Using models for transformation purposes

Photo: William B. RouseWilliam Rouse of the Tennenbaum Institute at the Georgia Institute of Technology explores why most business transformations fail and how modeling and simulation can help.

Interview conducted by Bo Parker, Karen Schwartz, Vinod Baya, and Terry Retter

Dr. William Rouse is the executive director of the Tennenbaum Institute at the Georgia Institute of Technology, which focuses on researching and developing the knowledge and skills necessary for enterprise transformation. In this interview, Rouse provides his perspective on the value of enterprise modeling to transformation, a view informed by the institute’s extensive research on the history of transformation efforts.

PwC: Can you tell us a bit about your background?

WR: I left the university for 13 years and started two software companies, the second of which had a suite of strategic planning tools for new product development and technology portfolio management. During that time, I worked with about a hundred companies, so that’s how I got immersed in this—I saw how difficult it was for the companies to fundamentally change.

I did a lot of background research. In fact, my book Start Where You Are goes back about 200 years and looks at three industries: transportation, computing, and defense. Of course, computing in the 1800s was cash registers and things like that, not computers, but the bottom line was that when it came to fundamental transformation, almost everybody eventually failed.

There are only a limited number of big success stories, and everybody of course wants to be one of those success stories. For transformation to succeed, there has to be a perceived or experienced value deficiency, or people don’t do anything. It’s really rare that people just decide to change, either as enterprises or as individuals. The value deficiency can be something that is relative to what you thought you could accomplish. In other words, you may not be not failing as a business, but you are not getting where you would like to get.

The way you turn companies around is by understanding the work processes in the organization and how value is created for the organization’s constituencies or markets. That often is very difficult for people to do. In healthcare, for example, people don’t really think in terms of work processes; they think in terms of functions or specialties. So healthcare is a particularly thorny problem, because beyond people’s reluctance to change, they also have difficulties thinking about it in a way that would enable change.

Beyond value deficiencies and work processes, a big issue is management decision making—executives’ and senior managers’ abilities, limitations, and inclinations to make decisions. In a lot of the companies I have worked with, the executives were not willing to make a fundamental change decision until the need was so patently obvious that everybody would go along with it, and at that point, their resources and time were lacking.

Beyond management decision making is the social network of the organization. The social network can be a great enabler of change, or it can be like an immune system. We found that large organizations tend to get insular over time. Because the majority of the people are not customer facing, they don’t see the outside world, and that begins to reinforce their beliefs about the way in which the organization achieved its success and will sustain that success in the future. Often, these perceptions aren’t valid, or at least no longer valid.

So, we study those four components: value—what it means, the nature of value deficiencies, and how these deficiencies are understood; work processes; management decision making; and social networks. And we feel that those four pieces are the keystones to successful transformation.

PwC: How does the transformation methodology you’ve developed differ from other methodologies enterprises generally use?

WR: In contrast to what I perceive to be the typical IT-driven approach, we tend to work top down. We start from business issues and understanding the strategic intent of the executive team and the board. And then, we focus on how you can develop an enterprise architecture—not necessarily an IT architecture, but an enterprise architecture—that can enable you to pursue the intent that you have. And then we address how you understand the relationship between the as-is enterprise and the to-be enterprise, which leads to IT eventually, as information is often a key facilitator for all this.

We have been involved with many companies who walked away from very substantial IT investments when they realized they hadn’t done sufficient thinking at the beginning from the top down. When you start top down, in some ways the interactions are easier to find, because you are looking at the overall context of the enterprise and how it’s put together. Engineering and manufacturing, marketing, and customer support—how do those all fit together in terms of the value that the customer perceives? The interactions of these processes are critical.

PwC: In your “Models of Complex Enterprise Networks” paper, you point out that the enterprise requires a balance between a holistic view and a reductionist view. Is that one of the main reasons why transformation is so difficult?

WR: Right. We have a wonderful case study going on right now in healthcare, and we are totally immersed in this, working with our partners in Mayo Clinic, Emory University, and other places. Everybody—all the different providers, the device companies, the pharmaceutical companies, the insurance companies, the hospitals, the integrated health systems—they are all seeing it from their perspective and arguing for change or a lack of change from that point of view. There isn’t really anyone taking a holistic perspective.

We can take that broader view to find out where you can actually gain leverage. In the end, you do have to do some reductionist work—you have to get down there and make stuff happen—but the question is can you pick the places where you will actually get the biggest leverage in the whole enterprise, and for that you need a more holistic view. It’s been very rare to find any instances where people have totally reengineered a large enterprise from the bottom up—even tried, never mind succeeded.

“The bottom line was that when it came to fundamental transformation, almost everybody eventually failed.”

PwC: Is getting a holistic view difficult? How are enterprises getting better at it, and does modeling fit in at all?

