Chief Data Officer Richard Wendell: Skills to Cultivate (Part 2)

In this second podcast, Mr. Wendell continues where he left off last time.

He explains the skills you’ll need in order to be an effective Chief Data Officer and we learn more about MIT’s International Society of Chief Data Officers.

Transcript
Skills a CDO Need: Information Technology, Mathematics, Change Management
Inside Out Security: You've given many examples how the CDO has a relationship with a CIO, the CMO, the CEO, and so you need a whole bunch of different skillsets. What are some skills you need in order to be an effective CDO?

Richard: Yeah, there are really three category of skills.

The first category is what I'll call an IT skillset, traditionally. The second is more of a math skillset, and the third is really, like you mentioned, around communication, and even HR and change management. So I could talk to each of those briefly.

Information Technology

It's an interesting role, right, because typically, people who are strong in IT may not have as much background or expertise in math or HR, and you could say that about the other two as well. These are the three different areas of skills that often do not overlap, and to be a good CDO, you absolutely must have all three skills areas.

The IT skillset is all about new data technology. So I think the number one, if you go online and you look at search terms, the number one phrase that's most commonly associated with the chief data officer is data science.

Data science, if you look at, again, it mean a lot of things to a lot of people. But chief data officers, chief analytics officers manage the data science function.

Data science takes place in most companies now on top of newer data technology stacks. There are so many new technologies emerging every day that are absolutely critical for managing the function of data science, data integration.

And so being able to go in and work with IT department on building out that technology suite, and even occasionally standing up IT infrastructure with these new kinds of tools that, you know, maybe the IT department is typically focused more on very, very proven technologies that are enterprise scale that could be deployed and maximized their IT ROI, that is what they should be focused on. That's the perfect focus for IT.

But if that's all you do as a company, then you're never going to experiment with somebody's new technologies that are really required to do data science well, and this is where a CDO comes in.

So, you know, it's really important to be able to stand up and manage some of these new IT technologies, and you have to be a hacker to make it work. I mean, a lot of companies I know spend a year and a half just trying to figure out how to productionalize their hadoop cluster inside their firewalls. So you have to know how to hack through these things and hack around them. I find a lot of my peers in the CDO community grew up as hackers, and you really have to have that hacker mindset and enjoy problem solving.

Mathematics

Secondary of math, I mean, this is really all algorithms so you need to understand machine learning. You need to know what are the different flavors of machine learning and how is it applied. And you need to be able to, I think in order to be good, you need to be able to get down to a fairly detailed level with your data scientist to talk about different packages, and how they're applied in different parameters that they're using in their model. Machine learning is just one area that's really hot right now but I think of advanced analytics, some statistics all have many different models that are used to solve different types of problems.

And operations research, frankly, is an area that's often overlooked. There are many powerful quantitative techniques that come out of operations research, and increasingly now, computer science that are all really important and all have their place, and they're just different tools for different jobs. So we need to have that tool kits of algorithms to know which one or ones are best applicable to different business use cases.

If you want to do a cluster analysis with a huge dataset, maybe you want to do a simple k-means. If you have a smaller dataset and you think you can get more insight out of it, then maybe you can do a hierarchical. It's just one simple example, but you need to be able to match the business use case to the scale of data to the algorithm.

Change Management

And then the third area, I mentioned the HR skillset, is really around change management. This is where most companies I see really fall down because most companies focus on insight.

Insights are great. Insights don't make money.

And this is where I'm talking specifically about like 20th century companies that are looking to be 21st century companies. Companies like that, they are filled with a lot of really great talent that's just not used to being data driven in their workflows. And quite often, sometimes, there's resistance to data.

Maybe folks, at the end of the day, feel that their gut is going to give a better answer than a computer or an algorithm, and look, I'm not saying let's throw away gut feel. There are still many cases where we have to use gut feel, but I do believe that increasingly, over the next few decades, there are going to be a lot of knowledge workers that...their jobs are not going to be automated but they're going to be augmented.

They will be using data. They'll be using analytics to more directly inform the decision that used to be made more on gut feel, not replacing them but augmenting. And getting folks who are not used to trusting a dashboard to trust the dashboard is really hard. It's really hard. It's change management, and there are reams and volumes written on change management.

I've done a lot of this. I did a lot at American Express. You know, change management in business units, to become more data driven, there's a lot of best practices that go along with this.

And by the way, the skillset though is nothing to do with IT and nothing to do with the math, the first two skillsets I mentioned.

So, yeah, you're right. It's a very heterogeneous job that requires a very cross functional skillset to be successful.

Inside Out Security: It sounds like you need to start working when you're 15 years old.

Richard: You know, it's interesting. There's a lot to learn, I think, to be an effective chief data officer, for sure.

Importance of Curiosity in Data, Technology and People
Inside Out Security: And to also be curious about the data, to be curious about technology, and to be curious about people, too.

