| If you don't have an account, Register here |
Moderator:
Reetika Fleming, Executive Research Leader, HFS
Panelists:
You can listen above or watch this HFS Videocast here:
Session Description:
Despite the dramatic recognition of data as the new currency or the new oil for global business, day-to-day business operations are severely underusing it. To design a data-driven, autonomous enterprise, business leaders must understand the data their enterprise needs to be successful, including understanding the digital detail of processes and interactions.
We’ll hear enterprise leaders across business functions address these questions:
You can also listen on:
• Apple Podcasts
• Spotify
• Google Podcasts
• Anchor
This transcript was auto-generated from the original recording and lightly edited for readability. We've done our best to catch errors, but names, technical terms, and company references may be misspelled or imperfectly captured. For the definitive version, please refer to the original audio or video. Views expressed are the speakers' own.
So I’m delighted to have a great panel of experts here to have this discussion about how data can really power this entire concept of the autonomous enterprise, or generative enterprise. All right, so if you guys will join me on stage, let’s start with Kevin Campbell, CEO at Syniti, and his purple vest today. Aishwarya Gupta, General Manager and Practice Head for AI Strategy and Solutions at Wipro. We’ve got Bets Lillo, Executive in Residence at Texas Christian University and board member at River Logic — we are in our monochrome coordinated vests also. We’ve got Amresh Mathur, who’s held multiple ecommerce and CX roles in firms including Samsung and Citizens Bank. Please join us on stage. And last but not least, we’ve got Abhishek Mittal, VP of Data Analytics and OE at Walmart. All right, thank you for joining us. So, to refresh your memories a little bit — I know it’s been a heavy morning already, a lot of discussion — Phil walked us through the core principles of the autonomous enterprise, and it’s pretty fundamental that data is very much at the core of this. I might be biased, I cover this space, but we’re seeing large enterprises seeking to radically rethink how business gets done, how their systems interact with employees, partners, and customers, and the way in which data moves in your organization really enables or inhibits all of that. Phil also showed you this slide today about what’s holding back companies from achieving their big strategic objectives. He talked through a number of issues, but I just want to point out that 3 out of the top 5 challenges are data challenges — it’s about data quality not really being as high as we need it to be, data governance, poor automation of processes and data. So we’ve known for a while now that we need to fix our data, and that’s what can possibly help us design our future with autonomous or generative enterprises. But how do we get there? So let’s quiz our panel. If time allows, I’m going to ask for some Q&A maybe midway through the session — let’s not leave it to the end, so start thinking. So we’ll start with, let’s see, Bets. Let’s talk about what makes data good.
You know, we think about data as if it exists for its own reason, but what really makes data good is its ability to be used in decision making. And one of the things we heard from the last panel is that what we’ve got with autonomous capabilities is the ability to reuse information that already exists. And what really enables that is an understanding of data. So I would argue that data itself, without that understanding, is useless.
And I completely agree with that. Additionally, if you look at what makes data good, it’s completeness, it’s orderliness, getting out of the silos, and consistency — with the help of governance and data quality, because data quality today is the most challenging task for organizations to overcome. And data quality comes with its ability to move around, out of the silos, with the consistency of flowing from one section to the other without any issues. Even if the data has been shared and it is not in order, it may not give you the relevant outcomes. So definitely these five areas come together to make the data good.
All right. Well, the autonomous enterprise principles start with leadership really needing to understand what is the data they actually need to win in their markets and their business. Amresh, what have you found in really knowing what is the good data that I need?
Yeah. So from my perspective, good data is something that can be accurate, because then you’re able to take a decision for the organization. And then, most importantly, can it be reliable? Because, like they said, quality is a challenge, so to take an informed decision it needs to be reliable. And these days, more and more, it needs to be relevant, because fortunately or unfortunately the data has moved faster than the humans have, so we need to make sure it’s relevant because of the explosion in data. And last, I think, is timeliness — you’re able to take a decision from a timeliness perspective, with real-time information, so that you’re able to imbibe all this data and do what’s good for the organization. That’s the way I see it.
