<img height="1" width="1" src="https://www.facebook.com/tr?id=901494157838465&amp;ev=PageView%20&amp;noscript=1">
Skip to content

Combining Human Expertise and Artificial Intelligence

Stravito Apr 14, 2023

In this episode of the Consumer Insights Podcast, we speak with Bhaskar Roy, Client Partner - APAC, at Fractal

The interfaces between insights teams and businesses are ever-evolving. But what role should artificial intelligence play in those dynamics? As an insights professional, you need to continue on a path of learning, relearning, and unlearning in order to stay relevant and to stay ahead.

In this episode of The Consumer Insights Podcast, Thor is joined by Bhaskar Roy, Client Partner, APAC at Fractal.

They cover:

  • Why insights are crucial for both customer understanding and business efficiency

  • How to take a predictive approach to solving business problems

  • The different elements of getting stakeholder buy-in

  • Why centralizing data is essential

  • The role of insights and analytics in ESG intiatives

  • The do’s and don’ts of incorporating AI into your insights toolbox 

  • Why human expertise is still necessary

  • The skills that make a successful insights team

  • How to elevate insights within the organization

  • How to leverage automation to move ahead in your career

  • The role of design thinking in working towards human centricity

  • Why & how to break out of organizational silos 

If you’re looking to learn more about how AI can help transform your organization and your career, tune in to this episode of The Consumer Insights Podcast.

You can access all episodes of the Consumer Insights Podcast on AppleSpotifyGoogle, or use the RSS feed with your favorite player. Below, you'll find a lightly edited transcript of this episode.

Thor Olof Philogène:

Hello everyone, and welcome to The Consumer Insights podcast. Today I'm excited to have an impressive data and insights leader joining me. What I know will be an amazing conversation. 

I'm thrilled to introduce today's guest, Bhaskar Roy, Head of ESG Data and Analytics at Fractal, a prominent player in the artificial intelligence space that offers AI engineering and design services to help Fortune 500 companies, Bhaskar has over 15 years of analytics experience, having previously worked at companies like Target and Accenture. Thank you so much for joining me.


Bhaskar Roy - 00:01:07:

Thank you. Thank you for having me Thor, really excited to have a conversation and share insights and stories with the listeners.


Thor - 00:01:14:

Well, let's jump right into it. Could you maybe, to get started, tell us a couple of minutes about yourself, your journey, and how you came to work in the role you're in today? How did it start?


Bhaskar - 00:01:26:

Sure. So I'll take you back in time, right? When I was back at college, university, as people like to call it in different parts of the world, I was always inclined towards quantitative subjects. But then I wanted to combine that always with the real-life context. And that is what led me to do a combination of things.

I did my undergrad and thorough studies in statistics to build the right background and followed that up with a degree in economics to build that context. And that, I think, set me up the right way to get started in the world of what was still at that point of time, the nascent world of analytics, AI, and Data. 

Started off with a company called GENPACT, which was incubated by GE as their in-house center initially, and then over time spun off. But then that's where my journey started, and it has taken me to different places, different kinds of industries, and I'm extremely happy on the journey that I have taken and the learnings that I have had along the way. 

I’ve worked with a number of large organizations as well as small ones throughout the career, but extremely passionate about it and love what I do on a daily basis.


Defining insights and their importance

Thor - 00:02:33:

And having had that experience, how do you define an insight?


Bhaskar - 00:02:39:

Insight is something - insight, to me, Thor,  is something that should help drive a business decision forward. Otherwise, it's just a piece of information that is good to know and that's about it. It should help you see things in a different light and make a decision that moves forward.


Thor - 00:02:57:

I love it. Succinct, brief, clear. And why is it that they're so important? You said it's about helping the business forward. But tell us more. What do they allow businesses to do?


Bhaskar - 00:03:11:

As you think about businesses, both the traditional ones as well as the modern ones that are coming out in the areas of tech and finance. Insight is a key part of what each business needs to do. It helps them not just understand their customers and consumers better, but it also helps you understand how to keep them engaged, how to ensure you are optimizing the experience that those customers and consumers have to not only drive your top line as an organization, but then over time, improve efficiency in different processes that you are doing as well. And that is what drives business impact eventually. It's a combination of what you do for your customers to drive both top line and bottom line.


