Citibank uses data, analytics, and AI to improve the customer experience, but that's only the beginning. We speak with Murli Buluswar, Citi's Head of Analytics for the US Consumer Bank, about how his team supports businesses from origination to acquisition to customer retention and beyond. He also shares practical advice on how to bring data and analytics into your own organization.

The conversation includes these topics:

As Head of U.S. Consumer Analytics (USCA), Murli Buluswar is leading the execution of the vision to ‘be a critical partner in achieving a quantum leap in customer intelligence’ across Citibank’s 30MM+ customer base. An ‘intrapreneurial’ analytical and strategic C-Suite Financial Services Leader, Murli has successfully built alignment and achieved commercial outcomes by challenging conventional processes while harnessing data-driven intelligence. A former member of The Operating Committee of The American International Group (AIG), Murli has influenced Fortune 100 board members and senior leaders to expand innovative thinking and execution through data insights. As Chief Science Officer for AIG, Murli built a (data) Science team to ‘serve as a catalyst for consistent evidence-based decision making at AIG.’ Science served as an internal data-driven innovation unit with a complementary set of capabilities in: problem structuring, data engineering, behavioral science, data science, and software development. The Science Team helped reduce operating costs, improve loss ratios, increase sales, and engage with customers to reduce risk.

Prior to Citi, Murli was a Senior Advisor to The Boston Consulting Group (BCG) and two startups. He helped shape clients’ ability to retool decision making powered by data.

Transcript

Murli Buluswar: Through data, we are essentially truth seekers. We can seek truth based on questions that people have already asked. We could also be asking, "How could the world be reimagined? How could a particular decision be reimagined, powered by deeper data insights in ways that we hadn't solved for before?"

On data science and analytics at Citibank

Michael Krigsman: That's Murli Buluswar, the head of U.S. Consumer Analytics for Citibank.

Murli Buluswar: My team's mission is to be a critical partner in achieving a quantum leap of customer intelligence across the U.S. personal bank. What does that mean? If you sort of think about the banking sector, it's a very high engagement sector, with quite a bit of transactions and interactions compared to many other sectors. How do we take all of those interactions and transactions, quantify them, and understand customers' stated and latent needs better and build more relevance with them on a near real-time basis through the power of machine learning and real-time intelligence?

Michael Krigsman: So, Murli, you just described to me two different things. One is the notion of technology and number two is the notion of people. Is that a fair assessment of how you view the world?

Murli Buluswar: Yes, I'd say it's the intersection of people, technology, and this notion of curiosity; this ability to imagine what decisions are made and how decisions are made; and connecting that to people and technology. It's really sort of that trifecta is what I think about quite a bit in my current role.

On the role of data science in financial services

Michael Krigsman: How do data science and analytics come into play in financial services, in general, and then, in particular, at Citi and the work that you do?

Murli Buluswar: There's really three things that I think about, Michael.

  • Number one is this notion of getting smarter in how we spend our precious marketing dollars in service of a myriad of objective functions. So, building deeper intelligence in how we understand marketing spend and how we understand customer behavior.
  • The second is in this notion of building deeper insights that are more contextual in how we understand customer needs and how we serve them, whether they've asked for it or whether they're implying that through signaling (on a near real-time basis again).
  • Then the third is the application of data-driven intelligence to superior risk controls in service of continuing to build a better bank that harnesses data in every aspect of what we do.

How are data science and analytics used in banking?

Michael Krigsman: What are the kinds of business challenges or business problems or opportunities that you look at when you're thinking about how to apply this kind of machine intelligence and data and analytics?

Murli Buluswar: In that first bucket, one simple example is answering the question, "What is that efficiency frontier of how much we should be investing in marketing, where, which customer, what offer, which product, which channel, and why?"

There's an enormous amount of science behind that, whether it is for direct mail, whether it is for digital, whether it is for above-the-line advertising. We build a science underneath that, and what we also do is we augment that science with software solutions that allow for better decision-making across multiple functions within Citi.

Michael Krigsman: Can you give us some examples of where this comes into play?

Murli Buluswar: At the end of the day, in any firm, marketing dollars and how we engage with customers are scarce resources. And the question really is, "What is your objective function?"

Your objective function could be maximizing the NPV, maximizing customer satisfaction, maximizing return on equity, or a myriad of potentially complementary as well as sometimes competing objectives. How do you make those decisions? What's the science behind that? And how do you build confidence that you're making the best decisions given the knowledge and the predictions that you have?

That requires the ability to have top-notch, granular data-driven insights, but it also requires the ability to ingest those data insights into software and tools that a variety of coworkers can actually engage with.

Michael Krigsman: So, you need to have real clarity both about the kinds of technologies that are available, the data that's available, and then the nature of the business problem and the strategy, so the underlying context into which you place or you use the data and technology.

Murli Buluswar: One hundred percent, and the other thing I might add to that, Michael, is what I was alluding to earlier. It's this notion of imagination. It's this ability to think through how would we want decisions to be made differently than they are being made today or yesterday.

What are the questions we should be asking that are a couple of layers deeper than what we may have historically, and how can data-driven insights, predictions, enable better decision-making all around? It's really, again, going back to those three dimensions that we were talking about earlier.

How Citibank selects which data problems to solve?

Michael Krigsman: The choice of, or the selection of, a problem is a very multifaceted one then.

Murli Buluswar: Absolutely, and we're just unpacking one problem statement, and it's a very important one (certainly for Citibank, but also across, frankly, industries and organizations) but it's a reflection of the kind of curiosity that my team brings on a day-in-day-out basis to rethinking problems, some of which have been identified and some of which maybe people haven't necessarily thought through in totality as to how advanced analytics could help solve those.

Michael Krigsman: I find it interesting that you've used this term "curiosity" now a couple of times. We tend to think about data as technology-driven, but when you talk about curiosity, that has nothing to do with the technology.

Murli Buluswar: My mantra for my team (and, frankly, for myself as well) is that I want this team to traverse the three dimensions of followership, partnership, and leadership every single day. What does that mean?

Through data, we are essentially truth seekers. We can seek truth based on questions that people have already asked, which could be followership and some partnership.

But in addition, we could also be asking how could the world be reimagined, how could a particular decision be reimagined, powered by deeper data insights in ways that we hadn't solved for before. That sort of dimension of leadership requires this ability to think about issues in a commercial lens, work with our functional partners in multiple areas, and then sort of bring data-driven insights in a way that maybe wasn't imagined before.

Michael Krigsman: That's a very interesting comment. You said, "Through data, we are truth seekers." I don't think I've ever heard anybody put it quite like that. Honestly, I don't hear bankers talk about their work that way so very often actually at all. Maybe you can elaborate on that. It's a very kind of compelling thought.

