 Live from Cambridge, Massachusetts, it's the Cube. At the MIT Chief Data Officer and Information Quality Symposium. With hosts Dave Vellante and Jeff Kelly. Welcome back to Cambridge everybody, Dave Vellante here with Jeff Kelly of Wikibon. This is Silicon Angles the Cube. The Cube is our live mobile studio. We go out to the events. We extract the signal from the noise. We're here at the MIT Information Quality Symposium. It's a chief data officer forum, probably the foremost forum of its kind and leading thinkers are here. Eugene Culker is here. He's a PhD and the chief data officer at Seattle's Children's Hospital. Welcome to the Cube. Thanks for coming on Eugene. Thank you for inviting me. So the role of the CDO has emerged in some cases like gangbusters, particularly in industries like yours. I wonder if you could talk about how you became the CDO and how that role emerged at Seattle Children's. Yeah, so we started seven and a half years ago and it was quite surprising for me. I got called and was invited to talk about what Seattle Children's is planning to do in the next five to ten years with regard to data. And the idea was that we're planning to treat data as strategic institutional asset. And to do that, you need to have somebody who's going to be responsible for this kind of activity. And we at Seattle Children's, and me personally, we believe that sharing best practices and sharing our lessons learned would be great for, important for other people in order to prevent them making the same mistakes we made and hopefully improve what they do and how. So one of the key important messages we're trying to get across in our entire organization is that decision can be made not only based on gut feelings, but also be data driven. And in many cases, those decisions, either strategic or business or operational, they could be better decisions just because they're based on data. And we're trying our best to make that reality on everyday basis or for major strategic decisions. We tried a lot of interesting approaches, how to do it, and some interesting lessons we learned. I'd say they're coming back to something which most people would say it's so obvious. We don't need to be Ph.D. to get to that point. But the deal is that data, information, technologies, approaches are extremely important. But the most important of all this are, of course, people. Because whatever you do with your analytics, with your approaches, with your methods, you need to make it actionable. Those insights should be actionable. You should have shared vision that people can act on that and make real impact. Our goal is to improve care for our patients and families. And it's not trivial in healthcare. It's extremely complicated. It's unique in many senses and many issues. It has a lot of similarities, though, with other industries through when this similar transformation healthcare is going through right now. Point would be that in addition to doing proper data science analytics analysis, you need to be able to communicate it properly, to make it working for other people to act on. And that's the key driver of what you're trying to do. So you mentioned Gut Feel before. There was an article in the early 2000s in the Harvard Business Review about how Gut Feel trumped technologies and other approaches that people were using. They had big data warehouses at the time. And then a number of people tried to build mathematical models and sampling and the like. And now, sampling's almost disappeared, right? You have all this data. So the technology has evolved greatly. You're saying technology is only one piece of the equation. It's people, and I'm sure process fits in there as well. But the technology has evolved very rapidly. So my question is how have the people side, or even the process side, has it evolved as rapidly, or is that the bottleneck? I'd say that people's side is always would be bottleneck, challenge, and also opportunity. As we heard in one of the recent presentations earlier today, that sometimes you don't need big data, you need smart data. You'd like to have better decision based on data. And it could be done on everyday basis, or it could be done strategically. And I think that issue of how to work with people who are not the data challenge like some of us, okay, to understand what we do, how we do, for them to believe trust in our data stories. We call them data stories, okay? This is extremely important. This is not done for people who are technology based. It's usually done for people who develop software or statistical analytics approaches. But this is part of where best business practices, consulting practices, extremely important. Marketing, communication is extremely important. And we're trying to learn how to do it, and we've made some major progress along those lines. So you're a data guy, but you're a storyteller as well, you're saying? Actually, you have to be storyteller. Otherwise, you know, you're going to be somewhere and people are going to be in another place and it's not going to be actionable. You're not going to make a real impact, not real, you know, improvement of care of patients and families. So tell us a data story. So, you know, in healthcare, like in other industries, you have, you know, long term beliefs and thinking about important issues that people are not even questioning, okay? Just one story, you have... The world is flat. Yes, yeah, so something like that. So, you know, nurses, right? It's very important, you know, in our place, which is approximately less than two billion a year enterprise, five and a half thousand employees, quarter of those are nurses. Nurses are so important to work with patients and families. So when we sometimes see signs of, you know, good nurses, you know, experienced nurses, you know, live in our place, we really worry about that. So we want to understand what's happening. There's some beliefs that people not even question. For example, if you have hospital, just very simple, hospital inside city, all conditions equal, you better hire a nurse which lives in the city versus the one who lives in suburbia. Obviously, right? This person gonna run, you