WR: Sometimes you gain the holistic view by talking with key stakeholders of the executive team and key customers, and it can initially be a fairly qualitative model. When we try to become quantitative, often we start with some simple analytic models that portray the relationship between the enterprise and its markets or constituencies. How they provide value, how value flows, the costs, and the revenues are associated with these flows. And from those simple analytic models, we get initial insights to some of the key tradeoffs.

Then we start scaling up the analytic models. Pretty quickly at some point, we can no longer deal with the models analytically, and so we have to go to organizational simulation to be able to simulate the enterprise. We have found success in simulating the social network in the organization—not just the business processes, but also who knows who and who relies on who. We have been able to show how you can make major improvements in the organization by exploiting those relationships. In the case of one major company we work with, we showed them how to use their social network to reorganize the flow of projects through the organization, and with zero investment, they could achieve a 50 percent reduction in time to market, just by exploiting what they already had.

We are modeling the work processes in the enterprise, but we are overlaying it with the social network, which then allows you to portray who works with who, who relies on who, and who will go to who to get additional information or help.

PwC: So the organizational simulation that you alluded to is leading to more of a complex adaptive system view of the organization?

WR: Yes. In healthcare, you really have to take the view of a complex adaptive system. That’s the only way you can constructively decide how you might address change, because you can’t command a complex adaptive system.

Once we go from our analytic model, which might be algebra and a few other kinds of calculations, there are three directions we can go: a discrete event simulation, a system dynamics simulation, or an agent-based simulation. For the Air Force, we’re working on how to combine those three simulations in a seamless way, rather than trying to force all the phenomena in the organization into one representation.

PwC: Would you compare and contrast each of those simulations?

WR: Discrete event simulation has the best off-the-shelf tools, tools that are very mature. There the key thing is the timing and flow of events, which could be people or products or information. In terms of control information, you are really looking at how capacities are allocated over time. By contrast, with the system dynamics approach, you are more concerned with feedback loops.

With the discrete event simulation, you might look for the steady-state, optimal solution to allocating resources. With the system dynamics simulation, you are looking for the time variations of the response. There are welldeveloped tools for this—they have been around a long time. However, they often don’t seem to be as useful to get a really fine-grain representation that the discrete event simulations allow, in terms of off-the-shelf tools. The agent-based approach allows you the greatest level of granularity, but the tools are not as mature.

We learned it can be very useful to use those representations and convert the simulation into a game. This way, executives can play with the organization and try things out. For example, we have a game called Health Advisor in which you manage 500 patients. You are not a doctor, you are just a health coach, and you are helping these patients get through the healthcare system. We are using that game to study how different levels of information and different levels of incentives impact what people do. With an online game, you can study a large number of people playing the game. Or, if you are only concerned with simulating an enterprise, you can do what-if experiments in the virtual enterprise first and then port them to the full-scale enterprise.

Once you get to a certain scale, you might want a mixed representation of discrete event, system dynamics, and agent-based simulations. Only recently have tools emerged that allow you to manage that well. One of the problems with these large models is that the maintenance of the model becomes an enormous cost. I was on a DoD [US Department of Defense] senior advisory group for modeling and simulation a few years ago, and they were spending a lot more money on the maintenance of agent-based models than they were in creating models, because they were so handcrafted. We need to get beyond that.

PwC: Where do emergent properties enter the picture?

WR: Usually we can spot some emergent properties from the modeling exercise. For example, in one modeling project we expected that as your workforce became more and more competent, even though they may be more expensive, you would get better performance. So, as a benchmark in this model, we had everybody be maximally competent. You could never afford to hire that workforce, but that was our reference. And then, as people played around with the model, we were surprised to find that in this company—at least in the mix of projects they had—a blend of half expert and half entry-level personnel got almost the same performance as having everybody maximally competent. We did not expect that, but it reflected the fact that many things that needed to be done in the organization did not require maximal competency.

PwC: One final question. Where in the organization does model-keeping reside? It seems to us that enterprise architects are the most active modelers within most organizations, but they don’t generally play a very strategic role. In your experience, where are you seeing the modeling capability emerge within corporations?

WR: We work with a lot of executives here at the university or in consulting, and the goal is to have the chief modeler, if you will, be pretty high up—a vice president or higher. I don’t mean that they are out there coding. I mean that they are so into this, they are champions for the model-based approach. Sometimes it’s the top person, sometimes it’s the CFO, sometimes it could be the chief technology officer. And, if all goes well, the model-based approach becomes one of the main ways the executive teams think things through.

PwC: Are you saying that this role is usually somebody with more of an operational or a strategic role? Not the CIO?

WR: Well, while this is a representative sample, our advisory board at the Tennenbaum Institute includes 18 people, all of whom are vice president or higher at major companies and some smaller companies. They are all very IT- and model-oriented, but are more likely to be CEOs, CTOs, or CFOs, rather than CIOs.