Latest / The Collaborative Canvas Podcast / Former Nestlé Global Insights Head Reveals The Future of AI & Consumer Research |Using AI Completely
Transcript
- speaker-0: How do you see, you know, the new era of AI into the market research world and how people are actually reacting to it? AI is speaker-1: Everywhere, it's not just in the world of insights. Intelligence and human insight and understanding need not be a static phrase. speaker-0: Human intelligence versus AI intelligence should not be a versus, but it should be a culmination of both. speaker-1: Wherever you know there is value in AI, make use of AI and you know it'll be able to do things maybe you know. speaker-0: Consumer surveys or passive behavioral data AI co-pilot or AI decision maker. One matric marketers don't pay enough attention to. speaker-1: Passive behavioral data, but They I co pilot always. speaker-0: Welcome back to the Collaborative Canvas Podcast. For years, businesses have relied on research, data, and consumer insights to understand people and make better decisions. But today, we're entering into a new era, one where artificial intelligence, connected data ecosystems, and digital twins are changing how organizations listen to the consumers, predict behavior, and act on different insights. What does the future of consumer understandings look like? Can business move beyond periodic research projects? Built always on insight systems. And as AI becomes the new powerful that role has already emerged, so we're thinking what role will human judgment continues to play. To help us explore these questions, I'm delighted to welcome someone who has spent decades at the forefront of insights, strategy, and marketing, and business transformation. My guest today is Chandan Mukherjee, a respected industry leader whose career span leadership roles at Nestle and Unilever. At the firm as the former global head of insights at Nestle, Chandan has helped shape growth strategies, strengthen brands, and build consumer-centric organizations and transform complex data into meaningful business intelligence. He has been recognized with numerous industry accolades, served on advisory boards across the marketing and insights ecosystem, and continues to be a strong voice on the future of marketing, analytics, technology, and human-centered decision making. Today we will discuss digital twins, always on insights, AI power, decision intelligence, and how organizations can build systems around insights that don't just report the past but help shape the future. Chandasar, welcome to the show. speaker-1: Hi, pleasure to be here. Thank you so much for having me. speaker-0: Thank you so much for c making some time out and ⁓ to start with again, you know, the same usual questions. how do you see, you know, the new era of AI into the market research world and how people are actually reacting to it? It's pretty new and ⁓ there are a lot of questions which are going around. speaker-1: I think, you know, we live in dynamic times, things are changing, and ⁓ AI is everywhere. It's not just in the world of insights, but it's everywhere. And therefore, AI is also coming and fusing into the world of insights, into the world of marketing. ⁓ there have been a lot of developments on the technology side, there have been a lot developments on the side of using AI. It's still maturing, I would say, but there's a lot of scope to kind of build it further and a lot of scope to embed it within the organizations. Because intelligence and human insight and understanding need not be a static frame. Now it's gonna be a little bit of a moving camera where you need to have things, you know, always coming in, always getting picked up. That's the heartbeat of the organization that organizations need to embrace and really embed insights in an always on way. speaker-0: So earlier when we see, you know, with the decades of market research that I've seen myself, it was very project ⁓ based. And we spoke about insights as a system. So how does ⁓ you know that reflects in the current ⁓ era are are people actually thinking and they're moving towards insight based systems or they're still thinking that let's wait and watch and see how it goes on? speaker-1: I think different organizations and different kinds of, you know, different kinds of groups are moving at different pace on this, ⁓ on this entire topic. But I think you need to move there. You need to move into a system where insights is always on. It is there, which is there by your side all the time. It's no longer a project or a study, it's a system. And that's the big shift that people need to have. I mean, there will be different kinds of, you know. ⁓ studies and projects and researches that will go on. There would be different kinds of methodologies that will be in place. There's so much of data around, you know, data from the world of understanding people, from the world of affinities, from the digital world, from the commerce world, and so on and so forth. I think incites us to really look at it in a very holistic manner, to pick up signals from all these different areas in a very, very precise way, in a very, very real way. And then use AI in the right way again to be able to see how you can make the most out of it. So it's it's it's a system where we have AI on the side always. It's a bit like, you know, you use any any of the regular AI tools, you know, on your phone or on your computer, you know, now, then, every time. And this is going to be one of those where as an insights professional, as a marketing professional, you will have access to these kind of tools around decision intelligence, which can be always there by your side. And you can ask a few questions, get a few things out, you need to understand what's going on. Human inside, of course, will play a very important role. It will still be the centerpiece. But with AI and all of that, of assimilating so much of data, of using so much of data in a very, very powerful way, and really kind of you know making that ⁓ into something which is meaningful from a business context point of view, ⁓ would have a lot to do with how we use AI and how we actually fuse the human world and the AI world together. speaker-0: Sir, I I think you've been ⁓ doing a lot of ⁓ these advisory boards and you've been talking to a lot of newcomers these days. What kind of industry shift are you seeing with respect to AI? And ⁓ for example, you know, we are seeing digital twins, we are seeing synthetic data, we are seeing qualitative being moderated. And I think I've been hearing that there would be ⁓ more of ⁓ AI ⁓ respondents who would be, you know, into these qual. So how do you see ⁓ you know, if we just talk about, you know, does it Does are people very excited about it? Because there's always a debate which is already going on. And what is the most concerning ⁓ part ⁓ you know, when we see those evolutions are happening, you know, whether it is digital term synthetic data? speaker-1: Well, I think it is a very fast evolving system, yeah. And there are a lot of experiments going on, different things are being looked at, different things are being examined. ⁓ there are validations going on between what you know different AI tools gave versus what some of the classical tools gave. Also to understand what is behind the