Richard: That's a really, really good way of putting it. You're absolutely right. It's curious across all three of those ends.

But it's something that I don't want to be discouraging.

I think that folks, more often than not, earlier on in their careers, specialize in maybe the first area, or the second area, maybe IT or math, and then expand over time to have all three of those critical areas checked.

But there are a lot of folks from data science increasingly are coming from liberal arts background. There's a whole new profession of data journalism. Data visualization is huge, and data science is coming in from that background. I would put that more on the HR communication skill bucket.

And I think if you're smart and you're really curious, as you said, there should be nothing stopping you from going out and acquiring all the skills that we're talking about.

Reasonable Timeline for Success
Inside Out Security: And so if you became a CDO, how much of a timeline would you give yourself? Because sometimes, when you join a company, people are like: "What are you doing there? Are you here to cut costs? Are you here to streamline things?"

What's a reasonable timeline you would give yourself as a CDO to validate and justify this new role?

Richard: It's a good question. I think that there are critical checkpoints along the way. I'll start at the beginning. I think that the first critical checkpoint is 90 days before you start.

And I'm not being facetious.

I actually believe that a good chief data officer is not going to take any CDO role that's not set up the right way.

And I've just found, with so many conversations I've had with companies that are looking to be successful in this area, they don't necessarily know how to structure the role for success. They don't know who...should the data warehouse and ETLT be under the CDO or not, or should that stay in IT.

And so I really think that it's incumbent on a CDO to make sure that that role is structured for success before they accept the job. Almost be a consultant to...and actually, a lot of CDOs who are in their roles were consultants or advisers to companies through this process.

I think that's number one, and then I think that, so coming in on day one with the right success structure and the right metrics is really important.

And then, you know, I think that it's interesting. A lot of people argue back and forth about should we be driving quick hits or no, actually, we should be driving transformation.

And it feels like there's people in one of these two camps, and the answer, I think, is actually both. You have to do both. I believe in taking a portfolio approach. I think that it's really important to very quickly, early on, identify a couple large transformational projects that are going to move the needle for the company, but are going to take more like, a year or two to do.

But in addition to that, there's got to be a whole hopper full of monthly click hits that come out, and they're just stoking the fire, feeding the engagements of the business. Because if you just do one or the other, then you're either going to lose engagement or you're going to lose the larger potential.

So I think, short answer to your question, the role needs to be set up for success. I think that within the first 90 days, a CDO needs to quickly assess the company, the situation, come up with an initial version of a road map, use a rigorous prioritization criteria to come up with a hopper of quick hits, data-driven quick hits, as well as a few big transformational projects that get the executive team and the board of directors excited.

MIT’s International Society of Chief Data Officers
Inside Out Security: I want to learn more about your role at ISCDO at MIT. Can you tell us a little bit more about this organization and some of the goals and objectives?

Richard: Sure. So the MIT International Society for Chief Data Officers, it came out of several conversations that a few of us in the community were having with a couple professors in the Sloan school.

Really, think of us as the IEEE for chief data officers. So, not affiliated with a vendor and really having more of a code of silence around our meetings. Place for folks who...chief data officers, chief analytic officers, really, think of the leaders of the data and analytics for $200 million plus companies, is more or less the range that we're seeing.

To come together and roll up our sleeves and talk about what's worked, what hasn't worked, which vendors have delivered, which teams within which vendors have delivered, and what are common challenges.

It always amazes me.

…. From MIT, we went down to DC, and we were talking to chief data officers in the public sector. And I walked into those meetings thinking, "Oh, these people are going to have nothing in common with us in the private sector."

They were talking about the challenges of data integration, and the challenges of driving quick hits, and getting engagement. And in the end, it amazed me at how similar so many of our challenges are, and just how many really of the best ideas, or I call them jiu-jitsu moves kind of ways of overcoming these challenges as an executive, really come from outside of our industry.

So I think a lot of...it's important to obviously work with folks within one's industry to think about, specifics to regulatory issues, for example.

But I think that it's equally critical to look outside of one's industry to find best practices in areas that other industries may be a little further along on some issues, and maybe they figured out something that we don't.

So MIT ISCDO right now, I think we're on at around 150 executive members, would invite anyone interested, who thinks they qualify as an executive leader of data and analytics in companies have at least $100 million or so in revenue to apply. And we do screen all the potential members. And if assuming that somebody is qualified to be a member of our community, we would absolutely love to have them join us.

Inside Out Security: Thank you so much, Richard. I'm wondering, if people want to follow you, how can people contact you or follow you on twitter?

Richard: Yeah, sure. So I would say, first, iscdo.org. My twitter handle is @reWendell, and my email address if anyone wants to get a hold of me is [email protected].

Inside Out Security: Thank you so much, Richard.

Richard: Thank you, Cindy. I enjoyed talking with you. I look forward to staying in touch.