 Ladies and gentlemen, to enlighten us on how hyper personalisation and artificial intelligence are ruling this space, let's call upon Mr. Tarun Katyal from Z5. A media veteran for over two decades, Mr. Tarun Katyal. But I would like to speak something about you when the media mogul is beside me. I'd like to take this opportunity. Mr. Tarun Katyal is a media veteran for over two decades. Mr. Tarun Katyal, sir, is the CEO of Z5 India in his current role. Well, he's a TEDx speaker and part of industry forums, including jury chair of Air Club AB and DMAI, committee member with FICCI and CII, XCOM at IAAA and vice president at Association of Radio Operators for India, AROI. So ladies and gentlemen, let's give it up for the super energetic and ever-enthusiastic Mr. Tarun Katyal from Z5 who's already here with us. God, that was a long introduction. Thank you so much. I know this is a session post, I didn't think post tricks, but post coffee. Interesting times and an interesting topic. I hope I can keep you engaged for the next 10 to 15 minutes. We'll keep this crisp, but I hope we can keep it interesting. They talk about hyper personalization and they talk about, you know, the need of the changing consumer and being served to a segment of one. Now, I think, you know, most of us as marketers and advertisers, as agency folks started in, you know, in the late 90s and early 2000 looking at target audiences, it was fairly broad. We were told, you know, you can look at four plus, then we were told you can look in the upper cap of 25 to 44. You could add gender to it. You could then add geography to it. Then there were some studies that could have some lifestyle attributes to it and so on and so forth. But then came digital about five to seven years ago, which started to do user level attribution and started to understand user cohorts and segments. And then, you know, we all got very excited. Google can tell us more than what we know about our audiences. They can tell us what they consume, what they do, what they eat, you know, and so on and so forth. And all of that data that we passed to Google search and many other platforms are on an everyday basis. And that then led to what we call the segment of one. Now, what is the segment of one and why do you need hyper personalization, right? Most of our user journeys today are, you know, taken for granted. So when you get onto a food delivery platform, you expect the food delivery platform to understand what you ordered last, to have descriptive data to be able to tell you where and what are your likes and preferences, to have prescriptive data and recommendations and lookalikes to be able to tell you what you possibly will like and where you could order from, to have your payment history and to be able to understand what kind of discounts you would like, what kind of cards you own, what kind of, you know, payment history that you've shown or you've displayed over a period of time, all that is kind of taken for granted. Why is it taken for granted? It's because each individual today believes that they must be served to their individual level rather than even a segment level, right? Now, where does that take, you know, us as marketers? Where does it take us as product owners? Where does it take us as businesses? Now, all of what looks very simple is actually very complex. It's complex because, you know, whether it's a video delivery platform like ours or whether it's a food delivery platform like Zoomato or Swiggy or some of the others, or whether it's a search platform like Bling or Chrome or Google or so on and so forth, each of them have to deal with millions and millions of users, millions and millions of profiles, much understanding of, of being able to predict how each of those users will behave, much understanding of where they will go from where, what are they going to be their user journeys, what are they going to be their likes, dislikes, and what do they really want to end up at? And you're not going to go right all the time, right? The 20 things that come up on your search on any platform are not always going to be correct. And so there can be intuitive search, there can be descriptive search, there can be prescriptive search, there can be recommended search. But all of it is continuously learning, right? Machine learning is, is a very broad subject and, and artificial intelligence machine, machine learning are almost two words that are most abused in data sciences today. Everybody believes that their tool or, or their product or their platform today is well equipped with each one of them. What do we expect some of these things to do? What do we expect the user to do with them? How do we really look at uplift numbers when we use them? Where do we go when we get to a point of saturation on using them? Is only descriptive data good enough? Are recommendations really the way to go and are they getting into data privacy? What happens when things like GDPR and data privacy laws come into countries like us? What happens when opt-ins become lesser and lesser? These are questions that all of us will and continue to answer. The world is extremely dynamic when it comes to machine learning. Today, you know, you act, you allow access to much of your data on an ongoing basis. Unfortunately, as you realize what is happening with this data and there are in, you know, enough and more reports about how your conversations are being recorded by some of the tech companies or how most people know what you interact with. Are you willing as users to be able to give as much as you give today? And if you don't give as much and if you don't believe that, you know, you want to part with your personal data, then how will the machine learn and come back to you? So the balance between machine learning and data privacy, machine learning and descriptive data is something that, you know, all product owners and all marketers today are struggling to keep. Now there are things that, you know, all of us have learned over the last, I would say, three to four years, one, that however, you know, you want, unless your user journeys are intuitive and unless you continuously look at data to be able to create more efficiency and improvise them, you're not going to be able to hold on to a user. Now, this also looks very broad, but I'll give you an example of, you know, our own platform called Z5. When we got onto the platform, we realized that there were three things that were essential to hold on to users, to be able to create better DAUs, better MAUs, better retention, better time spent. And