AI tools. You know, that's very, very important, to understand what's going behind what's really the sum and substance of what goes in into into the tools. Where does the data come from? How do you really look at it and so on. So You know, I think that the majority of the data coming in would still have to be real. We're living in a world of real people, real consumers. You know, human insight is very, very core. So if you have real data coming in, getting fused into AI systems and AI systems are able to kind of, you know, draw out different insights. You want to test something, you want to predict something, you want to synthesize something. You want to build something and join some dots together to see what's what's really going on in a particular topic. There's a lot of role of AI there. There's a role of digital twins, of course. You know, digital twins are not just a replica of a person, it's more about a replica of a decision-making model and a decision behavior, you know, thinking model. So as long as you know we understand what these are about and how could they be really useful. And sharpen them and evolve them, I think there's a lot of scope to be able to embed them and that could be a very relevant piece for people to look at. speaker-0: I think you made a very valid point. But considering that, you know, data already stays for there for y for a very longer time. I mean, every other corporate organization has abundance of data. But there's always a gap. I mean, I think ⁓ if we go back five years or ten years ago, I was actually attending a conference ⁓ earlier this year and somebody talked about that big data was already there, but still the gap was predominant. It was there, people were not able to use that big data in a very vital way. And then ⁓ now AI. Well, Westerners are actually using it. ⁓ you know, I think Asia Pack is still thinking w whether we should implement AI or not. My question is that, you know, if the data already existed, what is basically, you know, stopping them not to use say digital vent or synthetic data? Where is that trust being gone? What is that gap? speaker-1: Well, the data has been there for quite some time and data is also getting multiplied all the time. As we speak, there is more and more data coming in. And the velocity and the quantum of data over the last one or two years is far more than the velocity and the quantum of data in the next, you know, in the previous two years or the previous, you know, even a decade or two decades. So it's it's multiplying at a very, very rapid pace. Now what you need is not data in silos. You need to first of all connect different pieces of data. That's very important. And then to be able to draw meaning out of it. So while there is a lot of abundance of data, we still lack understanding, putting together all the data, and then insight from the understanding of the data. That there is a big gap. And that gap still exists. And that gap may be partly conquered by AI, which is able to digest and assimilate so much of data. But you all also need a lot of human ⁓ understanding going behind it, bringing in the business context, bringing in what's really ⁓ useful from a business point of view, what's really the human underlying motivations, what's the texture of the data behind and why is it coming through in that particular way? And to answer the the root causes and so on and so forth, you would need a lot of human overlay and human insight on top of that. But making sense of so much of data is something that AI can help to get to. And and therefore, you know, we need to find a way of how we look at AI systems which are able to take, you know, so much of data, maybe assimilate data at ten times the speed, you know, with maybe five or ten times the precision of the data put together and then see, you know, how it works. Of course, AI has its own, you know, shortcomings as well on hallucinations and other kinds of things, which also need to be addressed. But you need to embrace, you know, technology and you need to see how you can move insights to decisions. Because that's the other gap. You know, one gap is between data to understanding to insight and then from inside to be able to take decisions in an always-on manner because we live in a world which is very dynamic. So you can't have insights which are only coming in, you know, as waves or as pieces. You know, you need now more continuous pieces coming in and this is where ⁓ some of the EI tools can help, of course, with a lot of human strategy and ⁓ and understanding on top of that. speaker-0: So I think let's take the first example as it's a Kyapnikaki, there's a there's a gap. The f let's t discuss the first gap. Considering that we have data in silos, they're sitting everywhere in in within the organizations. And ⁓ let's say an FMCG company. Now converting that data to make a meaningful insights take a lot of time when we actually plug that and you know, basis on the organizational objective of understanding what consumer actually wants basis Again, you know, by doing surveys or, you know, finding out other avenues. We we have been talking about first party data, zero party data, second, third and whatnot. This whole gap can be easily filled by AI. We know that, of course, considering there are a lot of hallucinations which are also going. But isn't that, you know, it it does take a lot of time and when we say that AI is already in the play, ⁓ and you know, we can decrease the time by by a large margin, ⁓ does People still thinking that, you know, ⁓ we should collect survey data to rather faster the ⁓ you know, or probably like more like converting a organizational objective as early as possible rather than you plugging every single data to find out what else we can do with that data because it's pretty scattered. You're not too sure where this data is coming from. And that's again the biggest point of conversation which I see on the on most of those debates that I am not too sure where this data is coming from. So how would I plug that data and make a meaningful insight be even before I fill the second gap? speaker-1: I think there are two different parts to to your question. One part is about, you know, there's a lot of data coming in. What kind of data is really useful and what is the, you know, what are the different pieces of data? And the second part of the question is about, you know, what do we do with those different pieces of data? Are those meaningful or not? In the very first one, you know, you need to always start with the problem statement. What's the business question? What are what's the question you're asking, which is still a very human question. It's very it's still a very strategic question. And that organizations need to first understand, you know, what are the decisions that they want to take, for what? And there are two levers of essentially taking decisions. One is about, you know, you want to grow, and the second is maybe you you want to see what your investments are and how your investments are paying off or not. So in growth and investments, and you know, you can look at marketing investments or other investments, you need to understand what's the fundamental question. And then think about what are the data pieces that you need so that I can answer those questions. And