Tech Trends Roundtable Dr. David Douglas and Dr. Manish Agrawal on analytics Episode date: 01-03-2012 [ MUSIC ] LANINGHAM: This is the developerworks podcast. I'm Scott Laningham. developerworks recently published the results of a Tech Trends survey by IBM drawing on the input of the developerworks community of IT professionals, business partners, students and university faculty. The survey focuses on trends in the area of business analytics, cloud computing, mobile computing and social business. Today we talk about the business analytics results of the Tech Trends survey, and we're joined by Dr. David Douglas, who is university professor of information systems in the Sam M. Walton College of Business at the University of Arkansas. Also Dr. Manish Agrawal, who is associate professor of information systems in the College of Business at the University of South Florida. Welcome to you both. So grateful to have you here today. DOUGLAS: Thank you so much. Glad to be here. AGRAWAL: Great to be here, Scott. LANINGHAM: Now, before we drill down into the very interesting results of this survey, I'd love to just get thoughts from both of you about, in an overarching sense, about the state of business analytics at this point in time where we're at. I mean, are we on a rapid ascension curve in terms of technological change? What exactly is going on right now? And Manish, let's start with you. AGRAWAL: Yes. What is going on with business analytics is that companies have invested considerable sums of money over the last decade or two in digitizing all their data, getting inputs and investing in hardware and software, and professional skills within their organization. So now there's all this data and there's all this processing power within organizations, and now this analytics component has the promise of huge return on investment with relatively little additional technical investment. And so that is what is driving the change. We now have faster processors, so we have on desktops the power that was once on mainframes. We have simple tools like Excel, et cetera, are extremely powerful now. So that is what is driving much of this trend, in my opinion. LANINGHAM: Okay. Dr. Douglas? DOUGLAS: To add to that a little bit, the growth of data, IDC projecting growth of data at 60 percent annually since 2007. So what you're seeing is a tremendous increase in data, and the tools to analyze the data are getting better and easier to use. So the analytics is actually being utilized whereas in the past it was mostly for
strategic decisionmaking, it's now becoming more and more support for operational decisions. LANINGHAM: Well, that helps kind of set the stage for the times that we're in around this kind of technology and methodology. And clearly there's so much going on that is driving change that I think that is why it was such an interesting time to do this survey. So let's talk for a minute about some of the results of the survey, and I'd love to get your thoughts about what these things imply, if that's okay. Let's start with this one. Within 24 months, the survey says 90 percent of respondents believe their organizations will be using analytics; 45 percent of respondents see growth of 25 percent or more in analytics work over the same period. Talk about the popularity of this field from those kind of results we're hearing and how that...what you were talking about earlier is actually translating into this enthusiasm. DOUGLAS: Well, as more and more cases are available where they've been very successful and the decisionmaking it's making companies more competitive, because as has already been mentioned, the data is being captured almost automatically and growing at a tremendous rate of speed. The fact that you can now relatively easily, with the software and the vendors making it easier to use and moving where you do that on a more timely basis to support operations, I think the growth rate is probably pretty much on track. LANINGHAM: Manish? AGRAWAL: Yes, essentially the same idea that the return on investment is very high. What you really need are people who are able to bring their insights into business, who can leverage the technology. But otherwise all the raw components are well in place. There is also, as Professor Douglas mentioned, there's also organizational recognition that this has value. So I'm not surprised that data mining and analytics mining is getting so popular. LANINGHAM: We often talk at IBM about the cultural change that has to happen for all this stuff to be deeply embraced. Where do you think we are on that curve in terms of widespread cultural realization and adoption in organizations to move to more of this predictive thing as opposed to simply gathering a bunch of data to have in a storehouse in case we want to look at it for whatever reason? DOUGLAS: I think the culture is moving more toward analytics as a driver for decisionmaking. The days of doing everything from an intuition, which used to be the big enemy in the organization, as the younger group of people come through I think you're seeing that the culture is in fact changing. And that's not an easy thing to do. There's sure still a lot of decision- making that's called gut feel kinds of things and not based on data. But I do think it's moving more toward an environment and a culture, if you will, where they'll have to have analytics to support decisionmaking.