Using insights to fuel efficiency

Thor - 00:03:56:

And let's take some time and maybe learn from all your experiences. And could you tell us about a time in your career when you've uncovered such an insight and maybe share the journey with us on how that fueled innovation and allowed you not only to get those top line benefits, but also the efficiency you just mentioned?


Bhaskar - 00:04:15:

Absolutely. Let me take up an example from a time I was working in retail, right? And as you think about retail there is a large proportion of revenue that gets lost for retailers in what is known as shrinkage loss,  which essentially is either people stealing from you or things going missing or misplaced in the store. And that has a lot of process implications. 

Depending on what retailer you're talking about it could range anywhere from about 1% to 2% of your revenues and if you think about the larger ones that are out there, the Walmarts, the Targets of the world it is a significant amount of money.

Now how do I think about a way to start breaking that big problem down because 1 or 2% of revenue is huge. How do I break that down into manageable components? That was my first part. And it literally started with multiple levels of engagements and initiatives that we did. 

First, segmenting for example your stores in different ways. One of course is sales  that you're looking at, but then from a process perspective as well– how do shipments differ, how do the formats change depending on where the store is and how you're shipping to them, what is happening within those stores? et cetera. 

And then finally all of that led to not just the segment, but eventually looking at predictive elements that can be built into what you're trying to do. For example, a large part of that loss is literally, literally in the US is organized retail crime, which on the face of it is very difficult to predict, but if you can get ahead of the game you can have a large impact there. 

So we actually built out predictive models that let us have an impact significantly reduced, but we had an impact of about 0.2 to 0.3% on shrinkage by these different efforts that we did, all the way from understanding segmentation which is as is, to getting predictive and looking at what might happen in the future as well. So a combination of those that led to that impact and drive efficiencies overall.


Solving problems of the past and predicting the future

Thor - 00:06:17:

And I love what you're just describing, and I also love the fact that you were able not only to solve the problem, but you were also able to, as you said, get predictive and starting to look ahead, going from looking in the rear view mirror to looking in the windshield, right?

Tell us a bit about that. What was the journey? To go from understanding how to solve the problems of the past to actually having a view on the future?


Bhaskar - 00:06:42:

Sure. So in fact, one part that a lot of people don't realize is the insight. Whether it is as is or predictive in nature is one part of the journey. The other part of the journey that we need to also cover is engaging with stakeholders to actually implement those insights and solutions on the ground. 

I think that to me was the biggest learning as we kind of went along that journey. It took us about a couple of years to undertake that journey in itself. But then it had different elements to it. One, of course, was as we are looking in the rear view, as you rightly said, how do I get buy in from my business and have that alignment that this is the right way in which they want to look at their stores or any other way in which in which they are trying to try and run their business? That was the first part. 

So that the business will start trusting the data that is coming through. Once you have that right, then you can build on top of that number of different solutions, be it reporting in nature or predictive or even trying to optimize how your operations are being run. 

All of those elements become much easier once that reliance on data is built-in, the trust on the data is built-in with the business.


Mistakes when integrating insights

Thor - 00:07:55:

And let's spend some more time on that because I absolutely love what you just highlighted there and the reality of how the integration part becomes so crucial. So let's spend some time there. 

What is it that people get wrong? What is the advice you would give to yourself if you'd go back in time in order to make that happen, in order to integrate the insights into the business fully?


Bhaskar - 00:08:20:

Sure. In fact, I’ll bring in maybe another example as I'm talking through that. The key learning for me has been and if I can go back in time and change that is understanding the value of data, and the fact that data unfortunately will never be in the same format across the globe. 

You will always have different people storing and sharing data in different formats, although the worlds of APIs makes it a bit easier. But then you will always have data that is structured, that is unstructured, data that you're getting from external parties and suppliers which all needs to be integrated to create that single source that everybody looks for. I think that is the crucial part. 

We tend to miss that as organizations, and I'm realizing that more and more as I'm getting deeper into the world of driving benefits for ESG for organizations, right? And that is where I see that for example, depending on which organization you talk to today in the Fortune 100, Fortune 500, they all take anywhere from three to nine months just to get their reporting numbers ready. Basis SEC Guidelines and the European Commission on how sustainability and different measures within that is impacting their business. 