survey is going to be one part of that entire data spread. I mean, there are surveys, there's passive data, there is digital data from so many sources. You can have zero party, first party, second party, third party data, all different kinds of data. It's also about, you know, people talk a lot about first-party data, but first-party data. Also, of course, has to be, you know, collected in the right way with all the different regulations and the compliances and privacy and all of that. But also you need to see after you've collected all that, you know, compliant data and using all that compliant data, how do you really use it? How do you really enrich it? Do you have enough information to really use behind all that data? And that's where you know consumer data platforms and various other kind of things also come in where you can use. Different types of data for different business decisions and to meet different purposes. The second part is about, you know, what data are you getting? Where is this data coming from? What does it mean? And that's very, very important. You need to understand what does this data mean, where is it coming from? How is it collected? What does it really mean? Is it representative? You know, what does it really talk about? What's the signal you know that we are getting and and what's what's the conclusion that we can draw out of it? And that's where I think still a lot of human rigor has to come in. To understand what the data is and and where is it coming from and what can we do with it and what can't we do with it? Because sometimes you know just because you have some data, it doesn't mean that that's the data that will be really useful. Sometimes we collect some data and some many different metrics, but they do not necessarily tell you ⁓ what's the what's the benefit, you know, because they don't answer precisely what you're looking for. It's not a it's not a good enough signal to see what you're looking for, and that's where metrics that matter comes in. Which essentially means what are the metrics that will be really useful, which matters, which shows something about what the consumer is doing or behaving or what action I can take. So data understanding insight and then decision and action, this needs to be the whole piece of the puzzle that we we we need to solve for. And within that, of course, there will be a role of AI, there will be a role of human insight. And both artificial intelligence and human intelligence will will when fused together can really deliver in that direction. speaker-0: ⁓ you talk about human insights. ⁓ and in today's world, people are actually going AI everywhere. I mean, even a c normal consumer is searching about a different brand on on different chat GPTs or, you know, all other LLMs. So when we see ⁓ an overall consumer understanding, it is basically coming from research, insights, intelligence and then prediction. Are we entering a new era altogether? speaker-1: Together? It is a new era, of course. I think there are two dimensions of the era. One era is from a static and periodic picture, we are moving into an always-on picture, you know, more like in size as a system or something which is continuously there all the time, by your side, with you, along with you. And the second element that you're trying to have is that there are many more sources of data. So you need to really be able to connect all those different sources of data. And and see what is it which would really be meaningful to the business question and the strategy that you're looking at. And that's where, you know, the the entire area of predictive intelligence, of decision tools, and all of that comes in in a very, very clear way. Not just in terms of making sense of all the data, because it's just too much of data. Human, you know, human tools will or human thinking will not be able to manage so much of data in such a very a precise way in a short period of time. That's where AI can help. But AI can also help on the decision making and the predictive side of things, you know, on in in trying to see what sense can you make out of the data and what are the scenarios out there. speaker-0: So you talked about digital twins in in our initial conversation. And I remember a lot of qualitative ⁓ researchers used to talk about persona back then. When they were actually plugging the data. I mean, there were not many sources earlier. It is basically those passive data or the FDIs that we used to do. ⁓ Those were basically the data points to create a persona. Now they're there it has this data has been multiplied by a number of ways that it is getting collected. People who are actually creating digital twins across the globe, they're they're not declaring their sources, other than the fact that, you know, sometimes they collect survey responses, sometimes they collect from different apps. Now, collecting that responses, few of the researchers believe that these are not the right way to source the data. And I'm specifically asking about this because a lot many people don't even trust, you know, even if we are entering a new era, people are not trusting this data. And they wanted to validate it. How can this validation actually work in in an AI environment when we are saying that human intelligence versus AI intelligence should not be a versus, but it should be a culmination of both. speaker-1: So on the data sources, I think organizations have to become more and more transparent about what data they are using. Of course, need to collect data holistically, you know, understand what data is coming in from classical traditional survey sources or various other qual and quant or or tracking or any other kind of research. But more importantly, you know, some of the behaviors and the digital footprints that people leave behind on on digital ecosystems and some of the behaviors that they exhibit. There are also, you know, data like commerce data and other kinds of data which are also there. Now these are again with specific partners, with specific organizations. So you need to see, you know, what kind of data can you collect and what's meaningful and so on. but that's one part of it, which is essentially about trying to see what all data do you have. Because you need to have a holistic picture across consumption, shopping and engagement. So all the three different elements and looking at a human being as a human being, you know, to to try and understand from a human lens on different dimensions and then try and see, you know, how you could plug in. So you need, of course, a very holistic way of looking at all this data and so on. And people have to get more and more transparent about what this data is. Of course there will be limitations of what data you can access and what data you cannot access with the with the cookies and other things and all the other you know elements around it. There would of course be ⁓ you know, so a situation where you will not have access to all data. But you should try and see how much data you can access because data will continue to be power and and and therefore you need to have all the data together. The second part of it is trying to see, you know, what do you get? What does a digital twin really mean? You know, consumer personas or or pen portraits or whatever, they have been