this is what we call the strategy of three Vs, which was, the first of them was video. Now, video is kind of a broad word, but we realized that serving video was as intuitive as it, you know, as anything else. India is a country of humongous amount of device and OEM ecosystem, right? So it starts from something that is not even in most other parts of the world, which is a feature phone that delivers video. So the geo feature phone, which is the lowest end device, which you can buy at 50 rupees monthly and, you know, and a 1500 bucks one time cost, also delivers video at one end. And the smart TV, which delivers 4K with Samsung and LG and some of the others, also delivers video at the other. The adaptive betray technology that really needed to be put into place to understand what platform is the video going on to, what's the bandwidth of the platform, what's the ability of the platform in terms of serving video, what's the memory of that phone, what's the CPU of that phone, how will you deliver video in the best way, how do you make sure that there are lower and low video start failures, lower and low exit before video starts, and the quality of experience that can get better and better. That in itself was technology that India hadn't seen or even the world hadn't seen that we had to work towards over the last 18 to 24 months. We partnered with some of the, you know, cutting edge technology developers in the world like AWS Elemental and the others to be able to do just one thing, which is serve good video. Because we knew that on a video streaming platform, forget all the UI, forget everything else, people eventually come to consume good video. And if you can't give them a great video experience, you're not going to be able to retain them. And that had to be personalized to their device, to their memory, to their ability to hold on to downloaded videos, and so on and so forth. And that intelligence of every single individual needed to be fed back into a video player, into our transcoding and encoding systems, so that in the pipe, when the video came back finally to you, it understood how you consumed video and to be able to serve that well to you. The second one was language, what we call vernacular. We realized that India lived in its languages and Indians were not unidimensional. As most of you may not know, even Google and YouTube are only available in six languages. We are available in 11 to 12 languages today. And that was a big ask of us. So to be able to do display in those many languages, to be able to do UI in those many languages, and most importantly, to be able to create search in those many languages. And that was a big ask again, because most phones didn't support so many languages on their keyboards. So we had to then move to voice, which was our third V. So we created voice search in those many languages for everybody to be able to do content discovery. And that, between vernacular, voice, and video, was an extremely personalized, hyper-personalized experience that was built on artificial intelligence and machine learning for us to be able to get to the number we did. And how we've improvised this over the years has led to every quarter and every spin showing us enough delta, enough uplift for us to be able to serve more and more consumers and more and more languages in more and more content of their choice. And this, I must tell you, is an ongoing experience that has got us to having not one app, but 15 applications in the market today. You have a different application for iOS. You have a different application for iOS. You have a different application for Android, for LG, for Tizen, for Samsung, for Xiaomi, for AFS, for Apple TV, and so on and so forth. Unless you have those many applications that are catering to every single individual and with video players that can understand, and different video players that can understand adaptive bit rate, that can understand your ability to consume data, can understand what kind of data plan you're on, slow 3G, slow 4G, broadband, and so on and so forth, you will not be able to create user experience that would proliferate the brand into hundreds and millions of people in this country. This is really easier said than done. And increasingly personalization or hyper-personalization is something that we should take back into marketing, into building media plans, into being able to do targeting. So right from a 4 plus, 12 plus kind of a media plan to getting to cohorts and segments, to getting to an ability to personalize advertising at an individual level, it's a journey that all of us are taking. I'll also talk about another tool that I was telling a colleague right behind. We partnered with somebody, somebody called Minitly, to be able to create an auto preview technology. What is an auto preview technology? An auto preview technology is very different from what you believe in auto play technology. What's auto play? Auto play is when you get onto a video, it starts to play on its own. And it's a recommended personalized video that starts to play out and starts to show you what you want to get to without you having to even trigger the play button on the video. The auto preview technology takes a step further. You've got hundreds of curated tiles on video OTT platform or a video streaming platform. The auto preview technology scrubs all of the 30 minutes, the 30 seconds that you are going to most likely like to see out of the 30 minutes, and creates an auto preview of the video out of all the 10 videos that you can see, it will pick one or two videos, scrub those videos, and then auto preview the video on an ongoing basis at a hyper-personalized level to you. That is the level of personalization that people are getting to or video platforms are getting to. So while there's one level of understanding the kind of idly that you will want to order on Swiggy, the other level there is the kind of video and the kind of scrub off the video that we'll have to watch. And this is not just ending, right? Every single onboarding, every single video experience, every single payment plan is getting to a point that you can be able to personalize anything you want to. At some point in time, we want to get to a subscription plan that can allow you to choose the amount of content and build your own subscription plan on the fly. And things like this may or may not happen, but the journey to hyper-personalization has just begun, and I welcome marketers, advertisers, and everybody else to join this journey in being able to serve the segment of one. Thank you so much.