Now that in today's world should not be acceptable. And therefore the first big learning for me is get that data together, get it right, and get it in a format on which you can start building on top of after that.


The role of insights in sustainability

Thor - 00:09:48:

Let's spend some more time on ESG data and analytics where you actually currently specialize. Could you tell us a bit more about the role that insights play in sustainability?


Bhaskar - 00:10:00:

Absolutely, Thor, and it's an interesting thing and I started realizing that a couple of years ago as I got more interested in how sustainability is being looked at by different organizations. 

The first realization there for me was most organizations, even though they are looking to make that journey and start putting numbers out either as commitments that they are making to the external world on how much, for example, they will reduce their carbon emissions by or anything else around it. 

The key part there is having visibility on dat, and data for these organizations meant data that is structured internally, data that is lying literally as flat files or PDFs in terms of contracts that they have with their suppliers or even external data that they are getting in from their suppliers on a regular basis. 

And as you think about the broader Fortune 100, Fortune 500 world, you of course, have a lot of financial institutions, but you have a large set of manufacturers who are hugely impacted by what's happening within ESG, and the declarations that they need to make. Because as you think about production, whether it be textiles and the link to Southeast Asia, where a lot of different human indicators are not the right way, or sourcing from Africa in terms of raw goods that are being used, all of that needs to come together and there needs to be a platform to kind of enable that.


How to incorporate AI into your insights toolbox

Thor - 00:11:25:

And you've also shared a lot about the benefits of AI, and particularly with regard to innovation. What are some of the ways that insights professionals can incorporate more AI into their toolboxes to yield those benefits?


Bhaskar - 00:11:40:

Sure. So as I think about it, right, AI in itself is probably not enough. It is a good tool to have in the toolbox, yes, but it needs to be deployed the right way to have the impact that we are looking for, right? 

So for example, most work that you do today actually can get solved by just looking at the data, making it more clear, more transparent. But then as you start being more predictive and start bringing in optimization elements, that is where AI plays a significant part today as we think about it, right? 

And that is where, for example, if I build a little further on the ESG example that I was talking about, once you've helped an organization create that data platform, what you can then help them do is not just have visibility on where their KPIs today are, but you start going a little deeper. You help them understand how for each KPI there are different drivers that are impacting that KPI itself. 

And then it becomes an optimization problem after that on how do you interplay those drivers with your objectives on revenue and profitability also coming in and then make the right decision so that you are of course continuing to grow as an organization, but also staying committed to things that you're looking at on ESG and other elements.


Where human expertise remains paramount

Thor- 00:13:01:

And if we shift perspective, do you see any use cases where AI won't be applicable for insights professionals? I mean, you talked about the optimization role that it often serves, but in other words, in  what areas do you see human expertise being paramount?


Bhaskar - 00:13:18:

As I think about it, right, there are some areas in which AI is almost like a no-go, given the things that are at stake or the broader governmental/regulatory environment at play. 

So for example, Credit Models is a place you can deploy AI, but you are very limited by the choice of techniques or algorithms that you can use because you don't want it to be a black box. So that's one place where AI's impact is limited, I would say. 

The other place, I would say, where AI also gets limited a bit is places where you don't have too much data that's available, although those kind of spaces or white spaces are reducing over time. But that is where it is not the right place to kind of deploy AI. 

And finally, if I may add it, you asked about the human element. Most decisions that are today being made by humans can always do with some augmentation by AI. It needs to be an AI plus human decision. And that in most cases is the most effective. As you move forward, a machine alone or a human alone will not be as effective.


The composition of successful insights teams

Thor - 00:14:33:

I absolutely love the fact that you highlight that crucial role that we still play as human beings in order to actually make AI even stronger, actually becoming the augmenting layer as you just suggested. 

If we switch gears a bit and talk a bit about talent and you've of course seen the industry, the insights industry change. And as you see that change, how do you believe that we need to think differently about the composition of successful insights teams?


Bhaskar - 00:15:04:

As I think back, right, I think the common part or the common ingredient of any successful Insights professional always has been a combination of, of course, very, very top notch technical skills combined with the knowledge of the domain or the context in which they are operating. 