I would like to add that while we’re looking at what makes data good, there was a lot of hype around synthetic data generation. We’re looking at accommodating data where we do not have good synthetic data, and it has to be carefully driven — setting up some benchmarking around how much synthetic data is going to be generated, where we are able to differentiate between the real data and the synthetic data. It should not be filling in all the gaps to the point where it becomes very difficult to differentiate between the two, and the quality of the data will be jeopardized because we are generating what we want. It’s not going to give us the right kind of analysis from the real data.
Yep. So you need to keep confirmation bias very much in mind.
And actually, one of the things that struck me in one of Phil’s earliest slides — he highlighted that the two biggest enterprise concerns are cybersecurity and supply chain disruption. Those are enterprise concerns because they’re unpredictable and they rely on real events. So when we think about the data that we already understand how to manage in our organization, that’s predictable, and it’s historical, and it’s real. So this ability to understand what it is we’re looking for, without augmenting the data in a way that takes us past that understanding, is huge.
I’ll just add, we did a survey with HFS earlier this year that’s been out there. It was C-suite executives at enterprise companies, and 95% of them said their companies would be more competitive — better customer service, better decision making, more flexible — 95% said their companies would be all that if they had two times the data quality. And we joked when we came up here, we were going to say to that other panel we just heard, that talked about great things: if the data is no good, none of that works. So this is the fundamental piece. It’s the building block for everything. Yeah, absolutely.
OK, so Abhishek, let me kick you off on this one. Does the org structure, as it exists today in many large organizations, get in the way of us making good use of data?
All structures are never perfect, right? If you’re centralized, then you’re away from the customer, you’re focused maybe too much on the capabilities. If you’re decentralized, you’re close to the customer but you’re subscale. And if you have a hybrid, then a lot of time is spent just figuring out the allocation math and alignment. So structures are never perfect. The way I see it is like running labs: you focus on the use cases, you focus on the customer, but every time you solve a use case, if you can invest in building a foundational data capability, I think that will serve you better, because that is the balance. If you focus only on capabilities, then sometimes you’re doing tech-driven implementations — MDM solutions that take five years, and there’s no value, and the organization runs out of patience. Or, if you’re focused on just solving one problem after the other, you’re not really having the biggest impact of data and you’re not investing in foundational data capabilities. So focus on the reuse of the capabilities that you’re building. I think that’s the way to solve it. All structures are never perfect.
Bets, I think you had a view on that.
Yeah, I do have a view on that, because when we think historically about how technology has grown up, it’s grown up as a vehicle for helping us do functional activities better. And I think Tony, on the prior panel, got it absolutely right: it’s figuring out how to optimize something we know how to do. But one of the real challenges is how to work across those functions, because those are the areas where we need to make business decisions that we don’t yet understand. So this is really the conundrum around data. If I’m dealing with functional data, I know how to make it look good, because I’ve got an organization that confirms the data looks good. If I’m taking data across functions — and we see this with River Logic, where we started with supply chain network optimization, and now one of the things we’re doing is matching financial and operational data for ESG — we’re now mapping data quality across different functions. So figuring out what good looks like is really challenging, and I think the organization absolutely gets in its own way of making good decisions. So figuring out how to navigate those functional boundaries will be huge.
Kevin, how do you advise clients on this?
So the org structure makes a big difference. When we talk about governance, which is where you get into the data, I always tell C-level executives — if I’m talking to the CEO — I say, are you ready to have the organizational discussion? And they’re like, what does governance have to do with organization? And I say, just wait. Because as soon as you try to change how the data flow works, as soon as you try to say who owns what, every organizational discussion you’ve had for the last 15 years will come out. You want to see a food fight in the cafeteria? Start with that, because it’s the most valuable. So I think most companies have realized we don’t just give it to IT and ask IT to manage data, and we don’t put a CDO in place who’s an ex-IT person focused only on what we need to do. What we need to do is figure out what’s the most important data in the organization, and then what’s the most efficient way to do it. That’s why we always say, make sure you’re mature, make sure you’re ready to have those conversations. Start, as Phil said, with what’s the data that’s most important, and then just walk through it, because most CEOs will say to me, how hard can this be? And I’m like, OK, let’s just start with the most important data and walk around who touches it, who owns creation, maintenance, and all the other stuff of that data. So it’s fundamental blocking and tackling, but it’ll tell you a lot — and we get what we measure.