Murli Buluswar: This is an issue, in my experience, Michael, that cuts across multiple industries, which is that, in many areas, we might be making judgments, decisions, business decisions, based on heuristics, based on past expertise. Those are both very powerful and very pertinent.

The question really is, if we were to bring deeper, more granular data, how could that shape even better decision-making at that next level? That requires this ability to have a hypothesis on how you could imagine decisions being made, but then it requires the ability to say, "Look. That's a hypothesis. I don't know whether that is true or not. I don't know where or to what extent that is true or not. How could I tap into my granular transaction level data to be able to answer a question and sort of understand what we should do differently for 100 Michaels versus 100 Murlis and such in ways that are more relevant and more insightful and allows us as a bank to build deeper engagement with customers?"

On using data science and analytics to achieve specific business outcomes

Michael Krigsman: At each step then it sounds like you are taking the data and analytics opportunity (just to put it that way) and aligning it extremely closely or calibrating against specific business outcomes, business results that you're trying to achieve.

Murli Buluswar: Absolutely. That connects me to this notion, Michael. Oftentimes, we think about data and analytics, and those are just things. Analytics is an activity.

What I care about is this notion of decision sciences, which is, how is it that through science we are powering vastly superior decision-making? That's what guides what my team does, the questions that we ask, the nuances that we bring to questions that our partners are asking, and the deeper intelligence that we try to bring to the table in order to shape better decision-making.

Michael Krigsman: Please subscribe to our newsletter. Hit the subscribe button at the top of our website and subscribe to our YouTube channel so we can keep you notified about amazing, upcoming shows.

Again, I keep coming back to this, the calibration point is the result. That's where you're focused. It sounds like you're not focused on the mechanism or the mechanics, necessarily, of getting there, but you're focused on the calibration with the result.

Murli Buluswar: I think the mechanics matter very much. But the mechanics in a vacuum don't matter unless they're connected to the outcome.

At the end of the day, if you think about the role – and this goes back to my philosophy around follow, partner, leader – at the end of the day, we are enablers of better decision-making. And so, if we were to deliver analysis or a set of insights and recommendations, but we weren't engaged or had a line of sight into how those insights translate into decision and outcomes, then we've missed an opportunity to be more strategically relevant.

I think it's important for teams with this set of skills (across firms and industries) to have that DNA of asking the question, how are you enabling better decision-making? What is that connectivity? And what does it take to solve for that last mile problem of change management? Do you need to build tools that allow those models and those insights to be ingested and acted upon in a more seamless way? Really kind of taking a much more integrated approach to how do you measure success.

In my instance, for example, one of my primary categories is what is the revenue-neutral cost savings that this team has delivered. Those revenue-neutral cost savings cannot be delivered unless they actually manifest themselves in savings.

Another category is revenue-driven EBITA. That's a growth story.

How is data science and decision science enabling growth, and how do we measure the financial implications of that growth? Those are two large buckets of very tangible financial metrics.

Then we have three other buckets that relate to customer and risk controls, and speed and simplification that aren't necessarily financially measured, but each one of them has what I think of as material outcomes that clearly connects to significantly better decision-making and an important part of the bank.

Michael Krigsman: It's really fascinating to hear you talk about things this way. So, you're looking at that complete chain through the mechanics, meaning the tools and the technologies, again linked to precise outcomes that you're trying to gain.

Murli Buluswar: My challenge to myself and to my team is to take a general management or general manager perspective. At the end of the day, an insight, a model, or a prediction can be intellectually very stimulating and fascinating and have some beauty of its own. However, if we're not able to connect that to what decisions are made and how decisions are made, and we're not able to measure – sometimes it's financial and other times it may not be financial, but measure, nonetheless – what impact is this having, why is this consequential to the sustainability of what this institution is looking to achieve, that question is a very important one for us to have a line of sight into. If not, you just become part of a value chain where you're delivering a service but you don't have integrated connectivity.

Examples of data science in banking

Michael Krigsman: This is from Prashant Motewar, and he says, "Great insights, Murli. Can you give us one precise example where data drove the business decisions that were difficult to drive using traditional decision-making processes?" so where data really helped out, where traditional decision-making would not work.

Murli Buluswar: Absolutely. Let's take this conversation from conceptual to something very real.

You think about Citibank. It's obviously a very accomplished, large institution. We spend a lot of dollars on marketing. Up until now, much of that spend has historically been in individual sort of allocated silos.

The problem statement is, how do you know if you're objective is to maximize NPV, or if your objective is to maximize three-year earnings, or your objective were to maximize return on equity, or even for that matter customer account? How do you know that you're actually spending your precious resources in service of that objective function to its fullest potential?

It's really a problem statement of local optima versus global optima. That requires an ability to understand the math at its most elemental level to be able to certainly build better models so that you have a full potential view of what the math would be at its most granular level.

It requires the ability to standardize exactly how you might measure each of these programs. And it requires the ability to connect all of that and now be able to say what is your objective function. If your objective function were to change from X to Y, how would that shape your investment decision differently? What are the financial and P&L and possibly balance sheet implications of those trade-offs, and how do we quantify them? And how do we not only simulate choices that you're thinking of making, but also identify what that efficiency Pareto optimal frontier is in terms of the best allocation of your scarce resources in service of a defined objective function?

That requires this ability to think about the problem at a very deep level. It requires the ability to have data-driven intelligence and predictions at relatively granular levels. It requires financial acumen. It requires the ability to understand the process through which decisions are made, and how you might sort of connection a software solution that allows a myriad of functional partners to be able to understand the tradeoffs and to create transparency and more agility in decision-making.

Michael Krigsman: Murli, you sound like an artist with a palette of paints, painting the picture to come up with the outcome that you want.

Murli Buluswar: I have this belief, Michael, that the most complex, sophisticated problems that not only businesses but society as a whole face cannot be solved with a single tool. So, I think that the key for the future, whether it is global warming or whether it's issues in risk and controls that the banking sector faces, or whether it is something in pharmaceutical or consumer goods, these problems have to be solved by multidisciplinary thinking.

They have to be solved by connecting the dots through a multifunctional view. And they have to be solved by imagining or reimagining what the world could be versus what it is.

That requires that ability, in my view, to be curious and open-minded and to be declarative sometimes in what you believe, but also that humility to say, "You know what? Any given function, no matter who you are, will not have all the answers."

The magic is really in how you bring these functions collectively in service of a problem or opportunity. And so, being able to make sure that you are able to switch gears and not take a singular lens view of an issue or an opportunity allows you to be more effective in creating change.

Examples of using data science to improve long-term customer outcomes at Citibank

Michael Krigsman: This is from Lisbeth Shaw who says, "Can you give us an example of how better granular data has improved decision-making at Citibank?" She's interested in both the problem, the data, and how it led to a solution.