So, for example, if I'm working for a retail client, or I'm working in a retail organization, I need to understand thoroughly how different moving parts within that organization or enterprise work. Because that context needs to be kept in mind as I create any solution or as I engage with stakeholders for deploying that solution eventually and having them use that on a real-time basis. So that's second, I would say. 

And third, maybe I would also add, is the fact that continuing to learn continuously. I think that is a key differentiator as well because the industry that we are in, the kind of tools, techniques that are available to us, the way the interface between business and insights teams happens today is also ever-evolving. So we need to continue down that path of learning, relearning, unlearning as needed to stay relevant and stay on top of the game that we are playing.


Essential skills for elevating insights

Thor - 00:16:24:

And when you think about skills earlier in this conversation we talked about the importance of the integration parts, right, of Insights and to actually drive change. And if we talk about the elevation part, meaning that how you elevate insights so it actually plays the role it needs to play within the organization. What skills do you believe as essential to make that possible?


Bhaskar - 00:16:49:

It's a combination of skills, I would say, Thor. And that's something that I have seen changing in the last about five years earlier. It always used to be about getting the data in a sample file together, be it in Excel or a flat file. You run a model on top of it, you create the model, you share the insights back on a deck, but no longer is that enough. 

There are a multitude of other things that are important as well. First and foremost is skills on basic data engineering and managing. You should be comfortable with bringing that data together, having an understanding of it before you start doing any sort of insights or data science work on top of it. So that's one. 

Second is with the multitude of models and insights that are running around in organizations today. It is also important to start looking at how those get operationalized and how do you understand and manage them most effectively. And that is where the field of MLOps is significantly coming up these days because you need to understand what's happening within the enterprise. 

And then third on top of that, is just building a solution or creating an insight is not enough. And I referred to this earlier as well. How do you ensure adoption of that solution? How do you ensure buying from that end stakeholder so that adoption piece, the new skill set that needs to be specialized in as well. You cannot just leave it at that deck that you used to push out of the door earlier. So you have to combine engineering MLOps with the adoption to have that true impact.


The best career advice Bhaskar has ever received 

Thor - 00:18:24:

That’s such good advice. And if we stay on the topic of advice, again, going back in time, what's the best career advice you've ever received?


Bhaskar - 00:18:34:

It kind of goes back to my first job, right? As I think about it, the best advice was from my first or the second manager that I had in the organization, and it links back to the whole learning aspect that I mentioned earlier as well. He basically said, “For you to continue growing and continue learning in your career, it's important that you make yourself redundant to your current role.” And as I broke that down, in my mind, that essentially meant a couple of things. 

One, first and foremost, you should not need to be actively doing your current job. That's when you'll have time to learn. Now, to be able to get to that stage, you need to automate what you can in your current job and get it out of the way. 

And from that, comes in opportunities of then realizing where there are synergies across different kinds of work that you might be doing and then explore that to create sort of a product or an accelerator that can do more than just that small piece of work that you were doing. 

I think that to me was the single biggest piece of advice that helped me as I was starting off my career. And it has, in different shapes and forms, remained true as I've gone along.


Challenges on the horizon for insights professionals

Thor - 00:19:48:

That's such good advice because it's true no matter how far you progress and no matter what challenges you actually face. And if we continue talking about challenges and opportunities, what challenges do you see that could potentially face insights professionals and the wider industry in the near future?


Bhaskar - 00:20:12:

Sure. So the first and foremost thought there is the world around us continues to change. It has been referred to differently as VUCA and all of the fancy terms that people mention, but that is more true today than it has probably been previously. 

The last couple of years, three years now, with COVID and the related impacts has shown us that there will be times where you will be faced with situations that you've not seen before. How do you adapt what you are doing, given those new situations and circumstances, and build something that can understand the changing realities much more quickly than previously, so that your time to market, your time to response is much shorter than it previously was?

I think that is where all of us as insights professionals need to focus on: how do we reduce that time to market, time to time to action, and then continue leveraging that to drive the business forward?


Going beyond customer centricity

Thor - 00:21:13:

And a lot of us in the industry have been talking about human centricity going beyond customer centricity. Have you applied that and how have you thought about that if you take a look at your actual work?