know, using bus, go to the place, and that's basically... For the big commutes, that's it. Yes, exactly, don't need to fight traffic, the whole deal, right? No, it's completely wrong. And then the real issue is that those nurses, primarily it's women who live in suburbia, they have larger families, more kids, higher mortgages, they need stable jobs, right? And they need to be able to kind of adjust their lives around family too, right? So for long-term longevity for those, you know, it's better to have something, you know, if they like it and they experience they're good, so they would be valid on market. They prefer to stick with good place, right? Those nurses who live in city instead, if they don't like something, they're gonna change bus. This is long-term belief in our industry. More risk takers, right? Exactly, of course. And your data showed that. And we showed that, you know, very explicitly. It was eye-opener for us, for many people in our place, it was eye-opener for many people outside of our place, what we learned that other people also came to that kind of conclusion. Sometimes we're good at sharing. Those people, we are not so good, so we were not, we basically kind of learned it by ourselves. But this is, you know, a very simple example of those stories, right? And I can give you stories and stories like that. That's great. So the importance of stories like that, and first let me back up and say my wife's a nurse, I can very much relate to everything you just said. And we do live out in Suburbia and she worked here in the city. But talk about the importance of stories. So the ability to articulate and tell a story like that is, what's the main purpose? To sway stakeholders into taking certain actions, into adopting new ways of doing business. Is that really the goal of storytelling when it relates to data? I think there are multiple goals always, right? So in the first approximation of reality for the specific project you're working on, right? You're going to have not only stakeholders, but engaged stakeholders. And those who are going to act on whatever, you're going to be working with them. And delivering whatever data stories, right? Together with them. You know, for them to believe in those, because they were part of the whole story. Not telling, but producing this story, producing this analysis. That's extremely important. It's very different if you, you know, get your, you know, whatever problem and you're trying to do it in the corner. So there's data nerds, right? And you come back with solution and, you know, you lost everybody, right? So that's on surface level. Strategically, long term, of course, you would like to have more people engaged with these kind of activities. So they would have, they would become smarter users of data. So they would become advocates of what you do. And they would basically engage in these kind of activities and tell their co-workers, this is fantastic capability, we can do it together. This is not black box, like, you know, most of the time we feel like it is, okay? And actually those people, maybe data nerds, but maybe they're also data docs, doctors. So they help us and we help each other together to achieve, again, improve, you know, care for patients and families. This is our ultimate goal. So if I understand what you're saying correctly, really, the idea here is not to have, you know, maybe this great little collection or group of data scientists who go off in the corner and they come up with these great insights and then they just, you know, they tell a CEO or they tell a C level executive and that person says to the organization you're going to do it this way from now on. It's get those stakeholders who are actually going to have to affect change and allow them to actually help solve the problem using data. Yes, absolutely right. So, you know, one of the, another, you know, clear examples how it's working or not working is a standard, you know, A-B comparison or buy versus build, you know, another story I can tell you. And we've done it wrong way and we've done it right way. So FAIL project basically was similar with successful projects with regard to major components we thought. Data science analytics and actionable insights, both were of high quality. On FAIL project we had too many insights but we didn't worry about that, okay? That was very wrong, okay? Why? Because it turned out to be that the process was not really well defined, okay, in FAIL. So it was well defined in success project, okay? Customers were identified but, you know, in FAIL project we were unable to transform our concerned stakeholders into key, clear, you know, customers. And then I would say most important issue with regard to core question, core issue, we were unable to transform this, I'd say, free-floating anxiety into something which is, you know, testable and, you know, actionable. So as a result of it there was no shared, you know, vision, expectations were all over the map in FAIL project versus, you know, fully well articulated vision which is actionable for success project. And then this kind of issue after issue it's built on communication, our ability to communicate properly through entire process with transparency, engagement of our customers versus not. And of course great data science is important, you know, technology is important, you know, approaches are important. This is all important but you need to have shared vision, you need to have actionable in science behind the shared vision and you have to have it done with best business practices, best, you know, approaches to communicate with people including marketing, including your data storytelling capabilities. So in our case, in our team, it's not just about data scientists, we have business person program manager, we have somebody who help us with writing, so, and we are engaging from the very beginning trying to get well-defined project initiatives, clear customers and identified core questions from the get-go that everybody on the same page, whether we work within somebody within walls or outside vendor, it's, we treat them equally as collaborators co-workers. That helps. So