Analytics that power every customer conversation

Welcome. I'm Dave Edelman and I'm absolutely delighted to be here to host the fourth in a webinar series sponsored by Pega called Inside the Mind of the modern CMO. In these webinars, we have been talking with leaders who are taking the capabilities that are available now from AI, from digital analytics, from new, even new creative directions, and helping us understand how to put that all together to be a leader in this new environment who's driving substantial impact. And I have to say today, I am just incredibly delighted to have somebody who has done this in a number of places, not a CMO, but a leader in analytics across many different places. Murli has been head of analytics. He started at Capital One, he was at farmer's AIG, most recently at Citibank. And in what he's done, he's turned analytics not into a reporting function, but as a driver of real economic value. And in our prep discussion, as we as I started to learn more about what Murli has done, it just made me realize how so many organizations have incredible opportunity to elevate the leadership of their analytics teams. And so we're going to dive into that today. And in the spirit of the way we do discussions at the Harvard Business School, where I have been a teacher instead of just necessarily being straight.

Q and A, we use this thing called pastors where we take different topic areas and we go around in it and we explore what's possible. So first of all, merely thank you for joining us. Dave. Good morning. It's an absolute delight. Um, the best way I describe myself is I'm a CMO wannabe. So it's a real delight to actually be at this in this conversation with you. And I love your pastor's analogy because I'm actually vegan, so I love foraging the pastures. So here we go. I'm all excited to dive right into it with you. Thank you for the opportunity to share some perspectives with you. Sure. That's great. So I want to start at the highest level. Given how you have reshaped the way people think about the analytics function to be a driver of performance. And let's really explore what makes a modern analytics leader most effective. And, you know, one of the things you talked about in the prep is on one side, there are three traps you you cited to avoid. I'd love to hear more about those. And then that's what you avoid. But what do you actually do? Certainly, certainly. Um, so my understanding, uh, Dave, is that the average tenure for many of these analytics roles tends to be about two years. Um, and I think one of the reasons for that is, uh, people in these roles tend to fall into one or more of the following three traps.

I've mentioned the trap number one is this idea of being a pure follower, which is you ask me a question and I go answer it, and then you ask me a follow up question and I go excoriate the data and come back with an answer. Um, and it's, it's necessary, but insufficient in that it's not strategic, but it is tactically useful. That's trap number one. Trap number two is getting stuck in the world of beautiful lakes and clouds, and seeing our roles as being infrastructure enablers and infrastructure enablers only at the expense of being strategic and being more of a commercial force. And then the trap number three that I've seen people fall into is this this cycle of getting stuck in POCs, MVP's and pilots. I think it's important to have technical and commercial curiosity. On the other hand, if you don't have that operating, know how to think through, how are you going to translate that MVP or pilot or POC into full scale production? What needs to change and what needs to evolve in order to go from a concept to large scale value? Then you've got a problem. And so those are the three traps that I see. And these traps, I think, are even more urgent in the world of generative, energetic AI. And as I think about the broader AI ecosystem today, uh, the, the opportunity that I see is for people in these kinds of roles to bring that sort of perfect blend of, um, commercial imagination, technical skills, operating know how and the ability to lead cross-functional coordination and change to achieve a set of material outcomes that are relevant to a CXO or CEO, uh, that are measurable through the lens of a, uh, chief finance officer and that are attributable to the endeavors that you've set upon. So that last mile problem of being able to say, not only has this actually manifested itself in how and where and how decisions are being made.

But also there's a, uh, ability to quantify what the operating and or financial implications of that are. Yeah. It's so true. Murli and I've seen it in my own work. Um, when I was chief marketing officer at Aetna and when I worked with clients as well. The most effective pairings are a CMO with a senior analytics person. I do a lot of work helping people adopt agile practices, fast cycle test and learn. But it's not just setting up a test and getting the feedback, it's what are you testing in the first place, and what are you actually even thinking about going after? And when you have a senior analytics person paired with a marketing person at the top leading, you get smarter decisions about what to do as well as more effective. Whereas other models, you just have an analytics person on the team who, as you say, just does what they're asked to. Um, you know, sets up test cells, generates results, but that's insufficient. You've got to have analytics at the top, helping to set the strategy. I assume that's something you've always set up and it's part of your guiding principles. It's certainly something that I've got an enormous amount of, uh, passion for, which is that I think all senior executives need to be leaning into how are they nurturing their commercial imagination? How are they building a healthy level of technical aptitude? Even if they aren't technicians, how do they connect that curiosity in the power of data and AI driven intelligence into, uh, operating decisions? And how do you orchestrate that across functions? So my mantra, one of my mantras has been that in order for functions such as this to be super very effective.