Bhaskar - 00:21:27:

A couple of things right? One is whichever organization that I might be supporting today as part of my current role, there are always two parts of any solution that you're developing. One of course, as you rightly said, is starting with customer centricity or client centricity. How do you keep the customer at the center and then figure out the right decisions that need to be made or enabled? 

But then realizing that at the end of the day, even though you are creating that solution, for the customer. It is a human on the client side as well that is implementing that solution that is going to be looking at that solution to drive that forward. 

And it is that duopoly almost, if I may call it that, we need to keep playing with the customer and the implementer of that decision or system and how do you ensure that any solution that you create caters to both the needs whether explicit or latent? And how do you discover that in different ways is what we need to look at. 

A classic way in which we are doing that today in different engagements is for example bringing in design thinking. You've always had panels of customers and consumers which kind of gives you feedback on what they are looking for. But then design thinking so that you ensure that you are creating the solution, keeping in mind that stakeholder who will leverage that solution at the end of the day so that both those situations or both those facets get taken care of as you create something.


How insights professionals can challenge the status quo

Thor - 00:22:57:

I absolutely love that. And if we stay within the realm of impact and opportunities, what is it that you see that insights professionals could do to bring true business impact and to challenge what many describe as the status quo?


Bhaskar - 00:23:13:

The key part and differentiator that I have seen through my career is to be willing to put yourself on the line, hold yourself accountable, and sit with the business teams. We are all very comfortable sitting in our own data and insight silos and churning out the best models and the best insights that we can think about, and getting excited by all this new stuff that we bring out. But end of the day, it has to show value. Where the rubber hits the road is where the impact gets seen. Therefore, sitting with your business teams, being on the front line, and even sometimes maybe challenging yourself to see if you want to do a rotational role in the business, because that will give you a very different perspective than other professionals who might be doing your kind of role today. And then of course, you're more than welcome to come back to your original role, but that flavor of business gives a different dimension altogether to what you're doing.


Who in the world of insights Bhaskar would love to have lunch with

Thor - 00:24:11:

I absolutely love what you just said, Bhaskar, because it resonates really well with one thing I keep on hearing, which is having the CEO mindset, because if you are the CEO, you are really accountable for whatever happens, and it's a really great way of summarizing what that effectively means. So I absolutely love that advice. 

Unfortunately, Bhaskar I think it's been such a good conversation, packed with great advice and experiences, and we've come to the end of our conversation. But there's one more question that I always love to ask, and I'm going to ask you the same question, which is: who in the world of Insights would you love to have lunch with?


Bhaskar - 00:24:55:

In my mind, I define the world of insights a little broadly, right? And as I think of it, there are a number of different folks that I would love to meet and exchange ideas with, Geoffrey Hinton being one of them, of thinking of different ways in which we can challenge computational powers to drive impact for business. 

And second, politicians of today, any number of them, because they are the ones who are truly harnessing the value of data and insight to do what they want to do. Whether right or wrong, is up for debate. But they are the ones leveraging it to the hilt and doing what is possible out there today.



Thor - 00:25:36:

Wow, this has been such an interesting conversation, Bhaskar. You have a truly unique perspective to offer, and I've been especially inspired by how you've highlighted the intersection between insights and sustainability. 

I'm going to play back some of the things you shared with us today, which is starting with the definition of the insight. It's something that should help the business go forward. Otherwise it's just a piece of data. It should not only drive top line, but also efficiency in the business. 

When talking about AI, you reminded us that AI in itself is probably not enough. The question is how you deploy it the right way. AI's impact is limited in the regulatory environment. It's also very limited in areas where there is very little data at hand. But at the end, it needs to be an AI plus human decision, humans as the augmenting layer. 

And lastly, the best advice you have received, which I think we will all treasure, is for you to continue to grow and learn, you need to make yourself redundant to your current role. You should not be doing your active job, so you have time to learn. What is it that you can automate? What synergies are there around you? And if you're eager to make true business impact, put yourself on the line and make yourself accountable. 

Bhaskar, I know that I've learned a lot from talking with you today, and I'm sure our audience has as well. Thank you for joining me.


Bhaskar - 00:27:09:

Thank you so much for having me, Thor.