Yeah, absolutely. I think this org structure topic is very central to the design of an autonomous enterprise, and I think we’re going to get into that topic a little bit later, but it’s also important to tie it to data ownership and access. All right, next: is upskilling actually working? What else can we be doing to help people work with data more intuitively and better? We had a bit of a conversation on prompt engineering — is that really the future? Maybe, Abhishek, you want to kick us off on that?
Yeah, this is a very passionate topic for me, because I have a team of domain experts and data scientists. The domain experts are people who are lawyers, compliance professionals, or our blue-collar fulfillment workers working in our fulfillment centers. And then I have these data scientists who are talking about machine learning and AI. So I think what works, from my experience, is three things. Number one is the centralized training and learning programs — we have them, every company has them. They create awareness, and for the motivated few they’re a great opportunity, because they take those opportunities and upskill themselves. But a vast silent majority don’t, because they’re busy in their day to day; it’s unlikely they’ll take time out to say, OK, I need to learn Python or Tableau or BI or whatever it is. The way we’ve solved that problem is on two fronts. One is learning on the job: as part of transformation of our use cases, every time we’ve had an opportunity to look at a process, we’ve tried to make sure the domain experts lead the charge. I have a really good story here. One of our expert-service solutions, where lawyers provide value to other providers, started as an expert service and turned into a data service. And believe it or not, about 80% of the new data service is managed by these lawyers, because they became the users of the AI — they understood how to make value out of it. So when we started selling that AI model as a service, it’s those experts who are actually helping other experts in our clients’ companies, and they’re leading the charge. So that’s the learning-on-the-job piece. And then lastly, we as leaders also have to rethink our job structures. There’s a lot of low-code and generative technology coming in, but how do we create jobs where these low-code workbenches can really be used to create value? That’s the third thing that’s really made sure there’s upskilling — by creating jobs that may not be a data scientist, but where somebody can start with maintenance of the work, giving them a framework they can use to codify their expertise. So those are the three things that have worked for us.
That’s great. Aishwarya?
Phil talked about the empowerment of the employees as one of the major areas where leaders have to pay attention. When you’re looking at upskilling of the data competencies, yes, to a certain extent we’re able to provide the curriculum and understand what kind of skills are changing and that they have to adapt, but it becomes very important to attach a purpose to it — why we’re asking them to upskill. Empowerment is one of the areas which will help employees to look at it as: I am empowered to do this job, which will help my end client and bring business value. Now I have to look out for what skills are needed to deliver this job. Automatically it’s a driving function which encourages employees to go and upskill, because now you have empowered them to deliver and be responsible for the outcome, and the rewards you get in turn of that business value we’re adding to the organization. I think if we’re bringing this intuitively across the organization and putting up that vision, it will automatically start pushing and fueling the channel for upskilling, and you will no longer continue to push people without telling them why they need to upskill. That’s where I see a paradigm shift coming in at the job levels where low-code, no-code kind of jobs, which Abhishek just talked about, sometimes they’re not able to attach to this and they say, why am I wanted to upskill to this, because ultimately I’m able to deliver my job, whatever you’re asking me. So I think that could be a separate thing we can do intuitively. At the same time, if we’re enabling people intuitively to work with the data, it’s very important to tell them about the data-driven culture change, which is bringing in the insights of the past 50 years and the revolutions you’re seeing. Automatically it drives you that upskilling is one engine which has to be continuously fueled. And that is why they are required to intuitively work with the data, which will make them smarter, more productive, with faster turnaround time to delivery, rather than doing the work the brute-force way. So the moment they understand and learn how to play with the data, it ultimately gives them the empowerment we talked about — yes, I’m seeing the insights coming from this data, and this has to be fed into the channel of talent optimization. At the same time, looking at the previous, connected question: if you’re putting the right data functions in the organization, they are responsible not only to drive how the data flows; it’s very important to look at the upskilling path and the competency, because they all go hand in hand. You can’t be only looking at the technology side of data governance and data accumulation. I think what CP talked about in the previous session, it’s the animal kingdom — you’ve got to be continuously looking at how the forest is running the infrastructure around it. It’s not only the lion who gets the data; we also need to empower the monkeys and the deer to collect the data, get them inside, and learn how to make the forest a better place.
