 From around the globe, it's theCUBE with digital coverage of MIT Chief Data Officer and Information Quality Symposium, brought to you by SiliconANGLE Media. Hello everyone, this is Dave Vellante and welcome back to our continuing coverage of the MIT Chief Data Officer, CDOIQ event. Inderpal Bandari is here. He's a leading voice in the CDO community and a longtime CUBE alum. Inderpal, great to see you. Thanks for coming on to this special program. My pleasure. So when you and I first met, you laid out what I thought was, one of the most cogent sort of frameworks to understand what a CDO's job was, where the priorities should be and one of those was really understanding how data contributes to the monetization of station, aligning with the lines of business, a number of other things. And that was several years ago. A lot has changed since then. We've been doing this conference since probably 2013. And back then, Hadoop was coming on strong. A lot of CDO's didn't want to go near the technology. That's beginning to change. CDO's and CTO's are becoming much more aligned at the hip. The reporting organizations have changed. But I love your perspective on what you've observed as changing in the CDO role over the last half decade or so. Well, Dave, you know that I became a chief data officer in 2006, December 2006. And I've done this job now four times, four major organizations. I've created the organization from scratch each time. Now, in December of 2006, when I became chief data officer, there were only four chief data officers globally. And I was the first in healthcare. And there were three others, one in the internet, one in credit cards, one in banking. And I think I'm the only one actually left standing still doing this job. I don't know if that's a good thing or a bad thing, but like you noted, it certainly has allowed me to learn the craft and then also script it down to the level that, you know, I actually do think of it purely as a craft. That is, I know going into a new job, what I'm going to do day one, et cetera, et cetera. Now, the interesting things that have unfolded, obviously the professions taken off, there are literally thousands of chief data officers now. And there are plenty of changes. I think the main change with the job is it's, I think a little less daunting in terms of convincing the senior leadership that it's needed because I think the awareness at the CEO level is much, much, much better than what it was in 2006 across the board. Now, having said that, I think it is still only awareness. I don't think that there's really a deep understanding of those levels. And so there's a lot of confusion, which is why you will, you kind of, this is my theory, but you saw all these professions take off at the C titles, right? Chief data officer, chief analytics officer, chief digital officer, and chief technology officer, CIO of course has been there for a long time. And, but I think these newer C positions, they're all very, very related and they all kind of went to the same need which had to do with enterprise transformation. Digital transformation at enterprises, chief digital officer, that's enough. And people were all trying to essentially feel the elephant and they could only see part of it at the senior levels and they came up with whichever role, seemed most meaningful to them, but really all of us are trying to do the same job which is to accelerate digital transformation in the enterprise. Your comment about, you kind of see that CTOs and CTOs now partnering up much more than in the past, and I think that's inevitable. The major driving force for that is, in my view anyway, is artificial intelligence. As people try to infuse artificial intelligence, then it's a very technical field still. It's not something that you can just hand over to somebody who has the business chops but not the deep technical chops to pull that off. And so in the case of chief data officers that do have the technical jobs, you'll see them also pretty much heading up the AI effort internally. And as I do for in the IBM case, where we building the data and AI enablement internal platform for IBM. But I think in other cases where you've got chief data officers who are coming in from a different angle, they will partner up with the CTO now because they have to, otherwise you cannot get AI infused into the organization. So there were a lot of other priorities obviously. I mean, certainly digital transformation, we've been talking about it for years, but still in many organizations, there was a sense of, well, not on my watch, maybe a sense of complacency or maybe just other priorities. COVID obviously has changed that. Now 100% of the companies that we talk to are really putting this digital transformation on the front burner. So how has that changed the role of CTO? Has it just been an acceleration of that reality or has it also somewhat altered the swim lanes? I think it's bolt actually. So I have a way of looking at this. In my mind, the CTO role, if you look at it from a business perspective, they're looking for three things. The CEO is looking for three things from the CTO. One is, this person is going to help with the revenue of the company by enabling the production of new products, new products resulting in new revenue and so forth. That's kind of one aspect of the monetization. Another aspect is the CTO is going to help with the efficiency within the organization by making data a lot more accessible as well as enabling insights that reduce end to end cycle time for major processes. And so that's another way that they're going to have monetized. And the last one is risk reduction, that they're going to reduce the risk as regulations and as you have the cybersecurity exposure and incidents that just keep accelerating as well. You're going to have to also step in and help with that. So every CTO, the way their senior leadership looks at them is some mix of the three. And in