AGRAWAL: If I could add to that a couple of things -- take advertising, for example. We had no available tools to figure out what was happening in a campaign. You have a Super Bowl ad; what does it do to your company? It was very difficult to measure that. But as these days, we do class assignments now here at USF and University of South Florida and I'm pretty sure they do the same thing at Arkansas as well, we have students in undergrad programs who are using tools from companies to figure out what is the increase in visits to websites; how many products did they buy as a result of an online advertising campaign. So it's now become a fairly, it's becoming engrained. And if you look at the people who are doing these kinds of classes, not all of them are technical people. There are students from other majors who are taking these classes. They're becoming aware and they're going into companies and bringing out these ideas. And here again at the University of South Florida, if you think of education as a market, as I sit on different committees here I see faculty members from the arts colleges, the fine arts groups, they all have this awareness. They all know that somewhere there is data that will help them figure out what employers want. So in terms of culture, yes, I'm seeing it everywhere. And in fact, in terms of awareness, I would say that everybody's as aware as we in technology are. That's my impression these days. LANINGHAM: Okay. Now, of course you two are professors and you're at the universities, and the training gap is a perceived problem for some of this shift. And let's talk about one of the results in the surveys there. It said 61 percent of respondents are concerned about their organization's abilities to integrate business analytics into their existing architecture; 55 percent report concerns about training on the use of analytics capabilities. So let's talk a little bit about how these concerns are being mitigated, can be mitigated, and also the type...what you all are doing at your universities, but also the kinds of things that are going to have to evolve for it to do an even better job at that. David, would you like to start? DOUGLAS: Yes, kind of put it in perspective. Of course, we're trying to do education and training for students who are going to go out and work and be players one of these days. So we already have a fourcredit business intelligence certificate program, that's a fourcourse sequence. But in addition to that, because of the industry people's needs, we've also developed a non-credit business analytics certificate, which consists of three two-day programs. So we think those kinds of things are going to exist in many universities and areas to help the people that are already out there, but we also have to make sure that we do it for the students that are going out so we have both a credit and a non-credit orientation towards business intelligence and business analytics certificate. LANINGHAM: Okay. Manish?
AGRAWAL: Yes. See, what happens here -- and I'm glad you brought that up -- there are two things. When you're trying to do this technology in business, there are two components: one is, you need people who understand the business. They need to know what does the business want and what specific deliverables can the technology offer business. So you want to ask this question: what can analytics do? And it is not fair to assume that students know that off the bat. Many students do not have this large corporate experience, and at universities and at educational institutions, we see an obligation to educate students on that -- so, what can analytics do to a business? And we at universities are pretty good at that, and that is largely our focus, what can analytics do for a company. And we take small local organizations that are willing to put up with those students... And they see a lot of value from the presence of our students and we get students to do those little projects and the students become aware of some tools, some business needs and the gap and in general how these tools can be used to benefit companies. Now, there's also a role for companies themselves that are developing products and they have certifications and other things. So, for example, IBM, you have your Cognos, SPSS "Clementine" [Modeler], I think it's called. So these tools, there's a lot of need for people to really have expertise in these tools because tools expertise makes the difference between somebody getting the job done in two hours versus five days. So we at the universities can do a good job educating students on what can analytics do to a business and companies like IBM can do a great job at educating students on how to use the tools right. LANINGHAM: Do you see people drifting in a specific direction in terms of analytics? And I ask that because one of the things in the survey, it talked about that analytics seems to be incorporated throughout the business -- financial analysis, strategic planning, customer relationship and service being the three most common uses that were cited in the survey. But that we can see from this maybe that no single dominant process is surfacing where analytics is being deployed. Do you think that's going to change? Or do you think among those areas that students and maybe in your case in terms of how you encourage them, is there a particular direction that's stronger 00:12:26,486 --> 00:12:28,896 than another from their focus? Or do you think that this kind of cross-enterprise use of analytics is going to be the trend? AGRAWAL: If you see one common factor among all those -- finance, customer relation, service, et cetera -- is that they're already digitized. In each of these segments, the data is already available. So the managers in each of those segments have been wondering all these months and years, what can we do with this data? We have terabytes of data, and what can we do with it? They have been scratching their heads for a long time and they see analytics and we have some models and we have some frameworks for dealing with these large volumes of data, and so that's where people are going in, where the investment is relatively little and people see huge payoffs.