Can I actually ask you all a question? What do you all think? I think that maybe upskilling is the same as new tools in a different costume. And if we think instead, how do we build and create the ability to understand? In our last panel, the topic of an ecosystem versus an enterprise came up, and I think fundamentally what that is, is understanding context, in the same way that Kevin was talking about. All of the leadership capabilities are irrelevant if you don’t have the quality data underneath that gives you the ability to accurately understand the world as it is. So I’d almost argue that skills and tools are the same thing — one applied to people, one applied to technology — and they’re both useless without understanding.
Yes, and I would just add, really quick, that we own some of this problem as leaders and as an industry, because it’s no different. The analogy I like to use is, when I get my kids’ sports teams out there, what’s the first thing the kids want to do? They want to practice, if it’s American football, the flea flicker; if it’s soccer, they want to practice ripping the jersey off and being the one who scores. Well, that’s the end of the process, not the beginning, and we don’t focus enough on the fundamentals. What is data? What do I use it for? Even look at the college curriculums — now every college is going to have a ChatGPT, how-to-use-it course, but do people have the basics of data and how we do it? Are we teaching our people the basics and the fundamentals, and the fact that if you get bad data, good decisions with bad data are still bad decisions?
Yeah, so it feels like we need to do both: continue to empower employees as best we can and upskill them on general data proficiencies, but at the same time develop systems and workflows that bring them the data they need, so they don’t have to be data scientists.
Yeah, and I would like to add a last point: ultimately you are making people more cognitive, because they have the data at their disposal to take better decisions. So the moment they understand they’re becoming smarter, more cognitive, if they’re using the data more intuitively — I think that’s where everybody wants to become smarter.
Yeah. OK, so I have a couple more questions, but before that — anything from the audience? Just put your hand up and we’ve got mics going around. Yep, there.
So I was just thinking, listening to the discussion around data, back to my days as an auditor. Nowadays, they’re hiring CS kids to do the work of auditors, and the challenge I see, now that I’m far removed, is that if you don’t understand the data — what it’s supposed to be and what right looks like — it’s really difficult to let the tools do that, because if you don’t understand it from the foundational pieces, bringing in that person at the higher level doesn’t really do anything, because they don’t know what’s wrong. They just see the data and say, OK, here’s the data. So the problem I’m seeing today is that people are bringing data forward and doing analysis on it, but the fundamental data is wrong. So that’s just one observation I’ve had, and the panel, I think, is really bringing that to the forefront. Thank you.
That’s absolutely true, and I’ll just say, as an educator as well as a board director, my colleagues and I take very different points of view on the use of ChatGPT and generative AI in the classroom. I’m perfectly fine if my students want to do their homework using ChatGPT, because I have them write their exams with pencil and paper. So if it helps them learn the information, to stand on a scaffolding of technology, great; if it doesn’t, no harm, no foul. But when it comes down to them sitting down and talking about how they analyze a case, then they’ll be doing it with what’s up here.
And at the same time, if you look at ChatGPT responses — I’ve analyzed this carefully, because I have a teenage daughter who uses ChatGPT for her assignments — you’ll get the answers to the static questions, the history and so on, but ultimately you have to put your cognitive ability to analyze and take the right decisions. So the application, the sympathy, the empathy part which you bring into your answers, and ultimately what decision you should be taking in that situation, cannot be answered by ChatGPT. So definitely it’s a combination: you can use it for getting more knowledge and information, but the cognitive decisions and the final outcome of presenting and applying it in the circumstances, I think that is a personal learning that humans have to do.
So we’re actually already on this topic of culture, right? Because this is all about cultural change. I’d love your perspectives on what’s been working for you.