Murli Buluswar: Most certainly. One of the questions that we think of quite a bit is our customers engage with us on a variety of digital real estates: on the mobile app, online, and such. The question is, how do you understand the customer needs at that point in time when they log in, and how do you engage with them in a way that builds greater intimacy and relevance?

The way we've tackled that challenge – or we are, rather, tackling that challenge (because it's always a work in progress) – is we have more than 2,000 interactions and transactions that we stitch together at a customer level. We try and understand how that data, that near real-time data, is meaningful in terms of what could be most relevant and appropriate for a particular customer. And we translate that intelligence into the messaging and the engagement with the customer on a live basis in a way that again builds deeper engagement and relevance and intimacy.

That has a myriad of benefits. Some of it is financial. Some of it is non-financial. But it's really this notion of mirroring back to the customer that you understand them, that you care about them, that you know them, and the way you do that is by using all of the data that you have in terms of every channel, every interaction, every transaction that you have and drawing meaning from that as and when appropriate.

Michael Krigsman: You have not used the term customer experience, which kind of surprises me. Is there any reason you haven't used that term or it's just not how you think about it? Because that's what you're describing is this overall chain of customer experience (in a very broad way).

Murli Buluswar: That's absolutely right. I use the word. You heard me use the word engagement. You heard me use the word connectivity. You heard me use the word relevance and a couple of other synonyms. But it is very much about customer experience.

Now, customer experience is in service of customer engagement, intimacy, and building more relevance. I think of customer experience as a point in time. How can you sort of be the best version of yourself for the customer in a particular interaction? But if you stitch together a series of customer experiences, what you're doing is building deeper connectivity, relevance, and engagement with your customer.

In my definition, in my parlance, customer engagement is nothing but a series of powerful, positive customer experiences stitched together to create more of an emotional connectivity with the customer.

Michael Krigsman: You say that customer experience is the presentation or being the best version of yourself. I assume you mean that as a business. Can you elaborate? What do you mean by that?

Murli Buluswar: If you think about the sector of banking, it's a very emotive sector. It matters to customers.

Our financial health, our ability to make transactions – go to a branch, make withdrawals, or send transfers, or swipe our card – all of those things have meaning to us. If they don't actually work in a seamless way that we care about, then it causes us friction and pain. So, it's a high interaction, high transaction, high engagement, and high relevance, therefore, sector.

And so, what I really mean by that, Michael, is every single time you walk into a branch or you call the call center or you hang up after a call with the call center, you log into the app, or you go online, or you swipe a card, if I can understand what you need and how I can be more relevant to you and how I could serve you to the best of my abilities better, then I am – again, going back to what I alluded to earlier – reflecting back to you that we understand you and we're here for you.

To me, data and being able to take real-time data, real-time transactions and interactions, translating them into machine learning algorithms that guide what that next best conversation is, is not only powerful but it's mission-critical because that's the baseline expectation that all of us have from every facet of our interactions, not just with our banks.

Michael Krigsman: Again from Prashant Motewar, a very interesting question that he is raising. He said, "You have mentioned business outcomes a number of times," and here's the question. "Are these outcomes defined by the leadership goals or more of a business opportunity that's created by the data or driven by the data?"

Murli Buluswar: At the end of the day, it sort of goes back to the first principles of how do you measure outcomes, are those outcomes material, and are there multiple functional partners that are essentially validating those outcomes so that it's not my team making a claim but, rather, it's an integral part of our financial planning and accounting processes that ensures that there's an end-to-end measurement? That's how we think about it. In that construct, this notion of materiality is very, very important to me.

(200:25:15) Why is it important? Because too often I've seen there's almost like bookends for this capability in large organizations, in my view, that are at risk. One bookend is to work on problems that are conceptual and might be interesting sometime down the road but don't have any near-term or medium-term commercial relevance.

At the other end of the spectrum, you could have this issue of, "We're here to provide you data insights based on a question that you ask, and we're not really thinking more deeply about the question behind the question. Nor are we asking questions that you're not asking." And so, for me, that connectivity is important and that measurement is important because, at the end of the day, if it's not manifesting in outcomes that are material, it's just an activity.

Michael Krigsman: You really are taking this very global, broad perspective on the business challenge, the business opportunity that you're addressing at a pretty deep level to try to figure out its relevance and importance when you're selecting the problems to work on.

Murli Buluswar: That's right, and it requires. This sort of goes back to my belief that the most sophisticated problems these days cannot just be solved by a single machine-learning algorithm. I think that, increasingly, that skillset, while it's very important and critical, will be commoditized.

It's really that integrated thinking. It's that understanding of how are those decisions made today. How will your predictions connect to that, and how will those decisions actually translate tomorrow? What changes in the upstream and downstream need to happen in order to connect the dots, in order to ensure that you're achieving (or you're part of the broader team that is achieving) end-to-end outcomes?

Michael Krigsman: So, you're saying that the technology (over time), the models become commoditized and so, therefore, it's the human ingenuity of adapting the models and the technologies to the business problems is where the real opportunity and value lies.

Murli Buluswar: I mean, increasingly, we live in a world where machines can actually build predictive models, so this notion of driverless AI is very real. That does not mean that it doesn't require a human in the loop. It 100% does.

However, it does actually improve the productivity of human beings in that whole model-building cycle process. Now then the question really is, what is that model in service of, how do you kind of think about what it takes to implement that model to its fullest potential, and how do you ensure that you have a holistic discipline in answering the question of why did that matter?

Michael Krigsman: That's the hard part.

Murli Buluswar: Yes. By far that is the hard part because it is emotional. It is not necessarily a well-contained objective function that you could solve just on your own.

It requires a whole lot more integrated thinking and problem-solving. It requires deeper collaboration, and it requires understanding the human element of how does this sort of affect my colleagues in other functional areas, how do I understand that, and how do I make sure that I can mirror that understanding, that empathy in the idea, the analysis, or the insight that I have?

Using data science and analytics to manage risk and uncertainty

Michael Krigsman: This is from Wayne Anderson who is a regular listener. He asks really, again, such insightful questions. He says, "When does analytics and the resulting quants become systemic risk?" And then he has another question for you as well. So, analytics and resulting quants, when does it become systemic risk? It's sort of a little off to the side here.

Murli Buluswar: For me, the question isn't whether it's quants or not. Firms (in a myriad of industries) make decisions today, and those decisions are made under conditions of uncertainty. Those decisions have inherent risk associated with them.

The thing that I think about more is how are we reducing noise – not to refer to Daniel Conoman's latest book, but how are we reducing noise – and bias in decision-making? How are we being more systematized, and how are we creating more transparency and understanding in our models and in our predictions that might otherwise be susceptible?

That to me is a deeper question, and my view is, regardless of which industry you take today, there are people making judgments. Those judgments are subject to all forms of human bias and, frankly, randomness and such idiosyncrasies.