You have to have a follow partner lead approach, which is this ability to take the lead of other functional stakeholders and be guided by them in terms of what their needs are. Uh, this ability to co-create or co define problems and opportunities and solve them in conjunction with other stakeholders. And also this ability to conceptualize opportunities that others might not be conceptualizing, uh, and develop that, uh, in, in, in a little bit of an, in your own little sort of, uh, space, if you will. And then as it gets a little bit more of a form factor, bring it to your functional partners and stakeholders in ways that is tackling problems that they see, uh, but solutions that they don't have and questions that they might not be asking in terms of how those issues could be tackled with much greater creativity and technical sophistication. Yeah. And when we were talking, Murli, a great example of essentially shining a spotlight on something that people aren't thinking enough about leads us to a second pasture, which is your observation that people overweight emphasis on acquisition, but there's also a ton of opportunity in customer engagement. And now with the tools of AI, you can get a lot more data on engagement on customer journeys and start thinking about how to derive more value from making that something that's more seamless and more personalized. Tell us how you came to that observation and how you've seen it play out in your work.

Certainly, at some level, if you think about that observation, it's fairly self-evident in that we've all been talking about customer experience and customer engagement for decades at a stretch. The challenge that I see is that quite a bit of what we think of when we think about customer engagement and experience is soft metrics around net promoter scores and things of that nature. Those are helpful directionally. But the challenge with NPS as a single metric, not that it isn't useful, but is an isolated metric, is that it is based on sampling. It is biased because it requires respondents to self-select. Um, and it's not massively actionable because it's asking for a sentiment but doesn't actually tie to individual journeys and what you could be doing differently. And the way I would frame it is the role of any CMO, uh, is to be thinking of how could they play a pivotal role in nurturing sustainable, Mutually economically viable relationships with their customers. That is the mission statement. And frankly, that's a mission statement for the enterprise as a whole. Right? And it has to be mutually economically valuable. And it has to be sustainable not just at a point in time. And what does it take to be able to nurture that? That's the holy grail in my view. And I don't know that we actually think about that question in its deepest form, the way we should be or could be, but yet we live in a world today where the tools have evolved so phenomenally that they do allow us to live that promise in its broadest sense.

And so I think the, the, the, the, the observation that I have is that we all run, we've all been in roles in firms that have leaky buckets where customers disengage, whether they choose to end their relationship formally or their other forms of disengagement, that manifests itself in revenue. And we almost accept that as a baseline and then say, gosh, we need to go out and attract new customers in order to not only countervail the leak in the bucket, but hopefully to drive incremental two, three, four, 5% growth. And that's the game. The challenge with that is the customers that you've had with you for a longer period of time, they're in a much more flatter part of the loyalty curve. Um, and if you were in the risk or lending business, then their profile is even more superior. They tend to be much more digital. And so if you're able to continue to nurture that loyalty with them, the NPV, the net present value or the customer lifetime value of that, however you want to measure it is vastly superior to attracting somebody new. That is just getting started in their relationship with your brand. And and so how do you break that down? The way you can actually break that down, in my view, is then say, let's not just live in the averages and accept that leaky bucket, but through the power of science. Let's have that commercial curiosity to ask the question, when and why does that leak happen at a micro segment level? And that can actually be broken down into.

It happens because people's life circumstances change. It happens because they've gotten better offers from competitors. And largely speaking, those two are extraneous because they're not they're not easily influenceable by the, uh, by the existing, uh, relationship, uh, the firm. And then there's this third bucket of there's some level of, uh, tacit or overt disenchantment that you have with that particular brand because of some non positive experience. And that to me is very much within the bailiwick of what you could actually lean in to understand and engage with the customers in super sophisticated way through the power of whether you want to call it machine learning or agentic AI to be able to, uh, understand when that is happening and, and how you could actually claw back that, uh, disaffection that they might be feeling in that moment. And in my view, that sort of component of it is probably in the rough order of magnitude, at least a third to 50% of the overall leakage that you see in the bucket on an annual basis. And by the way, that's just from a customer standpoint, again, that does not take into account the embedded economic value because these customers are more digital and when you do actually save them, they stay with you longer. And so the economic value of that is vastly superior.