I think, from a data-driven culture perspective — like Phil said on stage earlier in the day, when he had all the leaders and CEOs — the culture starts right from them. They’re the ones who drill down how the culture across the organization has to be, and every one of them resonated with the fact that in their organizations they’re actively thinking about making it data-driven. So it starts from the top; that’s the first thing. The second thing that’s happening is that in the matrix you just talked about — the verticals and horizontals in the organization — everybody had an intersection point, and because they were so siloed, everybody had their own way of looking at data. So what’s happening now is organizations are evolving; we heard from the leaders that they’re trying to bring in all these data owners, and the owners work with every business unit to identify how the information they have is going to be a valid use case. That’s helping drive a very data-driven culture. The other thing we heard this morning is that, like Abhishek was saying — he has a set of data scientists and a set of lawyers — what’s happening in these data-driven cultures is that business is coming closer to technology and technology is coming closer to business. So no longer are the data scientists sitting in silos just making their theorems and statistical models; they’re actually interacting very closely with business to see how to make their models — whether it’s prediction, personalization, or anything — more relevant to the business, so that together they build what is the future of the organization. And lastly, what I’m seeing is that as these silos work towards having a common product owner or data owner, they’re also trying to co-create a roadmap — like Steve Jobs talked about a 15-year roadmap. They’re able to quantify the use cases, and even the uncertainty: while the data is used to say what’s happening and the CEO decides what to do, there’s also an uncertainty associated with every decision. Now, in the culture, they understand that we’re not 100% accurate, so they’re using the same data to predict the uncertainty as well. So together they’re not only making useful use cases, they’re also saying what we cannot handle with the data and how we predict the uncertainty in those worlds. That’s how I think the whole organization is becoming a data-driven culture.
You know, I’m guessing that when you were doing your auditing, one of the things you did — way beyond looking at the data in the accounting and business systems — is you looked at the artifacts of the organization. Were people parking in their assigned parking spots? Were they cutting in line in the cafeteria? And I think to really get at a data-driven culture, we have to take that very same approach. We have to look well beyond the specifics of the very narrow attributes of the individual components of data, and look at the ecosystem that surrounds it: is it helping the data to get better, or is it helping the data to stay the same or get worse?
One thing I want to add to what Amresh said is that sometimes the CEO really matters, and the top matters, but I think we also need to go back and show the value of what that culture has created. We’re good at creating these programs, but with any data program there’s always the question: OK, did this analytical team really create value, and how did it really help the operations? At the big level we know, but I think it’s important to look at what decisions were made and what impact those had, and showcase that, because then it creates the awareness that, OK, I need to do this in my area as well. In our company there are pockets which are advanced and then some that are not, and one of the things we’re trying to do is concerted awareness on where it’s working — talk more about it, showcase it — so that creates the culture sort of automatically.
Yeah, so a lot of sharing and replication where you can.
But at the same time, while we’re talking about driving the data more and more, we have to be very cognizant about responsibly sharing the data, because there’s a lot of GDPR, and those kinds of regulations are going to be in place. Sometimes, while you’re driving that data-driven culture, you have to put those checkpoints in place to ensure it’s driving and flowing responsibly.
Yeah. Kevin, I was just going to say, weaving together what we’ve heard here, the previous question, and the question we got from the audience — context is so important.
So I do think we’re missing something if we say we’re trying to have a data organization — and I know all of us have tried this. Remember, the fundamental thing is we’ve got to elevate everybody’s knowledge of data. When we look at it at Syniti, how we try to grow people: we take people equally from a data or more technical background and from a functional background, because in the end they need both. And that was your point about context. If all we do is create great data people, how are they ever going to have the fundamentals and the context? Because what makes fantastic data people is people who actually understand the context of what the data is being used in, so they can make the right judgments. And ultimately, we’ve always got to remember — my hashtag is always business benefits — we’re doing this all for a reason. I had an old boss who, every time a new technology came up, would always say, Kevin, as soon as it can pick up my freaking laundry, then let me know. And it was a skeptical way to look at new technology — is there a practical application for it? We’ll probably get there too, though.
Yeah, I’m hoping, I’m hoping. Any other questions? I think I saw a hand up earlier. No? OK, we’ll keep it moving. Actually, I think we only have a few minutes left, so, final question. Let’s start from Kevin and work our way all the way down. Can we ever be done?
