 It's theCUBE at the MIT Chief Data Officer and Information Quality Symposium. With hosts, Dave Vellante and Paul Gillin. Hi everybody, we're back. This is Dave Vellante and Paul Gillin is with me. We're here with Jeannie Ross, who's a director and principal research scientist at MIT and has written an article sort of that's got a lot of attention lately. It's called, you may not need big data after all. Of course, we always talk about big data on theCUBE and we go around to all these shows. We like to extract the signal from the noise which sort of came from the big data world. Well, first of all, Jeannie, welcome to theCUBE. Thanks very much for coming on. Thank you for having me. It's great to be here. So what was the motivation for this article? You're sure you got a lot of buzz around it and a lot of people probably have said, what do you mean we don't need big data? So where did this come from? The motivation was a concern about the hype. I was worried because people would say to me, oh, we have a new role in our organization. We gave this person $2 million and told them to do something fabulous with data and I thought it sounded like the recipe for disaster. So we went out and looked at companies getting real value from data and our sense was that more often they were taking the little data that was all over the place and they were using it really wisely. So I have to tell you that title is not mine. I wrote to HBR and submitted an article that basically said people can work smarter with their data and they said, great, we'll call it. You may not need big data after all. It's their title. So Paul, you came from the world of titles, right? You know the title is everything in that business, right? Well, I also know that hype is everything. This year, it's been all about big data and there's a lot of money being thrown at big data, huge market valuation, big data companies. In the moments before we went live, we said that you're playing the role of spoil sport. How real is the hype right now? And when all this shakes out, do you think that this is a major new way of doing business or are we in fact going to distill it down to the value that you were talking about? The value really is in just learning how to use the little stuff. Actually, I think you grow into big data. I really do believe the opportunities are out there. What's interesting is they've been out there for a long time and some companies have used them brilliantly. I don't think there's an oil company out there that wouldn't tell you big data has helped them decide where to put their next billion dollar well. And so if that's what you do on a daily basis, decide where to put billion dollar wells, I think you should be into big data in a big way. But for a lot of companies, it's about understanding their consumers better and to go out and get big data about those consumers before they're able to adapt to the little things they already know. Many times we know great things about our customers. We haven't figured out how to address them. So we go get some more insights. We won't be able to address those either. And I want to follow up on that because a lot of small companies are scared off by this big data phenomenon. They think we just can't do it. Is that misplaced fear? No, actually, if they think they can't do it, they probably can't. I think the thing that's limiting companies is their ability to respond to a great insight. It turns out you have a great insight. That's just the easy piece. The next thing you have to do is change your organization to take advantage of it. For many companies, that is a Herculean task. They've never learned how to transform their processes. And these new insights are basically demanding the same thing they haven't been able to do before. So when we have conversations with executives on theCUBE, a lot of times we'll ask them, are you a data-driven organization? Oh yes, we're a data-driven organization. Is that an illusion that there are a lot of data-driven organizations out there? I think so. I think what we all are truly becoming is metrics-driven. And we are collecting more data towards our metrics. And I think this is all a very positive thing. But I don't think most companies are yet data-driven where they say, what does our consumer want? Let's take the data we have and respond to it. I just don't think that's the inclination. So metrics-driven would be a division optimizing on a particular data point, whether it's, let's say bookings or customer retention or whatever it is. That's very popular. Right. But that's not data-driven. No, and here's the thing. I think we often find that we can respond to certain kinds of data, but to really be a data-driven organization, we start to ask ourselves, well, what data really matters? And right now, what we see is a lot of companies that have been very product-oriented, saying, oh my goodness, it's really all about the consumer. So one of the companies we studied was Foxtel. It's a pay-for-TV company in Australia. And they said, we had all the data in the world about our products. We knew which ones sold. We knew who bought them. We knew what stations people wanted to watch. What program they wanted to watch after the prior program. But we started to realize that that didn't help us make decisions. If we could look at the results, but it didn't tell us what to do next. When we went back and said, well, let's learn about our customers. And we started to understand our customer segments. And we started to understand what these different segments like to watch. Then we said, oh, okay, that would be really data-driven because we can anticipate their needs. But you know what happened? They realized they would actually have to redesign their entire organization around customers. And they said, that was just incredibly hard to do. They didn't have the stomach for that. So do I need a CDO to be data-driven? I think that's such a great question. My sense is no. But I'm sure there's exceptions. I am sure there are places where it's just they get the right person, they're in the right position and it makes all the difference. But the risk is you're going to hire a CDO and say, oh good, somebody will take care of data. I don't have to worry about it. Go figure it out. Exactly. I wanted to speak to one of the examples you used in the HPR article, which was Southland and 7-Eleven, and how they really changed their business not by creating big pools of data, but by investing more decision-making in the employees. Why did you choose that case study? What does that teach us? That case, I think, is my favorite. This is a company that said, okay, we have 200,000 part-time sales clerks. They know the customer better than any computer ever will. And what we're gonna do is we're going to give them good data on what's sold. So we're going to give them data so they can become data-driven, not just instinct-driven. But we expect them to supplement that data with their instincts, with what they've learned by talking to customers day in and day out. And then they said, we will coach them. We will teach them how to use data. And I think that's the piece that most companies don't realize is key. I mean, you can give lots of people lots of data and say, okay, guys, be data-driven. But if you don't teach them how to be data-driven, most people will not know how to respond. Well, in fact, most people are not numbers-oriented. They're not data people. So should your goal be to teach them how to be data people or to teach them how to use data without having to think about