there for quite some time. But we have always kind of thought of that as a very As a very static piece. Now it can be much more dynamic because a person, when you are going to work on a Tuesday morning, you're not exactly the same person, you know, in so far as your behaviors are concerned, to when you're going to, let us say, you know, play a game of tennis on Saturday afternoon or evening. Yeah. So you're you're, you know, kind of you might be the same person, but your context, your behaviors, your mindset, and these two different situations can be a bit different. ⁓ digital twins, you know, try and look at, of course, your consumer and try to create a twin out of that. But more importantly, it's not just who they are, but also about how they decide. And what are the behaviors they talk about. We understand a little more of that. Because that behavior and the example that I gave you between one day and another day on a different in a different context. or in a different, you know, kind of a situation or with a different need or or something else or maybe in a different category context, you might have different drivers playing in those two situations. And you really need to understand the drivers that are meaningful for that context. Then only would you be able to use that in a more definitive way to be able to take a business decision and to be able to generate some action out of it. And therefore it's it's, you know, it's it's got an added layer. And a more continuous layer and a more ongoing dynamic layer, for example, that could come in with some of the other forms of data. And I think there's a lot of potential to use that in the right way to be able to serve our consumers better. speaker-0: We talked about digital twin and validation overall, but considering brands, they already have that kind of a data. and they're companies who are creating digital twins. They collect data for over a period of time, but not in abundance of the data that they have, the kind of data which brands are having. Should brand declare those data points relevant to run a digital twin to a digital ⁓ twin agency or not? Or if yes, then what could the precautions they can actually take with them? speaker-1: Well of course, if you're going ahead with any form of data sharing, you need to have, of course, the right guardrails in place, you know, insofar as all the compliance aspects are concerned, you know, security aspects are concerned, privacy aspects are concerned, and everything else. You know, those can be done today. You can create the right kind of walls and the right kind of ways in which you can do that. And I think it's a good opportunity for brands to also partner and see, you know, how you could share that data and use it. In some cases, I also hear you know many, many different companies and brands also wanting to build their own, you know, systems and build their own. So it's a typical, you know, buy or build or partner question. And there are pros and cons of all these three situations, you know, when you want to buy something or you want to build something or you want to partner with someone. But whichever way you work with, you know, depending on your organization, depending on the category and the sector and the business. And the different guardrails around that and and some of the regulation aspects also, it's important to start looking at data together and to be able to connect different thoughts. That's where the power of the entire data is. And then if you have different agents, you know, sort of hovering around different pieces of data to be able to answer specific questions, and you could have a galaxy of agents trying to answer different questions and then join them together as well. So so that's the power of the new world of AI that it's that it's giving us and I think we need to take advantage of it and see what we could do with it. Of course, fully understanding and fully knowing where does that data come from, what is the process that's being used and what is the result or the output that is coming through, because only then would we develop the conviction to be able to draw some actions and then take it to the market. speaker-0: I'll take you ten to fifteen years, you know, before ⁓ anything actually happened. And back those days you you worked on multitudes of the data. When you were connecting dots to come up with that kind of a conviction, I I just wanted to be a little n nostalgic in trying to understand how this used to be be worked upon back those days when these technologies was not there. And do you really have to do multiple surveys? Do you really wanted to see, you know, that if I'm picking up a certain objective? What other category of data should we look at it? Now, as of as on today, if we compare data is coming from almost everywhere. We c we easily know that, you know, this data is not relevant, this data is relevant. Back those days, you we don't have that kind of ⁓ you know data ⁓ capturing. We really have to capture the data by doing surveys or by other means as well. So back those days, how you ⁓ how much time it it's by you guys were taken and you know how much ⁓ in how much time you were able to draw insights. when you have to s look at data at a very holistic view, but also at a very granular view. speaker-1: Well, I think drawing insights takes time. It takes time because you need to really understand all the data and you really need to connect all the different pieces. When there was primarily, you know, measurement data and primarily survey data. Of course, social media listening has also been there for a very long period of time, or data from search or other places have also been there for quite a period of time. One would have to still look at it in pockets and buckets and then you know, see how one could make sense of all the data and the different behaviors. But if you look at the world of surveys, you know, surveys are primarily question-based. Yeah, and and of course there are other ways of getting data like ethnographic work or qualitative work or observational work or social media listening and so on and so forth. And this has evolved over over the decades. You know, what more data sources have come in. Now in today's world, I think there are two things that have come in. Now Number one is that the data sources are multiplying and it'll continue to multiply. You know, if you look at the next few years, of course, there'll be more and more data, you know, which will be accessible, and one needs to figure out what data will help me answer what kind of a question to be able to take what kind of a decision. Yeah. So that will still very much remain. And the second part of it is that you have AI as a tool to be able to make sense of so much of data. If you train it the right way, if you have the robustness of thinking. if you have you know different ⁓ elements built in into the AI model, then you could make sense of it. Now, you know, I often hear a question that, you know, whatever could be done by by surveys can be done by AI. And I feel that, you know, many things that were done by surveys can be done by AI, but some things cannot be done by AI. For example, you know, this kind of a discussion, you cannot do it using AI. You know, we will not have It'll be a very mechanical and a