some cases, one is given more importance than the other and so forth, but that's how they are essentially looking at it. Now, I think what digital transformation has done is it's managed to accelerate, accelerate all three of these outcomes because you need to attend to all three as you move forward. And I think that the individual balance that's struck for individual CDOs really depends on their company, their situation, who their peers are, who is actually leading the transformation and so forth. You know, in the value pie, a lot of the early sort of activity around CDO sort of emanated from the quality portions of the organization. It was sort of a compliance weighted role, not necessarily when you started your own journey, you obviously had been focused on monetization, how data contributes to that. But you saw that generally organizations, even if they didn't have a CDO, they had this sort of back office clients thing, that has totally changed in the value equation. It's really much more about insights, as you mentioned. So one of the big changes we've seen in the organization is that data pipeline you mentioned end to end cycle time. And I'd like to dig into that a little bit because you and I have talked about this. This is one of the ways that a chief data officer and the related organizations can add the most value reduction in that cycle time. That's really where the business value comes from. So I wonder if we could talk about that a little bit and how the constituents and the stakeholders in that life cycle across that data pipeline have changed. No, that's a very good question, a very insightful question, Dave. So if you look at a company like IBM, and my role internally within IBM, is to enable IBM itself to become an AI enterprise. So infuse AI into all our major business processes, things like our supply chain, our lead to cash process, you know, our finance processes, like accounts receivable and procurement and so forth. I mean, every major process that you can think of is using Watson now. So that's the vision, that's essentially what we've implemented. And that's how we are using that now as a showcase for our clients and customers. One of the things that we realize is the data and AI enablement parts of the business, you know, the work that I do also has processes. And that's the pipeline you refer to, you know, we're setting up the data pipeline, we're setting up the machine learning pipeline, the deep learning pipeline. We're always setting up these pipelines. And so now you have the opportunity to actually turn the so-called AI ladder on its head, because the AI ladder has to do with, hey, first you collect the data, then you curate it, you make sure that it's high quality, et cetera, et cetera, fit for AI, and then eventually you get to applying, you know, AI and then infusing it into business processes and so forth. But once you recognize that the very first, the earliest pieces of work with the data, those themselves are essentially processes, you can infuse AI into those processes. And that's what's made the cycle time reduction and all the things that I'm talking about possible, because it just makes it much, much easier for somebody to then implement AI within a large enterprise. I mean, AI requires specialized knowledge. There are pieces of AI like deep learning, whether, you know, typically a company is gonna have like a handful of people who even understand what that is, how to apply it, you know, how models drift, when they need to be refreshed, et cetera, et cetera. And so that's difficult. You can't possibly expect every business process, every business area to have that expertise. And so you've then got to rely on some core group, which is going to enable them to do so. But that group can't do it manually because otherwise that doesn't scale again. So then you come down to these pipelines and you've got to actually infuse AI into these data and AI enablement processes so that it becomes much, much easier to scale across an enterprise. Some of the CDOs, maybe they don't have the reporting structure that you do or maybe it's more of a far-flung organization. Not that IBM is not far-flung, but they may not have the ability to sort of inject AI. Maybe they can advocate for it. Do you see that as a challenge for some CDOs and how do they sort of get through that? What's the way in which they should be working with their constituents across the organization to successfully infuse AI? Yeah, that's, it's in fact, again, a very good point. I mean, when I joined IBM, one of the first observations I made and I in fact made it to our senior leadership is that I didn't think that from a business standpoint, people really understood what AI meant. So when we talked about a cognitive enterprise or an AI enterprise as, as IBM, you know, our clients didn't really understand what that meant, which is why it became really important to enable IBM itself to be an AI enterprise. You know, that, that's my data strategy. You've, you had kind of alluded to the fact that I have this approach where there are these five steps. Well, the very first step is to come up with a data strategy that enables a business strategy that the company's on. And in my case, it was, hey, I'm going to enable the company because it wants to become a cloud and cognitive company. I'm going to enable that. And so we essentially, our data strategy became one of making IBM itself an AI enterprise. But the reason for doing that, the reason why that was so important was because then we could use it as a showcase for clients and customers. And so when I'm talking with our clients and customers, that's my role. I'm really the only role I'm playing is what I call an experiential selling one, where I'm saying, forget about, you know, the fact that we are selling this particular product or that particular product that you've got GPU servers, we've got, you know, what's an open scale or whatever. It