The answer is no. Like I said with my kids, the first question when you get in the car, before you even go the first mile, is, are we there yet? So that’s always there. It’s hard to answer the question — can we ever be done? No. Because we’ve got to keep improving. But that doesn’t mean we can’t celebrate the journey along the way, and celebrate the milestones and the things that we do. But with all the progress we’ve made, if we take an objective look, some of the stuff we did in the study with HFS said, again, we’re miles away from where we think we are. So 90% of people say the CEO sets the objectives, but only 60% of people trust the data that we produce. After all this time, only 60% of people trust the data we’re producing. So it says we’ve got a long way to go — but again, celebrate our successes and the progress.
No, I agree completely. Data is purpose-built. You’re using data to make business decisions, and the business decisions you need to make tomorrow will be different from those you make today; they’ll be standing on the shoulders of new data and new understanding and context of that data. So hopefully we will continue to evolve.
I can’t disagree with what Bets and Kevin said. Additionally, it’s a continuous journey and a 360-degree feedback loop, because while you’re working with the data and driving that data-driven culture in your organization, it has to be an agile, continuous feedback loop of: am I taking the right decisions with the data, and is this data still relevant for my continuous decision process? That’s where we need to be continuously working it, and it’s going to be a continuous journey. And having said that — how many times have we realized that, though data is the new oil of the industry, at certain positions leaders still take their decisions based on intuition. Sometimes the data is saying something, but they bring in their experience, their cognitive understanding, while taking the big bets. That’s where I think we have to continuously improve that data, and look at the day when we’re able to bring those intuition-related capabilities into the data, so it’s able to drive those kinds of decisions in terms of taking big bets, predicting with 100% accuracy, along with empathy and that human element while taking decisions. Because while we’re talking about digital employees and fully data-driven decisions, I think there’s still one thread which says, OK, I understand the data is showing me this trend, but based on my experience as a leader, as a head of function, I should be tweaking my decision. And that’s what I think all of us do at a certain point in time. That’s where I think it will continue to go.
I think I agree with all my peers that we’ll probably never be done — and I hope so, because as much as we talk about artificial intelligence and data and technology, human intelligence will always triumph over artificial intelligence. So with that, while we try to find solutions for all the problems we encounter — through technology, through collaboration, through tools — every problem we try to solve also creates a byproduct, which is a new set of problems. So there’s a solution for a problem, and there’s a problem for every solution you deploy. It’s a never-ending, iterative loop: you keep learning, iterating, and evolving, whether you apply it in the context of banking, or health, or wherever you can make a difference from a human-value standpoint. So it’s something we keep evolving, and every generation will have its own set of challenges — and the beauty is we’ll all work together to find new solutions.
Yeah, I’ll take the contrarian view — I think we can be done. The way I see it is as laps, or S-curves. If we break the problems into small laps or S-curves, whatever you want to call it — and I think that’s a critical skill for our data programs, defining milestones and considering that, OK, in this lap we’ve completed it, and celebrating that. Because otherwise the organization is like, when will I have perfect data? Never, from that perspective. But can we start using data to make decisions and showcasing them, as I said earlier? I think that’s one thing to celebrate. And then reuse: as data people, we sometimes jump from one tool to the other, and it infuriates the business people — OK, it was this, then it’s cloud, now it’s ChatGPT, then it was others. So yes, we want to move to the next generation, but also celebrating what we’ve achieved is important. That’s why I want to say yes, we can be done, but we also deliberately say, OK, now we’re moving to the next lap or the next S-curve. I think that clarity is needed, because otherwise it seems like we’re hoodwinking them. We’ll talk about this being the end of the era, and then we start the new era. So that’s how the eras will go on.
All right, we’ll do 4, 5, 6, 7 with you — ChatGPT 10.5, and so on and so on. All right, well, give it up for my panel, everybody! I’d like to quote from earlier: hype is about hope, at the end of the day. And I think with this data conversation, there’s certainly a lot of hope for how we’re actually making progress — even if it’s just 60% today on data quality, we’ll get there. All right, thank you. Thank you.
Your account has been created. You can continue exploring free AI insights while you verify your email. Please check your inbox for the verification link to activate full access.
With the exception of our Horizons reports, most of our research is available for free on our website. Sign up for a free account and start realizing the power of insights now.
Our premium subscription gives enterprise clients access to our complete library of proprietary research, direct access to our industry analysts, and other benefits.
Contact us at [email protected] for more information on premium access.