There's, in my view, much more sort of risk there than there is in modeling and increasingly, just creating more transparency and understanding. And the fact that models and predictions drive more consistency and being able to shed a bright light on where that consistency is and what are the implications of that is a science that is fairly well advanced (in my experience today).

And so, for me, I actually believe that this is reducing bias. It's reducing risk. I worry more about the human judgment error and the bias and such associated with that, as long as we've got the appropriate guardrails in all things data science and understanding the implications.

Michael Krigsman: To what extent do you think about the ethical implications, meaning you were just talking about bias? Obviously, the decisions that any financial institution makes based on data can have a real impact on people's lives. Where does that aspect come into this?

Murli Buluswar: This is applications to healthcare, insurance, and a variety of other sectors. The question that I would challenge us to come back to is there are ethical implications today of human judgment and the biases that we all have as human beings.

Those biases have consequential impact, whether it's in real estate, whether it's in insurance, whether it's in healthcare, or pretty much any and every industry that you could think about. And so, for me, isolating – I wouldn't isolate the impact of data science or machine learning in asking this question because those biases have existed historically as long as humankind has existed.

The question is, how can we continue to create more transparency, objectivity, understanding, and discipline with models, which I think is vastly easier to do than trying to unwire the complex, human mind that makes decisions in ways that aren't always logical?

How to build a data-centric culture?

Michael Krigsman: Another dimension to this is the notion of building a data-centric culture, data-centric organization. Any thoughts on that aspect?

Murli Buluswar: It's a curiosity-centric culture. Data is nothing but a set of facts and information in service of curiosity, imagination, and questions that people are asking.

For me, oftentimes when we think about data-centric culture, it tends to be a little bit too much of a binary world. For me, it is very, very important for the sustainability of organizations and even for the sustainability of people in their individual careers to really continue to challenge their thinking about curiosity and imagination-centric work that they bring day-in-day-out.

Data is in service of that. It whets your appetite. But data unto itself can do nothing if we don't have curiosity and imagination.

But if we do have curiosity and imagination, data can serve that. It can dispel any hypotheses we have. It can reinforce hypotheses we can have. And it can help us make better decisions.

For me, the imperative for every large organization is to ask itself, "When and where are we curious and imaginative in challenging what we understand to be true in this world, and how will we up the game on that front tomorrow versus today?" because without curiosity and imagination (in my opinion) no meaningful innovation has ever been achieved by humankind.

Michael Krigsman: I keep coming back in my mind to the idea that, for most people, for most organizations, when they think about data, it's much more mechanistic. You think about it in a very creative way.

As I said before, it's like an artist or a painter thinks about it. Most organizations don't do that, and I think that there's a leap that many organizations would have to make to get to what you're describing. I don't think it'd be easy for many companies to do that.

Murli Buluswar: Large-scale change is hard for all of us as individuals and certainly large organizations as well.

You'll notice that when I talk about decision sciences, I almost don't use the word data because, at the end of the day, for me, it is about thinking about how we could make better decisions and how will that sort of serve us in our mission to be of more service to our customers and be a better run institution tomorrow than today. Backing that into, "Hey, data can actually enable that."

I think that if we talk about data-driven cultures, it can become a little bit intimidating because, all of a sudden, we can identify as we're not data and someone else's data, perhaps. But if we were to say, "Forget about data. Can I imagine these ten decisions that I make today as part of my role? Could I imagine how I might make those decisions differently tomorrow so that I'm not taking a mechanized approach to my role, but I'm blending my expertise and my understanding with a little bit more imagination and curiosity?"

If I can blend that, then data can play a very pivotal role in informing that. But to me, it actually starts with curiosity and challenging our imagination and understanding. That's why you see me (over and over, Michael) coming back to that theme because, for me, data unto itself is just a bunch of digits. It's numbers. It's a thing. it's not really inspiring unto itself. Nor is it (in a blindingly obvious way) driving something better.

It's the people and their creativity, their imagination, their curiosity, and their hypothesis-driven approach, informed by data, is where the magic happens.

Michael Krigsman: This is from Arsalan Khan. He's a regular listener. He always asks great questions. This is a little orthogonal, but it's an interesting, good question.

He says, "How can consumers protect themselves without even knowing that they're being judged by data?" He's responding to your point earlier about making judgments about data.

Murli Buluswar: Data or no data, every firm or every industry that we're a part of, firms are trying to sort of figure out how to be engaged with us as consumers. Those decisions are happening with or without data.

Don't think of it as judgment but think of it as are they being more relevant with you, for you. If they are not, shut down that relationship. Go find a partner who is demonstrating that radical empathy for who you are as a consumer.

How Citibank uses data analytics in banking to improve customer service, loyalty, and retention?

Michael Krigsman: Another question. This is now jumping back to LinkedIn. This is from Debanshu Debroy. He says, "Do you see any opportunity to use the learnings from CX analytics in training or aiding contact center representatives to better serve their customers?"

Murli Buluswar: Yes. Every time we interact with a customer through our call center, there's an emotion element. There's a human element. There's an interaction element. There's a transaction element. That is all being codified by structured and unstructured data.

That is a golden opportunity for us to understand how we did, what we could have done better, and how we might potentially reconnect with that customer (post that call) because you have all of the data that happened before the call, anything that happens after the call, and you have all the data from the call.

There's lots of pixie dust there, you know, sort of nuggets of gold in that. And I think every firm that interacts with customers through those channels should take it upon themselves to say, "What can we do better at this point in time?"

Michael Krigsman: Wayne Anderson says, "If curiosity and imagination are the key," he says, "you can teach knowledge, but how do you encourage imagination?"

Murli Buluswar: You can create a culture that is supportive and encouraging. Not everybody is going to be at the same point, but everybody is on their own journey. What's important is to encourage it, to reward it, to recognize it, and support it.

Michael Krigsman: This is from Mike Prest, who is a CIO. He says, "Do you believe that how you harness behavior, data and analytics today will help your organization become a change agent through digital disruption?" It's a really good question, the role of data and analytics in driving change and managing disruption.

Murli Buluswar: Yes, I think of data and analytics as a mitochondria that's going to drive digital engagement, understanding, and intelligence in a virtuous cycle in how we connect with customers, certainly both within Citi and within banking, but also beyond.

Michael Krigsman: What advice do you have for business leaders on driving successful outcomes with data?

Murli Buluswar: Start with asking three problems that you might like to solve differently that are material for you and imagine why they matter to you. Challenge your decision sciences team to think about how they could inform that solution or that problem through deeper data intelligence and create a virtuous cycle of interaction and connectivity that can build on itself and that allows you to set a higher bar for what you get, what you achieve, and what you accomplish with this team, through this team.

Michael Krigsman: My advice to anybody that just heard that answer is that if it's relevant for you, there's a lot that Murli just said that needs to be unpacked there. Unfortunately, we're out of time, so we're not going to be able to unpack it now, but you can replay the video and parse it.