very transactional discussion. You want to have a deeper discussion on something, you know, you want to really understand what's going on. speaker-0: done that, but it was more scripted based. So it is being fed with the data. It is not an open kind of ⁓ conversation that we of course, you know. So AI is not everywhere there. But considering that a ⁓ the reason why I asked you the earlier question, you know, it insights take its own sweet time. I it's not that, you know, you plug the data. But a lot of people actually claim that with AI you can get insights within two to three days. speaker-1: Well, AI has definitely shortened the speed. But I and I think that's another big advantage of AI. But what one must be very, very clear about is that we are not just looking at a shortcut with AI. We are looking at a better decision-making mechanism with the help of AI. By ensuring that we have more robust data to start with, we have a very strong model or models to look at, and different agents, you know, drawing up different parts of it. And then we are able to look at Different variables and different decision-making vectors, which will be really useful to answer a certain business question, and then to be able to also take it to action. So it's not just a shortcut on speed that we are looking at, you know, AI to just reduce time. But I think AI should be able to help, first of all, digest so much data, you know, and then, like you asked about big data, you know, AI today can digest a lot of data. In a in a short period of time, in an efficient way, in a precise way, in a productive way, and so on and so forth. But you still need a human layer on top of that, because you need to understand human beings, you know, which is at the heart of heart of insights, and then you need to see how you you take it. speaker-0: Yes, sir. Tell me something since ⁓ this these LLMs are being made by humans themselves. So when you say we still need a human layering on top of it, there will be a times when everything has been, you know, pretty much robotic. We we have we would have fed the data so well that you don't even need a human layer. Will market research b will die those days because every now and then, you know, you would run it out and you know get the answers from AI itself? Will that ever speaker-1: That's a that's a scary part of AI, that AI, you know, as a tool developed by humans, starts taking over human beings in many different ways. And, you know, if you read around what's happening and so on, you know, that that's that's one dimension. That's definitely, you know, progressing very fast, and that's a very scary dimension. ⁓ so I don't know where that will go, but that's definitely, you know, one dimension, which everybody is talking about, and people are kind of, you know, thinking about that. you should not allow AI to get ahead of you. Yeah, you need to really be able to use AI, you know, because finally we are here to serve people, to serve societies, to serve communities, to serve, you know, ⁓ organizations, institutions, and everything else. So in that context, we, you know, my own view is that we should not allow AI to get ahead of us. But we should use AI in the right way because with so much of information, data, assimilation, challenges, et cetera, AI can definitely be a big aid to us, but we should look at it as an aid. But yes, AI can also start, you know, ⁓ looking at different things. But probably in terms of getting the context, the business question, the real strategy behind the vision part of it, you know. to those areas are still, you know, predominantly human. ⁓ AI could potentially, you know, maybe have something around a vision, but that vision needs to be textured in real business context and in real human context. And that still can you know is is quintessentially human. speaker-0: I still believe that, you know, ⁓ we are very dynamic in nature. I mean, since we have already built AI, we can build probably AI plus something else. I mean, from big data to social media listening and whatever or not technologies have come on. Anything which comes as a you know, cons consumable brand or consumable product to the human being, they always evolve through that. They know exactly whether they're liking it or not. there would be a period of time when, you know, they would start putting this onto the social media. And then we know that there's always a demerit because there are so many other brands who actually hire. I mean, just like online ⁓ data collection used to be criticized because there are online survey takers, you know, speeders. ⁓ I believe that there are actually not natural feedback givers. There are like robotic feedback givers. These data points are not validated. I mean ⁓ technically if I'm using this feedback, I'm actually spoiling the entire algorithm because then AI would give me a different insights altogether. That's where I believe that human intervention is definitely needed. And that's not going to go anywhere. But considering that ⁓ AI is becoming a better pattern ⁓ recognition these days, what becomes the unique role for human insights to evolve around? speaker-1: Well, to break the pattern. I mean AI can only learn from the pattern and can give you, you know, what comes next. But a lot of marketing, a lot of, you know, behavioral understanding, a lot of human insight, under st you know, and and all the different dimensions are about changing behavior, are about, you know, breaking a pattern that you have and maybe taking a consumer from one place to another. You look at, you know, let's say let's say health and wellness, you know, if you look at health and wellness, it's a discontinuous pattern breaking that is coming up with many different people. People are getting more conscious about health, people are taking more steps about health, people are connecting different elements about health, you know, on what they eat, how they exercise, how they have their lifestyle, you know, how do they use certain gadgets and data and other kind of, you know, things which are always available to them to understand their health metrics and so on and so forth. So all this is actually connecting different elements and about you know a new behavior about how a person looks at his or her health and what's what's the way in which he or she is managing or thinking about health. Now, AI can only take, you know, certain things and create the next point based on what it has learned. But if you are to break something, you need human thinking. You know, for example, if you look at the energy sector, you know, AI can start Telling you a lot about the energy sector. But if you were to think about okay, where do I go with green energy and where do I go with the next level of energy, then that's something that human vision has to still play a very big role to see you know where does the energy world go and what's the sector and what are the dimensions of it. You have any big ⁓ crisis or any big shift or or any big you know event that happens. AI may not have learning from patterns of exactly that event, you know, from the past. Like, for example, you know, COVID is something that AI at that time, you know, if you had had the same level of evolution of AI that we are seeing today, even at that time, because COVID was