LANINGHAM: All right. David, would you like to chime in there? DOUGLAS: Adding just a little bit to that, as indicated, the capture of the data is almost automatic. Digitation, if you will, as Professor Agrawal mentioned, is across all disciplines, all functions in the organization, and we also realize that organizations are integrated as well. So you're going to find that these pressure points are going to be working on all these at all these different locations throughout the organization because the data is there, the software is easier to use, they're more aware and there's also indications that it's beneficial. The ROI's there. LANINGHAM: Are there areas -- maybe a follow on to that question and your answer -- are there areas of the business that you think that are not as digitized now that will be coming on in the coming years that will change for them as well? I mean, does something jump to mind from that standpoint? DOUGLAS: There are some areas that are difficult still to analyze. It's being done. But as we move into more social media, the thing about social media is that it sort of permeates everything, and many companies have social media in their company as well as things like Facebook. So there are probably some areas that haven't...i don't know what they are, haven't taken as much advantage of that as they possibly could. But I think they will in the future. AGRAWAL: If I may add, as Scott put that question I was trying to think what areas there may be. Obviously there's huge areas. And once again going back to my role as university professor, you know, one of the biggest things that we worry about, we are concerned with are students learning? What are they learning? And if they're not learning, what can we do to improve that learning? These are very subjective. So there's a lot of information that's difficult to capture. LANINGHAM: Sure. AGRAWAL: And very important information. So, yes, I mean, the way things usually work is that once you find a solution to a problem, people become aware of the solutions and will hopefully see some of these very important but very subjective issues also being digitized in some form. There's a lot of concern on the campus about these issues. LANINGHAM: I wonder, too, predictive analytics may actually...it may work in reverse where it actually surfaces problems that we didn't even know needed tracking and solving sometimes, right? Very interesting space. Another point in the survey before we kind of do a wrap-up was 87 percent of respondents believe open source technologies like Hadoop and Linux will play key roles in the development of business analytic solutions. What are your thoughts about that? DOUGLAS: Well, playing key roles, that doesn't mean that commercial software will go away. But for example, IBM InfoSphere BigInsights actually incorporates Hadoop as part of that appliance that they use
for doing business analytics. And so you're going to see those things incorporated. It may not be that everybody as the user is doing it, but I think they're very key...they're going to be key, important areas that they're incorporated into other appliances and other things. So I do think they're going to be critical going along. But it may not be one as obvious to the people that are actually using it because it's built into an appliance. LANINGHAM: Dr. Agrawal, thoughts? AGRAWAL: I don't have much to add. Professor Douglas has captured essentially what I would say. If I may add, I think if I would just chime in a little bit, it would be that these open source platforms are becoming areas where the enthusiasts, the people at the leading edge, graduate students, faculty members, interested people...can try out their algorithms, their models and procedures and become testbeds and eventually for corporate use the proven models, the proven methods, et cetera, can be incorporated into commercial products. I think that's my...if you asked me what would happen two years from now, that would be my expectation: that in the real world in production we would see commercial products because they're tested, they're reliable, they're scalable. There's support behind it, there's documentation. Open source has a huge problem in terms of documentation. Nobody likes to...it's not fun to document. It's a pain, it's boring. And it's work and not rewarding work. And the problem is you have thousands of users that don't even know how to get started. So I see that as being the equilibrium, where the proof of concept happens on open source platforms, and once these concepts are proven they get integrated into commercial products which will then move the world forward. LANINGHAM: Research and development is a critical part of any development of technology and progress point and so maybe that's the most important way to view open source in this, is kind of an extended R&D field for all this work, right? AGRAWAL: Yes. And just to highlight what Professor Douglas said, the keyword "appliance." I think that's really great. The more we make these technologies work like appliances, turn them on and they work, reduced learning curve, things like that, that takes a lot of effort to take a technology and make it into a product that works like an appliance. And yes, that's what I'm hoping to see happen. LANINGHAM: Dr. Douglas, were you going to add something there? DOUGLAS: I was going to add to that a little bit that there's always going to be a group of people that utilize the open source. I happen to know of a company, for example, there are a couple of companies that are using these things directly. But I do think that they're certainly not going to push commercial software away. It doesn't appear to be possible. But there's going to be a small group that are
going to buy into it and work with these and make that their platform. LANINGHAM: Well, why don't we wrap it up, if we could, with your thoughts about kind of the future of analytics. And one way that it was put to me by a fellow developerworks staffer that I was talking with was considering capabilities to analyze large amounts of data like IBM Watson that we're all familiar with that project. What does that tell you or how does that...you know, that's the question. I mean, where do you see this all headed? DOUGLAS: I think what you're going to see is this kind of technology permeate things that are beneficial to society, from healthcare to water supplies, to traffic lights, to sustainability and so forth. And in addition to the organizations, those are going to continue. Within the organizations the hot new buzzword is called big data which talks about velocity, variety and volume; and like the appliance I just mentioned from IBM, the BigInsights. And you're going to see that term used a lot and the companies working that which means if you're working on data that's not what we're used to, which is data in databases, you're going to be working with multistructured data rather than data that comes out of databases. And I think you're going to see that trend for several more years until we get really good at that. LANINGHAM: Manish? AGRAWAL: There's not much to add to that. When we talk about Watson, the most obvious thing is natural language processing. We started with this idea that there's a lot of structured data in organizations and it's very interesting to end this way. There's also a lot of unstructured data when Professor Douglas talks about healthcare, you have all these medical knowledge and doctors' minds and how do you capture that and how do you use it for diagnoses and how do you use it to improve diagnosis from physicians assistants and professionals like that. So the opportunities are endless. If you ask about the future it will be about natural language process, unstructured data, this big data, and things like that. LANINGHAM: Well, I'm so grateful that you both have the time to spend with us today for this discussion about the IBM Tech Trends survey. Really appreciate it very much. Again, my guests have been Dr. David Douglas from the University of Arkansas and Dr. Manish Agrawal from the University of South Florida. Read more on this topic and others covered in the IBM Tech Trends survey at ibm.com/developerworks. This has been the developerworks podcast. I'm Scott Laningham. Thanks for watching. [ MUSIC ]