With that, a huge thank you to Murli Buluswar from Citibank. Murli, thank you so much for taking time to be with us today.

Murli Buluswar: My pleasure, Michael. I appreciate you, I appreciate the audience, and this has been a delightful conversation. Wonderful to reconnect. Thank you.

Michael Krigsman: Everybody who is watching, thank you for taking your time to be with us today and especially those folks who asked such marvelous questions from the live audience. Before you go, please subscribe to our newsletter, hit the subscribe button at the top of our website, and subscribe to our YouTube channel so we can keep you notified about amazing, upcoming shows that we have. Check out CXOTalk.com.

Everybody, I hope you have a great week, and we will see you again soon. Have a nice day.

Murli Buluswar: Through data, we are essentially truth seekers. We can seek truth based on questions that people have already asked. We could also be asking, "How could the world be reimagined? How could a particular decision be reimagined, powered by deeper data insights in ways that we hadn't solved for before?"

On data science and analytics at Citibank

Michael Krigsman: That's Murli Buluswar, the head of U.S. Consumer Analytics for Citibank.

Murli Buluswar: My team's mission is to be a critical partner in achieving a quantum leap of customer intelligence across the U.S. personal bank. What does that mean? If you sort of think about the banking sector, it's a very high engagement sector, with quite a bit of transactions and interactions compared to many other sectors. How do we take all of those interactions and transactions, quantify them, and understand customers' stated and latent needs better and build more relevance with them on a near real-time basis through the power of machine learning and real-time intelligence?

Michael Krigsman: So, Murli, you just described to me two different things. One is the notion of technology and number two is the notion of people. Is that a fair assessment of how you view the world?

Murli Buluswar: Yes, I'd say it's the intersection of people, technology, and this notion of curiosity; this ability to imagine what decisions are made and how decisions are made; and connecting that to people and technology. It's really sort of that trifecta is what I think about quite a bit in my current role.

On the role of data science in financial services

Michael Krigsman: How do data science and analytics come into play in financial services, in general, and then, in particular, at Citi and the work that you do?

Murli Buluswar: There's really three things that I think about, Michael.

  • Number one is this notion of getting smarter in how we spend our precious marketing dollars in service of a myriad of objective functions. So, building deeper intelligence in how we understand marketing spend and how we understand customer behavior.
  • The second is in this notion of building deeper insights that are more contextual in how we understand customer needs and how we serve them, whether they've asked for it or whether they're implying that through signaling (on a near real-time basis again).
  • Then the third is the application of data-driven intelligence to superior risk controls in service of continuing to build a better bank that harnesses data in every aspect of what we do.

How are data science and analytics used in banking?

Michael Krigsman: What are the kinds of business challenges or business problems or opportunities that you look at when you're thinking about how to apply this kind of machine intelligence and data and analytics?

Murli Buluswar: In that first bucket, one simple example is answering the question, "What is that efficiency frontier of how much we should be investing in marketing, where, which customer, what offer, which product, which channel, and why?"

There's an enormous amount of science behind that, whether it is for direct mail, whether it is for digital, whether it is for above-the-line advertising. We build a science underneath that, and what we also do is we augment that science with software solutions that allow for better decision-making across multiple functions within Citi.

Michael Krigsman: Can you give us some examples of where this comes into play?

Murli Buluswar: At the end of the day, in any firm, marketing dollars and how we engage with customers are scarce resources. And the question really is, "What is your objective function?"

Your objective function could be maximizing the NPV, maximizing customer satisfaction, maximizing return on equity, or a myriad of potentially complementary as well as sometimes competing objectives. How do you make those decisions? What's the science behind that? And how do you build confidence that you're making the best decisions given the knowledge and the predictions that you have?

That requires the ability to have top-notch, granular data-driven insights, but it also requires the ability to ingest those data insights into software and tools that a variety of coworkers can actually engage with.

Michael Krigsman: So, you need to have real clarity both about the kinds of technologies that are available, the data that's available, and then the nature of the business problem and the strategy, so the underlying context into which you place or you use the data and technology.

Murli Buluswar: One hundred percent, and the other thing I might add to that, Michael, is what I was alluding to earlier. It's this notion of imagination. It's this ability to think through how would we want decisions to be made differently than they are being made today or yesterday.

What are the questions we should be asking that are a couple of layers deeper than what we may have historically, and how can data-driven insights, predictions, enable better decision-making all around? It's really, again, going back to those three dimensions that we were talking about earlier.

How Citibank selects which data problems to solve?

Michael Krigsman: The choice of, or the selection of, a problem is a very multifaceted one then.

Murli Buluswar: Absolutely, and we're just unpacking one problem statement, and it's a very important one (certainly for Citibank, but also across, frankly, industries and organizations) but it's a reflection of the kind of curiosity that my team brings on a day-in-day-out basis to rethinking problems, some of which have been identified and some of which maybe people haven't necessarily thought through in totality as to how advanced analytics could help solve those.

Michael Krigsman: I find it interesting that you've used this term "curiosity" now a couple of times. We tend to think about data as technology-driven, but when you talk about curiosity, that has nothing to do with the technology.

Murli Buluswar: My mantra for my team (and, frankly, for myself as well) is that I want this team to traverse the three dimensions of followership, partnership, and leadership every single day. What does that mean?

Through data, we are essentially truth seekers. We can seek truth based on questions that people have already asked, which could be followership and some partnership.

But in addition, we could also be asking how could the world be reimagined, how could a particular decision be reimagined, powered by deeper data insights in ways that we hadn't solved for before. That sort of dimension of leadership requires this ability to think about issues in a commercial lens, work with our functional partners in multiple areas, and then sort of bring data-driven insights in a way that maybe wasn't imagined before.

Michael Krigsman: That's a very interesting comment. You said, "Through data, we are truth seekers." I don't think I've ever heard anybody put it quite like that. Honestly, I don't hear bankers talk about their work that way so very often actually at all. Maybe you can elaborate on that. It's a very kind of compelling thought.

Murli Buluswar: This is an issue, in my experience, Michael, that cuts across multiple industries, which is that, in many areas, we might be making judgments, decisions, business decisions, based on heuristics, based on past expertise. Those are both very powerful and very pertinent.

The question really is, if we were to bring deeper, more granular data, how could that shape even better decision-making at that next level? That requires this ability to have a hypothesis on how you could imagine decisions being made, but then it requires the ability to say, "Look. That's a hypothesis. I don't know whether that is true or not. I don't know where or to what extent that is true or not. How could I tap into my granular transaction level data to be able to answer a question and sort of understand what we should do differently for 100 Michaels versus 100 Murlis and such in ways that are more relevant and more insightful and allows us as a bank to build deeper engagement with customers?"