doesn't really matter. Why don't you come and see what we've done internally at scale. And then we'll also lay out for you all the different pain points that we had to work through using our products so that you can kind of make the same case when you apply it internally. And the same comment with regard to the benefits, you know, the cycle time reduction. Some of the cycle time reductions that we've seen, I mean, in my processes itself, you know, like this thing about metadata, business metadata, generating that is so difficult. And it's again, something that's critical if you want to scale your data because, you know, you can't really have a good catalog of data if you don't have good business metadata. So anybody looking at what's in your catalog won't understand what it is. They won't be able to use it, et cetera. And so we've essentially automated business metadata generation using AI. And the cycle time reduction there was like 95%. You know, I would actually argue it's more than that because in the past, most people would not, you know, for many, many data sets, the pragmatic approach would be, you don't even bother with the business metadata. Then it becomes just put somewhere in your, you know, data architecture, somewhere in your data lake or whatever you have data warehouse. And then it becomes a data swamp because nobody understands it. Now, with regard to our experience applying AI, infusing it across all our major business processes, our average cycle time reduction is 70%. So just tremendous amount of gains are there. But to your point, unless you're able to point to some application at scale within the enterprise, you know, that's meaningful for the enterprise, which is kind of what the role I play in terms of bringing it forward to our clients and customers. It's harder to argue or make a case for investment into AI within the enterprise without actually being able to point to those types of use cases that have been scaled and where you can demonstrate the value. So that's extremely important part of the equation to make sure that that happens on a regular basis with our clients and customers. I will say that, you know, your point is valid. A lot of our clients and customers come back and say, tell me when they're having a conversation. I was having a conversation just, you know, last week with a major financial services organization. And I got the same point saying, we're coming out at regulation. How do I convince my leadership about the value of AI? And, you know, I basically responded, yes, we've got the scale use cases, you can show that. But perhaps the biggest point that you can make as a CEO back to the senior leadership is can we afford to be left out? That is the, I think the biggest, you know, point that the leadership has to appreciate. Can you afford to be left out? I want to come back to this notion of 70% on average, the cycle time reduction. I mean, that's astounding. And I want to make sure people sort of understand the potential impacts. And I would suspect that many CDOs, if not most, understand sort of system thinking. It's obviously something that you're big on. But oftentimes within organizations, you might see them trying to, you know, optimize one little portion of the data life cycle. And, you know, having, okay, hey, celebrate that success. But unless you can take that systems view and reduce that overall cycle time, that's really where the business value is. And I guess my real question around this is, you know, every organization has some kind of North Star. Many are about profit and you can increase revenue or cut costs and you can do that with data. It might be saving lives. But ultimately to drive this data culture, you've got to get people thinking about getting insights that help you with that North Star, that mission of the company, but then taking a systems view. And that 70% cycle time reduction is just the enormous business value that that drives. I think sometimes gets lost on people. And these are telephone numbers in the business case, aren't they? Yes, no, absolutely. It's, you know, just tremendous amount of potential. And it's, you know, it's not an easy, easy thing to do by any means. So, and we've been always very transparent about that Dave, as you know, I mean, we put forward this blueprint, right? The cognitive enterprise blueprint, how you get to it. And I kind of have these four major pillars for the blueprint, there's obviously there's data and you know, you're getting the data ready for the transformation that you want to do. But also things like, you know, training data sets, how do you kind of run hundreds of thousands of experiments on a regular basis, which kind of then leads you to the other pillar, which is technology. But then the last two pillars are business process change and the culture, organizational culture, you know, managing organizational considerations and culture, because if you don't keep all four and lock staff, the transformation is usually not successful at an end to end level. Then it becomes much more what you pointed out, which is you have kind of point solutions and the role, you know, the CDO role doesn't make the kind of strategic impact that otherwise it could do. So, and this also comes back to some of the earlier points you're going to do. If you think about how do you keep those four pillars and lock sync, it means you've got to have the data leader. You've also got to have the technology leader. And in some cases, they might be the same people, right? But just for the moment's sake of argument, let's say they're all different people. And many, many times they are. So the data leader, the technology leader and the operations leaders, because they're the ones who own the business processes, as well as the organizational leaders, you know, they've got to all work together to make it an effective transformation. And so the organization structure that you talked about