I can actually unpack unpack that a couple more layers if that's helpful. But I want to pause for a moment. Well, one way I'd like to to unpack that merely is there's two things you're actually talking through as you're doing it. One is the analytics to understand why people leave when people leave, why people leave. And there's an analytic that informs things. But there's another thing you're talking about with this, which is an analytics engine that spots it in the moment. Yes. For an individual, yes. Which requires action. Yes. And I'd like to go on that vein and dig in there and actually understand how you set that up. How you've. Because it's one thing to find out in the moment. The other thing is to have an engine that actually acts on it. And how do you know? How do you link analytics into, you know, a fairly again, overusing the word agile, but an agile mobilization to spot and act on a trigger indeed. So this, this, this, this phrase of hyper personalization has been around forever. It's been bandied around in ways that frankly, um, it makes it sort of lose its meaning because while we use those words, I think very, very few industries and or firms have been able to figure out how to live that promise. And the way I think about that problem statement, Dave, is number one, uh, is can I get a real time temperature check inference on every customer and how they feel about their relationship with me at any point in time.

Uh, that's number one. Number two is can I understand when somebody has some level of friction depending on who they are as a customer and what that friction is? Can I actually discern the intensity of that pain and the financial consequences of that? If I put those two together now, what I have is an always on measurement that helps me understand how a customer feels about me at any point in time, and how that inference of how the customer feels about me can actually be tied to future financials, disengagement, revenue and Ebit, and of course, the economic embedded value. And I have the ability then to say, I also know why this happened. Mhm. So you bring those three together and you have sort of a level of magical insights that I think have been a little bit intermittent in many industries. And then the question is, I actually haven't had this level of intelligence granularly at that individual customer at a point in time. The way I do now, what do I do with it? And of course, you've got the power of AI tools that can now actually translate that into saying, we can actually have real time engagement in a very dynamic test and learn approach, uh, where the AI agent is now starting to interact with the customer and knows what not only what conversation to have, but over time learns, uh, what action could be taken in order to mitigate that risk.

And what is the financial risk at stake based on which customer was affected? How much were they affected? And what are the financial implications of the risks of that? So to me, that is hyper intelligence that allows us to move beyond soft measures of customer engagement and customer experience, which still have their place to bringing a whole new level of science that is highly actionable and measurable. And I think the more firms can do that, the more they'll have the epiphany that a much greater portion of their dollars should be in service of ensuring that they are very surgically driven by mathematics and enabled by AI. Engaging with customers in ways that would actually enhance that mutual loyalty. And there's much more value there to be extracted and to be preserved than investing hundreds of millions, if not billions of dollars in acquiring new customers. It's not that it's an either or, but that mix shift in service of this. If you are working with a finite pool of dollars. I think it's phenomenally more impactful on a whole different stratosphere than chasing acquisition. Yeah, I totally agree and have talked about this in terms of really thinking through that personalization. I want to take what you said, Marley, and actually break down because you went through a whole bunch of steps, each of which have different AI agent capabilities associated.

I just want to make sure everybody understands there's one part which is going through the whole base, understanding some kind of score in terms of their happiness with the brand, their intensity of, you know, friction, effort, any kind of negative, however you want to define it. Then there's also a second engine that's actually spotting moments with individuals, creating triggers that merely had something that wasn't good happen and therefore is vulnerable to attrition. Then there's another engine, which is what do we do about it? Um, and so there's another engine that goes through and actually says based on situations like this before, based on what we know about merely because it could be different how to merely versus Dave, based on what we know about merely, here's what we should do. Then another engine to generate that. Um, it could be prompts for the call center in the moment. It could be follow up content, email, text, whatever offers on the backside. And then it takes whatever has actually happened. There's another engine that gathers all that intelligence and optimizes things. And along the way, you might even be setting up split tests to try to learn more. So there's all there's about five different engines. You talk through and I just because that doesn't happen all in one glob. It's different. It's discrete things that as an analytics leader, you have to help people understand.