new, you know, AI wouldn't have been able to assimilate what was really happening at that point in time because there were no previous, you know, data or learnings or other things built in. So therefore, those kind of things have to be still. You know, human first in its approach, but AI can help you in a very, very strong way, of course. And that is where I think people should embrace AI, but of course, you know, keep human oversight and human guidance on top of that. speaker-0: So what would be an idle ⁓ insights operating system look like if we we say ⁓ and specifically for the large enterprises from from five years from now? Because when you know I think you gave a very good example about COVID. We didn't we never had this ⁓ inclination towards what is coming, how people can overcome that. Now we have a lot of data which ⁓ AI is telling us about COVID. Again, it's a past data. I think when I started using ⁓ a ⁓ Chat GPT was twenty twenty three. I think that's when, you know, Chad Chat GPT started ⁓ amongst us. So ⁓ what exactly that insight operating system should look like af from five years from now? I mean, considering all these things. speaker-1: Well, I think if you were to build an insights operating system, you should first build your business operating system and your decision operating system. You know, what's the decision architecture in so far as let's say growth is concerned or investments are concerned, or whatever else are your different dimensions coming on from the strategy of the business. And then based on that, you would have to think through okay, if we were to input into this and this is the decision. or decisions we need to take, then what is the data and what's the insight that we need underneath? It's a very different and how can we connect? Because many things are all connected today. You know, it's not that these are all in in in silos where where certain raw materials come from, how they are used, how they're manufactured, how things are produced, and therefore what the consumer gets at a particular, you know With a particular product, pricing, and everything else, you know, all around the piece and so on and so forth, the brand, everything depends on the entire chain. So you will have to rewrite the entire value chain of insights to essentially understand, you know, what is it that is the growth operating system and what's the decision operating system. Once you have these two layers, then you could think of what is the data and the insights that you need. And a lot of it will probably still be, you know, ⁓ continuous. There could be some which could be periodic as well, depending on your decision-making ability. You know, if you are only going to take a decision once in a month or once in two months on something, you know, then you need not have that data on a daily basis if you're not being able to influence that decision. Whereas if you're able to really influence that decision in a much more frequent basis. Like for example, you look at e-commerce, you know, they could be different day parts and different, you know, data that moves in a particular day itself in a different way. You know, and there you need operating systems to be able to help take those decisions on availability, on visibility, on on other aspects ⁓ at that point in time. So therefore it all depends on your growth operating system and your decision operating system, and therefore your marketing intelligence operating system we should support. speaker-0: So considering that, you know, there's a new brand which is coming up, and ca now I'll put probably ask this from a context. If larger brands are actually making these in insights operating systems, and down five years they will be ready for whatever comes to them, there would always be a be a scope for smaller brands to come up or new ⁓ generation brands to come up. Should they also have or probably those sh they should also prepare themselves to have an insights operating system from now onwards? And to give you a more context, ⁓ probably you can explain this by an example. That there's a new company which is coming out and ⁓ they're kind of establishing a new coffee flavor into the market. And that's the for the first time they're doing it. What is your guidance or advice to them considering how they can build upon their insights operating system and what they should be aware of? speaker-1: Well, if any company is looking at a new category or, you know, entering a new speaker-0: To to to add, they're fairly new. They have not started or had any data background b behind them. And they're they're probably like searching on AI. That I am kind of starting this. So what are those advices that you would give them to start with? speaker-1: Well, I think you know to start with, fundamentals are fundamental. So you of course need to understand in any category, you know, whichever category it be, you need to understand the market, you need to understand the consumer, you need to understand the channel landscape, you need to understand the engagement landscape. So all this is very, very fundamental. You want you need to understand the value chain and you know how it really works. So some of it, of course, AI and data available will help you. there, but you will have to do, you know, different elements to get together, different pieces to get together to understand where it is. And then, you know, what should you be launching which is ⁓ you know attractive and differentiated and you have the right to win in that particular case. So so therefore, you know, how how will you really carve out a position for yourself in the market? You know, what would the brand be? What would the different dimensions of your entire marketing piece be and so on and so forth? You know, your entire route to market, go to market, etc. So all this, you know, is not a ready-made solution by pressing one AI button that you will get all of it. Yeah, you might get some different pieces, you know, from different sources. You can, of course, use some AI tools to assimilate some data. You can understand what's going on in in some of the social media feeds and all of that, but they will be all small pieces of the entire puzzle. The total puzzle at this stage, and probably will continue for some time, will have to be still stitched together by humans. But yeah, I can accelerate individual steps of this process. And then you need to bring in, you know, what kind of data and insight would you really need, you know, once you develop. your entire proposition and then when you're launching it, what do you need to really track? Your entire proposition in the market and how it is moving and and so on and so forth, what are the key success factors and how you're tracking and and getting there. So EI can of course help in every step, but EI will be one of the tools, you know, which you have to still, you know, put together and use your overall human speaker-0: But Insights operating system or IOS I would probably call it as ⁓ is basically a entire road work plan or road plan, ⁓ sorry, road map plan where they would precisely have the business accelerators in the on the line where they have to change gears. They can also mention that. And based on these data they can still read where they can head to next five years. Is that