Eugene, you're on a panel tomorrow, how to establish CDL function within your organization. Before we get into the sort of how to do that, I've been sort of taking notes, you got to be a story teller, you're a data guy, you got to be good communicator, you got to at least have that function in your organization. So I have to be a data person to succeed at being a CDO or can I just have an appreciation for data and bring in skill sets, what are your thoughts on that, what are the attributes of a successful CDO? I'm not sure and I think it depends on different industries actually, there are a lot of similarities but if you're talking about companies like Amazon and Google, it could be different from somebody who is working in a small bank or in an advertisement agency, right? I think you need to have, you know, there's some standard skill sets, you need to understand data, you need to understand analytics, data science, you have to do that, you need to understand how to communicate with people, you need to know your industry too, otherwise there are going to be a lot of misalignments and we learn it hard way as well. When we became a board, we didn't know a lot about healthcare, we were coming from a completely research biomedical research angle and you cannot buy that, you need to have this experience, you know, it takes years to get that one. So I'd say you need to have, you know, an ideal case, an optimal case, all these different capabilities and have that also projected into your team and you need to like working with people, otherwise, you know, you're going to be sitting in the corner and successes would be very rare. Now you, to whom do you report within the organization? So I report to the president of research and directed to CEO, but what we figured out what's really, in addition to that important, you know, reporting schema, we also figured out optimal for our organization system how we're going to work with both business and side, NIT side. So in our organization, any major project initiative goes through executive sponsors. So we work primarily with, on business side, with our business clinical leader, which is medical director, and on IT side, we work with CIO. That helps a lot because we get best of the both worlds, right? And as any healthcare organization is extremely complex, so you need to have your business driver and your IT driver working with you closely, then you're going to be successful. Otherwise, it's going to be more complicated, more convoluted, you're going to limit yourself. So obviously, you're separate from the CIO, who does the CIO report to? To CEO. Okay. So you have sort of a dotted line to the CEO, right? Yes. Yes. And then you talked about projects. How does that work? Do you work through a PMO or do you actually own projects? Are there data projects that you have full responsibility for? So basically, we have an entire process how we're going to work on project or not, okay? And our executive sponsors, you know, they work with us, you know, and their decision makers, if we're going to take on it or not, we also need to agree and be able to do it, right? So the process includes, I would say, best business practices. You know, first, you know, we kind of have, you know, a step of identification and qualification if this process is going to be qualified for what we can deliver, okay? If that's checked, if that's okay, by our executive sponsors and stakeholders, we go to next stage, which is confirmation stage. And then, you know, we do a lot of due diligence and work with, you know, people to want to define it very clearly, okay? Next stage, we go into planning, right? And again, on each and every step, we do communication transparent with customers, with the exact sponsors, with, you know, everybody relevant, you know, to this specific project or initiative. Then we go to implementation stage. I'm not telling you anything new. Yeah, it's a classic PDIM, right? Exactly, exactly. And then the next one is, of course, measurements, right? And documentation and reporting. So, but I would say one of the major lessons we learned, you cannot do it solo. You cannot do it by yourself. Even if you have, like, Uber key sponsors, you need to have your stakeholders and make them your customers, advocates, and work with them all along the way. Because that way, your data story is going to be trustable, believable, and people would like to act on them. How big is your organization? Organization is, you know, is less than two billion a year. And specifically the CEO office, yeah, sorry. So, we have dozen to, dozen and a half people of different expertise. We have people who are data scientists. We have software folks. We have informatics folks. We also have, again, program manager, business person, somebody who is our technical writer. So, we're trying to blend these capabilities. And who's the communicator? You? That's your job? Yeah. We're trying to actually, depending who is lead on this project, to work it on team level, not only on my level. So it's going to be great for everybody who is participating in those projects. But who's your PR person? Is that you? Or is that you have a particular, you know, head of comms within your group? So we work with consultants and they help us to communicate our messages better, too. And I can tell you, we're trying to engage with people like internal consultants. And right now we are at the point that we're ready to share those expertise and lessons learned with people outside, too. Because we think that improving care for patients, family, you know, nationwide or worldwide is extremely important. And some lessons we'd like people to get from us, not to make their own mistakes. All right, Eugene, we'll leave it there. Thanks very much for coming to theCUBE and sharing your stories and congratulations on your success and good luck going forward. Thank you very much. All right. Keep it right there, everybody. Jeff Kelly and I will be back. Paul Gillin is also here. We'll be doing some of the hosting. We're at the MIT Chief Information, Chief Data Officer, CDO, MIT IQ hashtag MIT IQ on Twitter. This is theCUBE. Right after this.