Each of those things have data requirements, operational requirements along the way. Do I have that right? Marley 100%. Dave. And this sort of ties back to my core thesis that I was referencing earlier, which is the, the, the gap that exists in the world today is less of a technological gap. It's less of a gap in the capabilities of language models and machine learning and other sort of AI, semi-supervised and supervised techniques and such. The gap that exists is this combination of commercial imagination asking the fundamental question and challenging your assumptions and not just being tethered to how things have been for the last decade or two decades or what have you, but reimagining how the world could be and should be with the first principles approach. Then there is having a healthy level of technical aptitude to be able to say, from a business analysis, from a statistical sciences, and from a generative and a genetic standpoint, how do I orchestrate this? Then there is the operating know how skill set and mindset of saying, all right, so if I do those first two, how do I embed this into either existing or brand new workflows that I create? How are people going to be interacting with this? And what is my measurement? How do I ensure that my AI engine has a super sophisticated test and learn capability, both in terms of what conversation to have, how to have the conversation, and what call it offer to make peace offering, if you would, to the customer in service of renewing that relationship and salvaging some disservice that you might have done inadvertently.

So all of those and then those three capabilities sit on the foundation of being able to orchestrate cross-functional coordination, where typically in many large institutions, uh, the, the metrics of success that you have in individual functions can be a bit too narrow. So going back to your previous comment, it's, I almost think of it as we're trying to put together a thousand piece jigsaw puzzle and let's say there's ten people and maybe each of them has 100 pieces. And so you've got to know what piece you have and how it could work. At the same time, you've got to be able to collectively step back and imagine what that complete picture should look like. And so for me, being effective is about having this ability to zoom in into the, uh, gritty, granular technical and operating details and at the same time being able to step back and sort of see the connectivity and imagine the connectivity in service of what you want to be able to create. So it all for me, Dave starts with a, uh, commercial, uh, sort of a curiosity, I call it, you know, sort of the, it's not, uh, intelligence quotient in my parlance, it's imagination quotient. It is. And then though very important is the operating model around you, as you said, the cross-functional team and how they're set up. and, you know, maybe ten people are too many these days. Um, too much hyper specialization.

And because of the AI capabilities, each individual can have a wider swing. And you can have, you know, I always say it shouldn't take 13 people, 13 weeks to get a marketing program out the door. Have three teams of 5 or 6 doing it every week. Um, and that's what we should be going to with these tools. So it is a change process. It's a change process, as you say, in terms of how you think as a leader and your willingness to step up and actually lead creatively, even though this is about analytics, it's left and right brain coming, and then it's the operating model. There was something you said actually also Murli, when we talked about that change process because, okay, you need that, but how do you get there? Um, and you talked about three things a mountain, a monastery and a metropolis. And that was it. Was it? Very intriguing. And I'd love for you to share that with the audience. Thank you. Um, the first thing I have to say is, um, even though I, uh, share that freely with many, um, it's not something that I created, so I'm shamelessly borrowing it, uh, slash stealing it from, uh, someone else that taught me about it who may not, who may also not have actually created it themselves. But the principle behind it really stuck with me, which is if you're trying to drive large scale, uh, disruptive innovation in a particular area, function or process, uh, in a large institution, um, the three M, uh, approach can actually serve you in good stead.

The concept is that you go to the proverbial mountain, you be your toughest critic and you play with the idea and you think about how could you actually take a counterpoint and sort of challenge your own understanding or your own thesis, uh, be your toughest, uh, self sort of critic and put some form factor to that based on what you understand about other functional areas and how they might actually see, uh, how your, uh, identifying, uh, shaping and wanting to tackle the problem. So that's sort of the first step. The second M is, uh, monastery and the thesis there is, uh, once you've actually, uh, built your thinking to some level and put some form factor to it, uh, you should then bring it to a close cadre of your colleagues, ideally that bring orthogonal perspectives to this, uh, that care about you as an individual, care about the firm, uh, and care about the broader collective mission. and therefore will be as objective as they can. Uh, but yet also speak their mind and, and provide, uh, tough criticism and, and not suffer from groupthink in a way that can actually build greater strength and precision and adaptability to the solution that you would like to actually develop. And if you do those two, then you're setting yourself up to go to the metropolis, which is this ability to sustain pressures, challenge us from a broader range of colleagues and functions.