what I understood is the correct? speaker-1: Well, yeah, so there'll be different elements of the system. You know, there would be different pieces of data that you would still get. A lot of the data will still be in the offline world, especially in a country like India, if you look at the overall market. But if you were only to operate in the online world, then of course, you know, you can precisely focus over there and still need some pieces of data to come in. So operating system should be one where you're continuously getting these different signals and data. And you're able to really assimilate them quickly, and you're able to, you know, take use that to take decisions. If you want to like, you know, predict what will be the response to, let's say, A concept, B concept, C concept, maybe you know, with some of the data already stored with the help of AI, it could help you predict if you have good data getting stored over there to be able to predict, you know, what could be the likely response. You probably need to still validate from a human perspective. because it will give you a lot of directional, you know, ammunition. ⁓ and then you can accelerate the entire mechanism by using, you know, strong human judgment as well as as well as AI. speaker-0: So, ⁓ we s we spoke about how a new age brand can step into the market and how what precautions or what plan do need they need to, you know, emphasize on when they're building an iOS. What about researchers? How can ⁓ they can develop their own capabilities ⁓ in such a way that they remain researchers, say in in next six years or five years? speaker-1: Well, I think as a researcher, like everybody else, you know, I always keep saying that, you know, of course you need to embrace AI. Sometimes, you know, when I talk to certain people and they ask me a question, I in fact ⁓ give them an answer that this question can be answered by AI. You know, why are you asking me that question? ⁓ because if you want a certain view or a judgment or a certain, you know, point of view or or or connected dots around different elements, then yes, you should ask me. And I could, you know, based on my experience and and whatever is my lived experience, I could talk to you about that. if I know what it is about. But if it is about, you know, what's the formula for something XYZ, then of course the answer lies in AI. I mean, you just need to put it in in a system and it'll give you what the formula is for something, yeah. So therefore make use of AI. Yeah. Wherever you know there is value in AI. Make use of AI and you know it'll be able to do things maybe you know X times better or even faster. But AI will not, you know, you still need human insight and you still need human understanding on top of that. Because you know, AI will not get into the details of human texture, which you still need to understand. AI may not look at it, you know, that holistically like you would like it to think about. You still need to understand that. So it's a very useful tool, it's a very good productivity tool. It can help your thinking, but don't outsource your thinking to AI. I mean, you should think still be thinking on your own feet and ⁓ help use AI as a help or an aid to have a better output. And that's what researchers should be doing. And why researchers, you know, people from all walks of life should be should be doing. speaker-0: I mean I think I remember one of the conversation and this is specifically to the digital twin ⁓ which which I am re-emphasizing on. Since you mentioned, you know, researchers need to ⁓ pick up the right way of using the AI. But there's there's already an ongoing debate that digital twin can be seen as a more you know, if the if a organization is using it, it can be seen more of a ⁓ consumer-centric insights that they generally get from digital twin. So ⁓ is it Shall we call it more of a trust that those organizations are embracing it, ⁓ into these digital twins and or is it more of a you know, a false alarm to a certain certainty, to do we call it that way? Well speaker-1: It depends a lot on how you're creating these digital twins. You know, what data goes in, like I said before, how you're really putting together different stuff, what holistic view or full brain view or whole view do you have about the about the person? And then more importantly, how they decide and what behaviors and other elements of data you're feeding in. Garbage in will be garbage out, yeah, like always. So therefore I think it's important to have a very strong you know, input to creation of these, you know, digital twins. Now once you have that, that is then something which is which which if you have a robust, you know, set of different digital twins and so on, it could be an enterprise grade, you know, kind of ⁓ consumer, you know, digital twin which is there across the organization to help you take decisions across. And with lot of new data coming in all the time with the dynamic changes that are happening, it could also be always self-learning and understanding and all of that. But pressure test the digital twin before embracing it. And once you have embraced it and it it sounds meaningful and so on, have a continuous way by which you are always validating and understanding whether this twin still is reflective of what you were talking about or not, or what you had had in mind or not. But at the same time, you know, also see what are the kind of results that are coming in out of these lab, you know, based digital twins, let us say, for example, but always, you know, kind of validate them with what I call the validation with the consumer in the wild. And and that needs to be there. So so all the time it's it's a continuous learning system. One is feeding off the other, the other is validating, challenging. thinking about what next, you know, if it if it if it's not moving in the right direction, maybe there are different elements that needs to be brought in. Maybe a certain dimension is probably missing in it, which needs to which needs to come in as well to understand, you know, where it's going and so on. But once you have that, then it could be an enterprise grade, you know, I wouldn't say assistant, but you know, enterprise grade living digital twin, which is there through the organization. And you can use it in different ways depending on what questions you have. speaker-0: So are you drawing a line to a ⁓ to a digital domin creation that after a certain point of time, you know, after all doing all those pressure testing run, now we know that this validation is accurate enough to what we would have gotten from ⁓ doing surveys or collecting data from other sources. Are you drawing a line or is it it is does it mean there's there's no line? You know, digital domain needs to be tested every now and then, even if we are hitting that validation mark. speaker-1: Well, I think you need to continuously keep on pressure testing it. Because the world is dynamic and things are changing. If you look at simple behavior of of people, you know, insofar as online shopping is concerned, we have seen the acceleration in quick commerce. And it's accelerating at a pretty fast pace. You know, the kind of number of times people are ordering, even it even within a day, is going up. So that's a new behavior that you're