On using data science and analytics to achieve specific business outcomes

Michael Krigsman: At each step then it sounds like you are taking the data and analytics opportunity (just to put it that way) and aligning it extremely closely or calibrating against specific business outcomes, business results that you're trying to achieve.

Murli Buluswar: Absolutely. That connects me to this notion, Michael. Oftentimes, we think about data and analytics, and those are just things. Analytics is an activity.

What I care about is this notion of decision sciences, which is, how is it that through science we are powering vastly superior decision-making? That's what guides what my team does, the questions that we ask, the nuances that we bring to questions that our partners are asking, and the deeper intelligence that we try to bring to the table in order to shape better decision-making.

Michael Krigsman: Please subscribe to our newsletter. Hit the subscribe button at the top of our website and subscribe to our YouTube channel so we can keep you notified about amazing, upcoming shows.

Again, I keep coming back to this, the calibration point is the result. That's where you're focused. It sounds like you're not focused on the mechanism or the mechanics, necessarily, of getting there, but you're focused on the calibration with the result.

Murli Buluswar: I think the mechanics matter very much. But the mechanics in a vacuum don't matter unless they're connected to the outcome.

At the end of the day, if you think about the role – and this goes back to my philosophy around follow, partner, leader – at the end of the day, we are enablers of better decision-making. And so, if we were to deliver analysis or a set of insights and recommendations, but we weren't engaged or had a line of sight into how those insights translate into decision and outcomes, then we've missed an opportunity to be more strategically relevant.

I think it's important for teams with this set of skills (across firms and industries) to have that DNA of asking the question, how are you enabling better decision-making? What is that connectivity? And what does it take to solve for that last mile problem of change management? Do you need to build tools that allow those models and those insights to be ingested and acted upon in a more seamless way? Really kind of taking a much more integrated approach to how do you measure success.

In my instance, for example, one of my primary categories is what is the revenue-neutral cost savings that this team has delivered. Those revenue-neutral cost savings cannot be delivered unless they actually manifest themselves in savings.

Another category is revenue-driven EBITA. That's a growth story.

How is data science and decision science enabling growth, and how do we measure the financial implications of that growth? Those are two large buckets of very tangible financial metrics.

Then we have three other buckets that relate to customer and risk controls, and speed and simplification that aren't necessarily financially measured, but each one of them has what I think of as material outcomes that clearly connects to significantly better decision-making and an important part of the bank.

Michael Krigsman: It's really fascinating to hear you talk about things this way. So, you're looking at that complete chain through the mechanics, meaning the tools and the technologies, again linked to precise outcomes that you're trying to gain.

Murli Buluswar: My challenge to myself and to my team is to take a general management or general manager perspective. At the end of the day, an insight, a model, or a prediction can be intellectually very stimulating and fascinating and have some beauty of its own. However, if we're not able to connect that to what decisions are made and how decisions are made, and we're not able to measure – sometimes it's financial and other times it may not be financial, but measure, nonetheless – what impact is this having, why is this consequential to the sustainability of what this institution is looking to achieve, that question is a very important one for us to have a line of sight into. If not, you just become part of a value chain where you're delivering a service but you don't have integrated connectivity.

Examples of data science in banking

Michael Krigsman: This is from Prashant Motewar, and he says, "Great insights, Murli. Can you give us one precise example where data drove the business decisions that were difficult to drive using traditional decision-making processes?" so where data really helped out, where traditional decision-making would not work.

Murli Buluswar: Absolutely. Let's take this conversation from conceptual to something very real.

You think about Citibank. It's obviously a very accomplished, large institution. We spend a lot of dollars on marketing. Up until now, much of that spend has historically been in individual sort of allocated silos.

The problem statement is, how do you know if you're objective is to maximize NPV, or if your objective is to maximize three-year earnings, or your objective were to maximize return on equity, or even for that matter customer account? How do you know that you're actually spending your precious resources in service of that objective function to its fullest potential?

It's really a problem statement of local optima versus global optima. That requires an ability to understand the math at its most elemental level to be able to certainly build better models so that you have a full potential view of what the math would be at its most granular level.

It requires the ability to standardize exactly how you might measure each of these programs. And it requires the ability to connect all of that and now be able to say what is your objective function. If your objective function were to change from X to Y, how would that shape your investment decision differently? What are the financial and P&L and possibly balance sheet implications of those trade-offs, and how do we quantify them? And how do we not only simulate choices that you're thinking of making, but also identify what that efficiency Pareto optimal frontier is in terms of the best allocation of your scarce resources in service of a defined objective function?

That requires this ability to think about the problem at a very deep level. It requires the ability to have data-driven intelligence and predictions at relatively granular levels. It requires financial acumen. It requires the ability to understand the process through which decisions are made, and how you might sort of connection a software solution that allows a myriad of functional partners to be able to understand the tradeoffs and to create transparency and more agility in decision-making.

Michael Krigsman: Murli, you sound like an artist with a palette of paints, painting the picture to come up with the outcome that you want.

Murli Buluswar: I have this belief, Michael, that the most complex, sophisticated problems that not only businesses but society as a whole face cannot be solved with a single tool. So, I think that the key for the future, whether it is global warming or whether it's issues in risk and controls that the banking sector faces, or whether it is something in pharmaceutical or consumer goods, these problems have to be solved by multidisciplinary thinking.

They have to be solved by connecting the dots through a multifunctional view. And they have to be solved by imagining or reimagining what the world could be versus what it is.

That requires that ability, in my view, to be curious and open-minded and to be declarative sometimes in what you believe, but also that humility to say, "You know what? Any given function, no matter who you are, will not have all the answers."

The magic is really in how you bring these functions collectively in service of a problem or opportunity. And so, being able to make sure that you are able to switch gears and not take a singular lens view of an issue or an opportunity allows you to be more effective in creating change.

Examples of using data science to improve long-term customer outcomes at Citibank

Michael Krigsman: This is from Lisbeth Shaw who says, "Can you give us an example of how better granular data has improved decision-making at Citibank?" She's interested in both the problem, the data, and how it led to a solution.

Murli Buluswar: Most certainly. One of the questions that we think of quite a bit is our customers engage with us on a variety of digital real estates: on the mobile app, online, and such. The question is, how do you understand the customer needs at that point in time when they log in, and how do you engage with them in a way that builds greater intimacy and relevance?

The way we've tackled that challenge – or we are, rather, tackling that challenge (because it's always a work in progress) – is we have more than 2,000 interactions and transactions that we stitch together at a customer level. We try and understand how that data, that near real-time data, is meaningful in terms of what could be most relevant and appropriate for a particular customer. And we translate that intelligence into the messaging and the engagement with the customer on a live basis in a way that again builds deeper engagement and relevance and intimacy.

That has a myriad of benefits. Some of it is financial. Some of it is non-financial. But it's really this notion of mirroring back to the customer that you understand them, that you care about them, that you know them, and the way you do that is by using all of the data that you have in terms of every channel, every interaction, every transaction that you have and drawing meaning from that as and when appropriate.

