guest saw that I'm doing a trick and I said, no, I'm not doing a trick really, because sometimes, you know, I do the show daily, so of course I take my notes. I have the question that I prepared, you know, but we are human beings and maybe I have missed something down the road. So, just to make sure that maybe an idea that you wanted to highlight, I mean for the guest, they can have this window, let's say, to share this. But anyway, thank you very much. I and I really enjoyed this. I learned a lot about ATTEC today from you, and you know some of the things that we are expecting to see in the future. So thank you very much for enlightening us around this. 

And of course like same here and, of course, like you know, it's always good for the audience. Yeah, so, so it's good. Yeah, so thank you very much for tuning in and thank you, I am again, for joining me today and, as usual, this is how I end up my episode. Please, if you have any questions, any feedback, don't hesitate to reach out to me. And thank you very much for tuning in. We'll meet you soon. Thank you, bye, bye. 

0:38:13 - Ian
It's been a great pleasure. Thank you. 

Transcribed by https://podium.page