seeing of people. Now that data. You know, if you are able to infuse into your digital twin, if that's meaningful to what you're looking at, you know, if you are a CPG player or whatever in that particular category or market, then that dimension needs to continuously keep coming in. So therefore, a digital twin is a liv living and breathing system. Yeah, you need to always infuse it with data. You need to periodically, of course, you know, also assess and validate certain final steps. If if it's an irreversible decision, I would still think that you need to do some human, you know, final decision making based on that. And probably you need to do some strong validation in the more classical sense because it's a big decision, irreversible decision, and so on and so forth. But for a large number of continuous decision making, because the time from insight to action is also collapsing. And organizations need to gear up. to be able to get the insight and then be able to kind of, you know, also take decisions. speaker-0: So this is one last ⁓ basic fundamental understanding I need from you. We tested a digital twin, we got the insights, and then we validated it through, say, qualitative or quantitative approach. We knew that you know doing a qualitative or quantitative takes its own sweet time. We know that you know digital twin has given us that bandwidth of lesser time frame of say one to two days, and then you know spending again ⁓ three months in validating that, isn't that too much of a time for any researchers to to give away a brand insight that brand needed in lesser than three months? speaker-1: Well yes, so digital twin, of course, will be able to collapse your time. If you have got your tin t twin properly, if you've got all the data in there and if you've validated a few results, you need not validate it every time because I don't think that's even practical to be able to do, you know, something on a predictive decision intelligent system and then do something, you know, like the offline or the online researches that you're talking about and validate every one. But I think if you have a robust system You can take the decisions based on what your predictive decision intelligence tool is telling you. But very importantly, you know, all the data that is getting fed in also needs to come in from some of the other consumer work that you will be doing continuously. Otherwise, you know, your digital twin, if you don't feed in with anything, it's like, you know, we eating all the time. If we are not fed with food, you know, on a daily basis, then, you know, we will sort of you know not be able to. speaker-0: That would actually minimize the time over a period of time. I mean to start with it may take it. speaker-1: Takes some time to get there, you need to of course, you know, test and validate it, but once you have a good living breathing system by your side, then make full use of it and periodically of course you need to also see how it evolves. speaker-0: No, this this this definitely makes sense. And I think that would give a lot of ⁓ viewpoint to a lot of researchers as to how they can actually envis and envision the use of AI in the you know in the forthcoming years. But ⁓ before we end this conversation, there are a couple of rapid fire rounds that I I generally do with my guest and it's more of summarizing the entire ⁓ the podcast that we did. So to start with one word that describes the future of Insights. speaker-1: Anticipation. speaker-0: Consumer surveys or passive behavioral data. speaker-1: passive behavioral data, but also consumer insights surveys. speaker-0: Prediction or understanding. speaker-1: I would say between the two prediction, but if you don't have a strong understanding you cannot get to prediction. speaker-0: AI co pilot or AI decision maker? speaker-1: AI co pilot always. speaker-0: One matric marketers paid too much attention to. speaker-1: I think you know marketers pay a lot of attention to reach and maybe distribution. I think these are important. Like for example, you know, if you look at reach, reach is important, but it's also equally important who are you reaching and what happens after reaching. What is the outcome of the reach? speaker-0: One matric marketers don't pay enough attention to. speaker-1: Attention. speaker-0: The the biggest myth about consumer behavior. speaker-1: The difference you need to recognise the difference between the behavior and the attitude. speaker-0: Okay. A little more explanation. speaker-1: There's always a difference between what people say and what people do. Correct. Yeah. And that's what you need to understand. speaker-0: Okay. Got it. One book every inside professional should read. speaker-1: Thinking Fast and Slow by Kanuman. I think it gives you a lot of insight on behavior and ⁓ and you know, what is classically called system one, system two, and you need to understand the conscious and the subconscious. speaker-0: One skilled researcher should learn in the next twelve months. speaker-1: Well, a skill that researchers should always learn is why, getting to the why. But I think in the next twelve months, more precisely, getting more ahead on AI. speaker-0: You have to complete the sentence. The companies that will win in twenty thirty are the ones that speaker-1: that will embrace human intelligence and artificial intelligence together. speaker-0: That's great. So eventually we spoke about, you know, how much embraces ⁓ we need to embrace on AI and the usage of AI. And I'm glad that Chandan I and you, you know, you mentioned ⁓ we actually have a very deeper discussion on you know how AI usage is is there and ⁓ and and a lot deeper into understanding where we are actually heading to. So there are a lot of things that you would probably, you know. get take ⁓ key takeaways from this ⁓ podcast. So thank you so much and then for taking time. speaker-1: Well, thank you, Ankesh. It was a pleasure talking to you. speaker-0: So, ladies and gentlemen, Chandan today shared his perspective on the future of insights, intelligence, and decision making. What stood out for me was ⁓ the conversation is around ⁓ what kind of a technology we are actually continuing to evolve at such an incredible pace, and that the goal remains still unchanged. Understanding people is much better needed, whether it's through research, behavioral data, or AI models or even digital to win. The real value only comes from connecting information to the human understanding and the meaningful business action. As organizations move towards always on insights ecosystem, the challenge will not be about collecting more data, it will always be asking better questions, interpreting signals wisely, and balancing artificial intelligence with the human judgment. To everyone listening or watching, I hope this conversation has encouraged you to think differently about the future of the consumer understanding. And the role insights can play in shaping smarter decisions. If you're enjoying this episode, please subscribe to the channel, follow us on the on your favorite podcast platform, and share this conversation with someone who's passionate about insights, marketing, innovation, and the future of the business. Until next time, stay curious, stay collaborative, and keep building what next. Thank you.