Michael Krigsman: You have not used the term customer experience, which kind of surprises me. Is there any reason you haven't used that term or it's just not how you think about it? Because that's what you're describing is this overall chain of customer experience (in a very broad way).

Murli Buluswar: That's absolutely right. I use the word. You heard me use the word engagement. You heard me use the word connectivity. You heard me use the word relevance and a couple of other synonyms. But it is very much about customer experience.

Now, customer experience is in service of customer engagement, intimacy, and building more relevance. I think of customer experience as a point in time. How can you sort of be the best version of yourself for the customer in a particular interaction? But if you stitch together a series of customer experiences, what you're doing is building deeper connectivity, relevance, and engagement with your customer.

In my definition, in my parlance, customer engagement is nothing but a series of powerful, positive customer experiences stitched together to create more of an emotional connectivity with the customer.

Michael Krigsman: You say that customer experience is the presentation or being the best version of yourself. I assume you mean that as a business. Can you elaborate? What do you mean by that?

Murli Buluswar: If you think about the sector of banking, it's a very emotive sector. It matters to customers.

Our financial health, our ability to make transactions – go to a branch, make withdrawals, or send transfers, or swipe our card – all of those things have meaning to us. If they don't actually work in a seamless way that we care about, then it causes us friction and pain. So, it's a high interaction, high transaction, high engagement, and high relevance, therefore, sector.

And so, what I really mean by that, Michael, is every single time you walk into a branch or you call the call center or you hang up after a call with the call center, you log into the app, or you go online, or you swipe a card, if I can understand what you need and how I can be more relevant to you and how I could serve you to the best of my abilities better, then I am – again, going back to what I alluded to earlier – reflecting back to you that we understand you and we're here for you.

To me, data and being able to take real-time data, real-time transactions and interactions, translating them into machine learning algorithms that guide what that next best conversation is, is not only powerful but it's mission-critical because that's the baseline expectation that all of us have from every facet of our interactions, not just with our banks.

Michael Krigsman: Again from Prashant Motewar, a very interesting question that he is raising. He said, "You have mentioned business outcomes a number of times," and here's the question. "Are these outcomes defined by the leadership goals or more of a business opportunity that's created by the data or driven by the data?"

Murli Buluswar: At the end of the day, it sort of goes back to the first principles of how do you measure outcomes, are those outcomes material, and are there multiple functional partners that are essentially validating those outcomes so that it's not my team making a claim but, rather, it's an integral part of our financial planning and accounting processes that ensures that there's an end-to-end measurement? That's how we think about it. In that construct, this notion of materiality is very, very important to me.

(200:25:15) Why is it important? Because too often I've seen there's almost like bookends for this capability in large organizations, in my view, that are at risk. One bookend is to work on problems that are conceptual and might be interesting sometime down the road but don't have any near-term or medium-term commercial relevance.

At the other end of the spectrum, you could have this issue of, "We're here to provide you data insights based on a question that you ask, and we're not really thinking more deeply about the question behind the question. Nor are we asking questions that you're not asking." And so, for me, that connectivity is important and that measurement is important because, at the end of the day, if it's not manifesting in outcomes that are material, it's just an activity.

Michael Krigsman: You really are taking this very global, broad perspective on the business challenge, the business opportunity that you're addressing at a pretty deep level to try to figure out its relevance and importance when you're selecting the problems to work on.

Murli Buluswar: That's right, and it requires. This sort of goes back to my belief that the most sophisticated problems these days cannot just be solved by a single machine-learning algorithm. I think that, increasingly, that skillset, while it's very important and critical, will be commoditized.

It's really that integrated thinking. It's that understanding of how are those decisions made today. How will your predictions connect to that, and how will those decisions actually translate tomorrow? What changes in the upstream and downstream need to happen in order to connect the dots, in order to ensure that you're achieving (or you're part of the broader team that is achieving) end-to-end outcomes?

Michael Krigsman: So, you're saying that the technology (over time), the models become commoditized and so, therefore, it's the human ingenuity of adapting the models and the technologies to the business problems is where the real opportunity and value lies.

Murli Buluswar: I mean, increasingly, we live in a world where machines can actually build predictive models, so this notion of driverless AI is very real. That does not mean that it doesn't require a human in the loop. It 100% does.

However, it does actually improve the productivity of human beings in that whole model-building cycle process. Now then the question really is, what is that model in service of, how do you kind of think about what it takes to implement that model to its fullest potential, and how do you ensure that you have a holistic discipline in answering the question of why did that matter?

Michael Krigsman: That's the hard part.

Murli Buluswar: Yes. By far that is the hard part because it is emotional. It is not necessarily a well-contained objective function that you could solve just on your own.

It requires a whole lot more integrated thinking and problem-solving. It requires deeper collaboration, and it requires understanding the human element of how does this sort of affect my colleagues in other functional areas, how do I understand that, and how do I make sure that I can mirror that understanding, that empathy in the idea, the analysis, or the insight that I have?

Using data science and analytics to manage risk and uncertainty

Michael Krigsman: This is from Wayne Anderson who is a regular listener. He asks really, again, such insightful questions. He says, "When does analytics and the resulting quants become systemic risk?" And then he has another question for you as well. So, analytics and resulting quants, when does it become systemic risk? It's sort of a little off to the side here.

Murli Buluswar: For me, the question isn't whether it's quants or not. Firms (in a myriad of industries) make decisions today, and those decisions are made under conditions of uncertainty. Those decisions have inherent risk associated with them.

The thing that I think about more is how are we reducing noise – not to refer to Daniel Conoman's latest book, but how are we reducing noise – and bias in decision-making? How are we being more systematized, and how are we creating more transparency and understanding in our models and in our predictions that might otherwise be susceptible?

That to me is a deeper question, and my view is, regardless of which industry you take today, there are people making judgments. Those judgments are subject to all forms of human bias and, frankly, randomness and such idiosyncrasies.

There's, in my view, much more sort of risk there than there is in modeling and increasingly, just creating more transparency and understanding. And the fact that models and predictions drive more consistency and being able to shed a bright light on where that consistency is and what are the implications of that is a science that is fairly well advanced (in my experience today).

And so, for me, I actually believe that this is reducing bias. It's reducing risk. I worry more about the human judgment error and the bias and such associated with that, as long as we've got the appropriate guardrails in all things data science and understanding the implications.

Michael Krigsman: To what extent do you think about the ethical implications, meaning you were just talking about bias? Obviously, the decisions that any financial institution makes based on data can have a real impact on people's lives. Where does that aspect come into this?

Murli Buluswar: This is applications to healthcare, insurance, and a variety of other sectors. The question that I would challenge us to come back to is there are ethical implications today of human judgment and the biases that we all have as human beings.