it in analytical terms? Well, yeah, I think you don't want to say, now I'm gonna teach you how to be data-driven. The beauty of the 7-Eleven approach is these counselors that come in twice a week, every week, and say to each individual sales clerk, what was your hypothesis last week? And they actually know the answer when they ask the question, because they can pull out the sales. Their hypothesis is that the things they ordered would sell. And then they can say, how did you do on your hypothesis? And they know the answer to that. They know if it's sold or not. And then they ask the magic question, what can you do better next week? And that's when they sit down and they have a conversation about, well, we know this sold, we know this didn't. What have you heard from people? What do you see in this data? What's going on in some other stores? Let's figure out what you might do next week. And many people will respond to that by becoming without knowing that data-driven. I think it was Gartner that came out and said that the CMO is gonna spend more money than the CIO that started the Big Data Hyper. The modern era, the last six months of the Big Data Hyper, I should say. So one of the best practices that you put forth in your article was you got to have a single, I'll call it the single version of the truth. I think you called it a single source of data truth. A lot of people say with this move toward so-called Big Data. And I look at Big Data as a different approach to analyzing data that can't be stuffed in a single box. That's what's different and that's what Google sort of created and Hadoop created. And many have said we're further away from a single version of the truth than we ever have been because of that. So help me square that circle. I think it's a great question. And what we realized as we were studying this is there's actually two kinds of data that are really important here. One is the performance data. And that's where you run into real issues because everybody keeps their own. I think we included in the article the example of Epna where Ron Williams comes in, the company has lost $600 million and every one of his business unit has a spreadsheet that says my division made money. And he says, okay, I'm not using your spreadsheets. And that's where you have to say this is the data we're using and not worry about whether it's accurate or not because it'll be made accurate if we're all required to use it but rather say this is what we'll rely on and then everybody knows what the metrics are that matter. So that's more around performance data, that single source of the truth, but you also need really good data about your customer. And to your point, it won't all be in one place but we will all agree on what data we're relying on and how we're going to keep it up to date and how we're going to make it reliable. And that data also, single source of truth may not be the right start. Well, essentially you're saying you've got to have a single source of truth for the different types of business processes that you want to optimize. Yeah. Yeah, okay, well that makes sense. So is the role of the chief data officer then not so much to manage data but to teach people how to use information in ways that are meaningful to the business? I think so, but our risk here is that we'll all think that the chief data officer will take care of that. One of my favorite examples of really good use of data is a little company called Protection One. It sells security systems. And the CEO said, I want people to start understanding their real results and what actually will make this business successful. So what he does every morning is check the numbers. He has a scorecard. He calls all of his direct reports and they talk about anything that is a potential concern in those numbers. And he always ends with, what do you need from me so that you can make this better? And then he said, I am modeling the behavior. The next thing they are to do is call their direct reports and ask the same questions about what the data is saying and how we should interpret it and what are the potential issues and what do you need from me? And this goes all the way down to the branch chiefs who are responsible for bringing in their people every day and saying, this is my understanding of how you did yesterday. What do you need from me? And it changes the culture. It takes time. He said he was actually surprised how long it takes but it does change the culture. I want to come back to the CDO question. When you consult with organizations and they ask you do I need a CDO? It's an it's depends question. But when you get to the point of yes, the CDO makes sense in this organization. Certainly financial services and a lot of government organizations and healthcare are moving in that direction. Do you buy the premise that the CDO should be independent of the IT organization? Ah, well it does depend exactly what you're trying to get from the CDO. My guess is yes, that if you're looking for someone to take a very targeted strategic initiative forward. Here's some data we want. Here's how we want to use it. I would imagine it's if someone you want to partner with IT, not put in IT. But you know these things I think do depend a lot more on the people. And we've talked to all kinds of CIOs who are not reporting to the CEO which is supposedly what every CIO would want to do. And they say well it doesn't matter who you report to and it matters who you influence. And we've found many of the best CIOs are actually not reporting to CDOs but they're having enormous impact. I think that would be true of the CDO. You don't want to be a CDO if people aren't very clear about what you want to accomplish and give you the authority to accomplish it. It won't matter what organization you're in if they haven't set it up that well. And you feel unbalanced the CIOs will get over it. Oh yeah. I think CIOs more than any thing want to have a positive impact. So if they have somebody who says I am going to make data valuable in this organization. You've got a CIO who's very excited. All right I think we got time for one more here Paul. I just want to go back to an example because I think it's so illustrative. Everyone having their own version of the truth that together rolls up to a big lie. Doesn't it seem the companies need to overcome the culture against failure, the prohibits failure and then encourages CYA mentality so that everybody has their own version of the truth because they don't want to fail. Isn't that where we have to start? We're really going to have to start there and that is going to have to be modeled at the very top of the organization. This has to be about we need to make mistakes. Otherwise we won't learn fast enough. Let's understand the mistakes we've made, what we expect to learn from them, how we're going to go forward. And yeah I think we have to not only allow mistakes but I think we have to encourage mistakes. I think we want to talk about how we get smart about making mistakes and moving forward but it's going to have to start at the top. You can't take somebody in the middle of the organization and say, you know, don't be afraid to admit you made a mistake. I don't think that goes well for a lot of people. All right, Jeannie Ross, we have to end it there. You have to get back to the main stage. Thanks very much for coming on theCUBE. It's really a pleasure having you. Thank you, it's great to be here. All right, keep it right there everybody. We'll be back. This is theCUBE. We're live at MIT in Cambridge. We'll be right back.