that in some cases, my peers may not have that, you know, that is true. If the senior leadership is not thinking overall digital transformation, it's going to be difficult for them to then go down that path. You've also seen that culturally, historically, when it comes to data and analytics, a lot of times the lines of business, their first response is to attack the quality of the data because the data may not support their agenda. So there's this idea of a data culture. And then I want to ask you how self-serve fits into that. I mean, to the degree that the business feels as though they actually have some kind of ownership in the data. And it's largely their responsibility as opposed to a lot of the finger pointing that has historically gone on, whether it's been decision support or enterprise data warehousing or even, you know, data lakes, they've sort of failed to live up to that promise, particularly from a cultural standpoint. And so I wonder how have you guys done in that regard? How did you get there and any other observations you could make in that regard? Yeah, so I think culture is probably the hardest nut to crack out of those four pillars that I talked about. And you've got to address that, not just top down, but also bottom up as well as peer to peer. And I'll give you some examples based on our experience at IBM. So the way my organization is set up is there is obviously a technology arm and they are the people who are doing all the data engineering work kind of laying out the foundational technical elements for the transformation, you know, the AI enablement, the deep learning networks and so forth. So there are those people. And then there is another senior leader who reports directly to me and his organization is all around adoption. So he's responsible for essentially taking what's available in the technology and then working with the business areas to move forward and make an infused AI into the processes that the business area is working in. It's done in a bottom up way. It's deliberately set up, I designed it to be bottom up. So what I mean by that is the team on my side is fully empowered to move forward provided they find a like-minded team on the other side and go ahead and do it. They don't have to come back for funding. They don't have to, you know, they can just go ahead and do it. They're basically empowered to do that. And that particular setup enabled us in a couple of years to have 100,000 internal users on our central data and AI enablement platform. And when I mean 100,000 users, I mean users who are using it on a monthly basis recount them. They're, you know, so if you haven't used it in a month, we won't count you. So it's over 100,000, even very rapidly to that. It's kind of an enterprise wide story. That's kind of the bottom up direction. The top down direction was the strategic element that I talked with you about where I said, hey, our data strategy is going to be to create, make IBM itself into an AI enterprise and then use that as a showcase for clients and customers. That kind of, and we reiterate that, you know, I work the senior leadership on that view all the time and talking to customers, et cetera, and our senior leaders. And so that's kind of the air cover to do this, you know, that makes, gives you that possibility. I think from a peer to peer standpoint, you get to these large scale end-to-end processes. And there are a couple of ways I work that. One way is we've kind of looked at our enterprise data and said, okay, there are four major pillars of data that we want to go after. You know, there's data about our clients, data about our offerings, a data about, you know, our financial data that we, you know, and then our workforce data. And then within that, there are obviously sub pillars, you know, there's like some sales data that comes in and some, you know, within workforce, you could have contractors versus employees, et cetera. But think for the moment about these four major pillars of data. And so let me map that to end-to-end large business processes within the company, you know, the really large ones like enterprise performance management, end-to-end or lead-to-cash generation, end-to-end, risk insights across our full supply chain, end-to-end, things like that. And we've kind of tied these four major data pillars to those major end-to-end processes. What, yes, that there's a mechanism there, obviously, in terms of facilitating and to some extent, one might argue, even forcing some interaction between teams that otherwise might not talk. But it also brings me and my peers much closer together when you set it up that way. And that means, you know, people from the HR side, people from the operations side, the data side, the technology side all coming together to really move things forward. So all three tracks being hit very, very hard to move the culture forward. And am I also correct that you have chief data officers that report into you, whether it's a matrix or direct, within the divisions? Is that right? Yes, so IBM, you know, in terms of our structure, as you know, we are a global company. We're also, you know, a far-flung company. And we have many different products and business units and so forth. And so one of the things that I realized early on was we are going to need data officers in each of those business units. And the business units, you know, there's obviously the enterprise objective. And, you know, you could think of the enterprise objectives in terms of some examples, based on what I said in the past, which is so an enterprise objective would be, yeah, we've got to have our data foundation by essentially making data along these four pillars. I talked about clients, offerings, et cetera. You know, very accessible self-service. You had mentioned self-serve, actually, this is where the self-service piece comes in, right? So you can get at that data quickly and appropriately, right? I mean, you want to