Those biases have consequential impact, whether it's in real estate, whether it's in insurance, whether it's in healthcare, or pretty much any and every industry that you could think about. And so, for me, isolating – I wouldn't isolate the impact of data science or machine learning in asking this question because those biases have existed historically as long as humankind has existed.

The question is, how can we continue to create more transparency, objectivity, understanding, and discipline with models, which I think is vastly easier to do than trying to unwire the complex, human mind that makes decisions in ways that aren't always logical?

How to build a data-centric culture?

Michael Krigsman: Another dimension to this is the notion of building a data-centric culture, data-centric organization. Any thoughts on that aspect?

Murli Buluswar: It's a curiosity-centric culture. Data is nothing but a set of facts and information in service of curiosity, imagination, and questions that people are asking.

For me, oftentimes when we think about data-centric culture, it tends to be a little bit too much of a binary world. For me, it is very, very important for the sustainability of organizations and even for the sustainability of people in their individual careers to really continue to challenge their thinking about curiosity and imagination-centric work that they bring day-in-day-out.

Data is in service of that. It whets your appetite. But data unto itself can do nothing if we don't have curiosity and imagination.

But if we do have curiosity and imagination, data can serve that. It can dispel any hypotheses we have. It can reinforce hypotheses we can have. And it can help us make better decisions.

For me, the imperative for every large organization is to ask itself, "When and where are we curious and imaginative in challenging what we understand to be true in this world, and how will we up the game on that front tomorrow versus today?" because without curiosity and imagination (in my opinion) no meaningful innovation has ever been achieved by humankind.

Michael Krigsman: I keep coming back in my mind to the idea that, for most people, for most organizations, when they think about data, it's much more mechanistic. You think about it in a very creative way.

As I said before, it's like an artist or a painter thinks about it. Most organizations don't do that, and I think that there's a leap that many organizations would have to make to get to what you're describing. I don't think it'd be easy for many companies to do that.

Murli Buluswar: Large-scale change is hard for all of us as individuals and certainly large organizations as well.

You'll notice that when I talk about decision sciences, I almost don't use the word data because, at the end of the day, for me, it is about thinking about how we could make better decisions and how will that sort of serve us in our mission to be of more service to our customers and be a better run institution tomorrow than today. Backing that into, "Hey, data can actually enable that."

I think that if we talk about data-driven cultures, it can become a little bit intimidating because, all of a sudden, we can identify as we're not data and someone else's data, perhaps. But if we were to say, "Forget about data. Can I imagine these ten decisions that I make today as part of my role? Could I imagine how I might make those decisions differently tomorrow so that I'm not taking a mechanized approach to my role, but I'm blending my expertise and my understanding with a little bit more imagination and curiosity?"

If I can blend that, then data can play a very pivotal role in informing that. But to me, it actually starts with curiosity and challenging our imagination and understanding. That's why you see me (over and over, Michael) coming back to that theme because, for me, data unto itself is just a bunch of digits. It's numbers. It's a thing. it's not really inspiring unto itself. Nor is it (in a blindingly obvious way) driving something better.

It's the people and their creativity, their imagination, their curiosity, and their hypothesis-driven approach, informed by data, is where the magic happens.

Michael Krigsman: This is from Arsalan Khan. He's a regular listener. He always asks great questions. This is a little orthogonal, but it's an interesting, good question.

He says, "How can consumers protect themselves without even knowing that they're being judged by data?" He's responding to your point earlier about making judgments about data.

Murli Buluswar: Data or no data, every firm or every industry that we're a part of, firms are trying to sort of figure out how to be engaged with us as consumers. Those decisions are happening with or without data.

Don't think of it as judgment but think of it as are they being more relevant with you, for you. If they are not, shut down that relationship. Go find a partner who is demonstrating that radical empathy for who you are as a consumer.

How Citibank uses data analytics in banking to improve customer service, loyalty, and retention?

Michael Krigsman: Another question. This is now jumping back to LinkedIn. This is from Debanshu Debroy. He says, "Do you see any opportunity to use the learnings from CX analytics in training or aiding contact center representatives to better serve their customers?"

Murli Buluswar: Yes. Every time we interact with a customer through our call center, there's an emotion element. There's a human element. There's an interaction element. There's a transaction element. That is all being codified by structured and unstructured data.

That is a golden opportunity for us to understand how we did, what we could have done better, and how we might potentially reconnect with that customer (post that call) because you have all of the data that happened before the call, anything that happens after the call, and you have all the data from the call.

There's lots of pixie dust there, you know, sort of nuggets of gold in that. And I think every firm that interacts with customers through those channels should take it upon themselves to say, "What can we do better at this point in time?"

Michael Krigsman: Wayne Anderson says, "If curiosity and imagination are the key," he says, "you can teach knowledge, but how do you encourage imagination?"

Murli Buluswar: You can create a culture that is supportive and encouraging. Not everybody is going to be at the same point, but everybody is on their own journey. What's important is to encourage it, to reward it, to recognize it, and support it.

Michael Krigsman: This is from Mike Prest, who is a CIO. He says, "Do you believe that how you harness behavior, data and analytics today will help your organization become a change agent through digital disruption?" It's a really good question, the role of data and analytics in driving change and managing disruption.

Murli Buluswar: Yes, I think of data and analytics as a mitochondria that's going to drive digital engagement, understanding, and intelligence in a virtuous cycle in how we connect with customers, certainly both within Citi and within banking, but also beyond.

Michael Krigsman: What advice do you have for business leaders on driving successful outcomes with data?

Murli Buluswar: Start with asking three problems that you might like to solve differently that are material for you and imagine why they matter to you. Challenge your decision sciences team to think about how they could inform that solution or that problem through deeper data intelligence and create a virtuous cycle of interaction and connectivity that can build on itself and that allows you to set a higher bar for what you get, what you achieve, and what you accomplish with this team, through this team.

Michael Krigsman: My advice to anybody that just heard that answer is that if it's relevant for you, there's a lot that Murli just said that needs to be unpacked there. Unfortunately, we're out of time, so we're not going to be able to unpack it now, but you can replay the video and parse it.

With that, a huge thank you to Murli Buluswar from Citibank. Murli, thank you so much for taking time to be with us today.

Murli Buluswar: My pleasure, Michael. I appreciate you, I appreciate the audience, and this has been a delightful conversation. Wonderful to reconnect. Thank you.

Michael Krigsman: Everybody who is watching, thank you for taking your time to be with us today and especially those folks who asked such marvelous questions from the live audience. Before you go, please subscribe to our newsletter, hit the subscribe button at the top of our website, and subscribe to our YouTube channel so we can keep you notified about amazing, upcoming shows that we have. Check out CXOTalk.com.

Everybody, I hope you have a great week, and we will see you again soon. Have a nice day.