Uh, perhaps more often than not with the best of intentions, uh, but that you've actually anticipated the challenges that people might have or the questions that they might have. And you're weaving that sort of potential, uh, challenge into your thinking into how you've developed this so that it can actually sustain the vagaries of all of the forces of nature. And I guess human, uh, that could put this concept at risk. So that's the three M principle. And for me, one of the quotes that sticks in my mind quite often when I think about this, Dave, is F Scott Fitzgerald and I'm synthesizing this, which is the idea that true intelligence is the ability to hold competing ideas in your mind at the same time, and yet function effectively. And to me, that ties very closely to the imagination quotient concept that I was referencing. You know, where I see that playing out so tangibly merely is when companies are shifting from a product perspective to a customer perspective. Um, especially companies that have multiple products because from a customer's perspective, hitting them with multiple products is often a mess. But if you want to move to a customer perspective, you have to hold multiple things in your mind. You have to challenge yourself.

What really makes sense is it in the best interest to try to find some way of coordinating things? Oh, but that adds bureaucracy. How could that work? You've got to bring it to the business, figure out how to do it, and then try to make that happen. You know, I saw that I was chief marketing officer at Aetna. People get health insurance. They're onboarded as a new member. They get hit with all kinds of stuff from all the different divisions. It's totally confusing. And we had to rationalize it. And without knowing that framework merely, I had to go through those steps of really challenging why? Why does this happen? Is there other ways of doing it? You know, we ended up coming up with personalized videos that actually could coordinate everything that merely would have as a member. But it took a lot of organization. The data was all there. The tools were there, but working through the organization and having those tough conversations that we want to move from just selling products to creating a solution for the customer, which is what AI is really about. It takes that kind of introspection first and then interaction. Indeed, I almost, uh, in my view, the concept of the three M's has existed for a very long time. You and many other leaders have actually lived it, uh, without necessarily, uh, putting that sort of form factor to it. And I think at the crux of it, the question that in my view, executives and senior leaders should be asking themselves is, am I task taskadmin or am I mission driven? And of course, people would say I am mission driven at all points in time. That's the right answer.

But then that would beg the question, how are you spending your time and energy and setting goals and guiding your teams in service of being mission oriented versus being task oriented? I'm going to read to you my mission statement for my team, if I may, because I do. Yeah. I was just going to ask you, how do you turn that into something tangible? Go ahead. So I'm literally, I keep this laminated card on my computer and I'm going to read it verbatim. Uh, who are we? We are bold thought leaders, driven by outcomes, guided by curiosity, data, and learning. That's our mission. And then then there's a set of principles and values associated with that. Uh, and then there's a set of not sort of granular metrics, but how we define success in, in a reasonably, uh, in a reasonably concrete way that I think is important. And in the, in the background, I've got this image of rowers in eights. And the reason why I think that's a very good metaphor is that rowing is the ultimate sort of team coordination sport. Uh, there is zero room for individual brilliance because if you are individually brilliant relative to your teammates, you're actually doing disservice to the team mission. Uh, whereas in many other sports, I think you can actually counter any weaknesses in your team. So there you have it.

Yes. And though to the team, you can be creative outside the actual rowing to say, why don't we try this? Why don't we try that in order to get better? Because you're noticing what's working and what's not working. It's a constant dynamic. Well, Merlin, this has been amazing. Um, I love the mindset that you bring of transforming how analytic people should just think internally about their purpose and their mission and how to bring that to their teams to have impact. Of course, it takes a village. It's not just what one person as an analytics can do. And so you've got to think about the metropolis, um, or the village that you're trying to, to mobilize and bring together. But it certainly takes a very different kind of mindset. And, you know, that's why this is called in the mind of the modern CMO or CEO in your in your case. But I think as a CMO who's worked with CEOs, it takes that kind of mindset too. So thank you. That was just really, really inspiring. So I hope everybody took a lot of notes on this. I think there's much to share and discuss with your teams in terms of how you and they as individuals are thinking about analytics and the role with the company from both the right and the left side of the brain, the creativity of the right side with the analytics of the left and bringing that to bear. So I want to also thank Pegah, our sponsor, for this discussion. Um, this was the fourth in our series and I hope all of you will join us in Pegah world in June this year in Las Vegas. Thank you very much.