be make sure that the access control and all that stuff is ironed out. And you're able to change your policies and it's not manual, but, you know, those things get implemented very rapidly and quickly. And so you've got that piece of the puzzle to go after. And then I think the other aspect of this is that when you recognize that every business unit also has its own objectives and they are looking at some of those things somewhat differently. So I'll give you an example. I mean, we've got data and AI product units. Now, those CDOs, right, their concern is going to be a lot more around the products themselves and how we're monetizing those products. And so they're not per se concerned with, you know, how you reduce the end-to-end cycle time of IBM's internal supply chain. So this is my point. But they're gonna have substantial considerations and objectives that they want to accomplish. And so I recognized that early on and we came up with this notion of a data officer council. And I helped staff the council. So this is why that's the matrix reporting that we talked about. I selected some of the key players that we have in those units. And I also made sure they were funded by the units. So they report into the units because their paycheck is actually determined by the unit and which makes them then aligned with the objectives of the unit, but also obviously part of my central approach so that I can disseminate it out to the organization. It comes in very, very handy when you are trying to do things across the company as well. So when we, you know, GDPR, when we have to get the company ready for GDPR, I would say that this mechanism became a key, key aspect of what enabled us to move forward and do it rapidly, you know, from within my organization. Because you had the structure that perhaps the lines of business weren't maybe as concerned about GDPR, but you had to be concerned with it overall. And this allowed you to sort of heighten their importance. Right, because think of, in the case of GDPR, they have to be a company-wide policy and implementation. Right. And if we did not have that structure already in place, it would have made it that much harder to get that uniformity and consistency across the company. Right. You know, so you would have to invent that structure, but we already had it because we, you know, we said, hey, this is around for data, we're going to have these types of considerations that play out. And so we have this regular, you know, this network that meets regularly every month, actually. And, you know, when things like GDPR hit, then much more frequently than that. Right, so that makes sense. So we're out of time, but I wonder if we could just close if you could address the MIT CDO audience, the probably, this is the largest audience, believe it or not, now that it's virtual, it definitely expanded the audience, but it's still a very elite group. And the reason why I was so pleased that you agreed to do this is because you've got one of the more complex organizations out there and you've succeeded and a lot of the hard work. So what message would you leave the MIT CDO audience, Inderpal? So I would say that, you know, it's this particular profession, the CDO profession, is if I have to pick one trait, let me pick two traits. I think one is you're a change agent, so you have to be really comfortable with change. Things are going to change, the organization is going to look to you to make those changes. And so that's one aspect of your job that may or may not leap out immediately, but those particular set of skills and characteristics is something that one has to develop over time. And I think the other thing I would say is it's a continuous learning job. So you're continuously learning and things keep changing around you and changing rapidly. And, you know, if you just even think just in terms of the subject areas, I mean, this CDO today, you've got to understand technology. Obviously, you've got to understand data. You've got to understand AI and data science. You've got to understand cybersecurity. You've got to understand the regulatory framework. And you've got to keep all that in mind and you've got to distill it down to certain trends that's happening, right? I mean, so this is an example of that is there's a trend towards more regulation around privacy and also in terms of individual ownership of data, right? Which is like very different from what's before, but that's kind of where the puck is going. And so you've got to be on top of all those things. And so the characteristic of being a continual learner I think is a key aspect of this job. One other thing I would add, and this is post COVID-19, you know, pre-COVID-19 in terms of those four pillars that we talked about, you know, which had to do with the data technology, business process and organization and culture. From a CDO perspective, the data and technology were obviously front and center. I would say post COVID-19, post the civil unrest and so forth, you know, the other two aspects are going to be critical as we move forward. And so the people aspect of the job has never been, you know, more important than it is today. Right. And that's something that I, you know, I find myself not regularly doing this talking at all levels of the organization one-on-one, which is something that, you know, we never really did before, but now we find time to do it. So obviously it's doable. I also think it's just, it's a change that's here to stay and it should stay. Well, to your point about change, if you were in your comfort zone before 2020, this year has certainly taken you out of it. Indapal Bhandari, thanks so much for coming back in theCUBE and addressing the MIT CDO audience. I really appreciate it. Thank you for having me, Dave. My pleasure. You're very welcome. And thank you for watching everybody. This is Dave Vellante. We'll be right back right after this short break. You're watching theCUBE.