Hosni Adra, Product Manager and Partner, CreateASoft Inc., Naperville, Ill., USA
BY MARK INGEBRETSEN • PHOTOGRAPHY BY CHIP WILLIAMS
- Simulation programs enable users to model and forecast less tangible information such as the behavior of individuals and groups.
- Rather than outputting results as a stack of interlinked spreadsheets, simulation programs can display findings via a single interactive flow chart.
- Besides allowing companies to plan or rehearse strategies, simulations can work as team-building exercises.
- Someday, simulations may take on an even more important role, modeling a company's operations and offering lightning-fast forecasting.
GENERATION AGO, spreadsheet programs revolutionized the way organizations planned their futures. Today, simulation programs—offering an equally easy-to-learn interface—promise to take desktop strategic planning to the next level. While it's no magic bullet, this software can help executives answer important questions such as the optimal resource allocation among corporate divisions, how an acquisition might affect customer loyalty or what information flow within the organization best suits its core competency.
Unlike spreadsheets, which simply crunch numbers, simulation programs use sophisticated algorithms to deal with less-definable information—such as people's attitudes and behavior.
For a passenger airline, a standard financial analysis might predict the cost consequences of adding new service routes. However, simulations can ferret out answers where “numbers of people are changing through time,” says Kim Warren, founder and chairman of Global Strategy Dynamics Ltd., Princes Risborough, U.K. “If I served this passenger on this route in a way that they perceived as being of good value, what's the probability that the passenger will remain loyal to the company?”
Using these tools, management can see beyond the dollars-and-cents consequences of its plans. “We can put in a variable that predicts a merger's effect on people's trust levels and productivity,” Warren says.
Simulations can produce equally exacting results when they're unleashed on process-oriented problems, says Hosni Adra, product manager and partner with CreateASoft Inc., Naperville, Ill., USA, which helped Ashtabula, Ohio, USA-based Molded Fiber Glass Companies model its existing production process. The simulation showed how simply shifting its available production-line resources could save one man-hour per part.
The same techniques can be adapted to companies that move information instead of products. “Service businesses still have a production flow, they're just moving different types of parts,” Adra says.
Graphics, Bandwidth and Competitive Play
Whatever the problem, the software's graphical output makes the solution crystal clear. Rather than out-putting results as a stack of interlinked spreadsheets and charts, simulation programs can display their findings via a single interactive flow chart. “You've got a visual map of how everything is connected to everything else,” Warren says, “and you see what's happening to each variable over a defined period of time.”
These representations are called bandwidth boosters, according to Michael Schrage, author of Serious Play: How the World's Best Companies Simulate to Innovate. In fact, bandwidth boosters can be more effective if they're produced by a group of managers working together on a simulation, an idea that may be catching on.
European companies increasingly want simulations that involve groups, says Daniel Paul, a founder of Logicia, Neuilly, France. That's because as much as they want to rehearse strategies, the companies also look to build teamwork and give participants a better sense of the organization's entire operations.
Paul says the participants who take part often are fast-tracked engineers in their 30s and 40s who need a generalist background to complement their technical know-how before moving to upper management. The simulations allow them to receive that training and build relationships in the hands-on environment.
What we do is study an industry and find out what makes it tick, and then we build the algorithm to describe what drives sales.
Professor Emeritus, Florida Atlantic University,
Boca Raton, Fla., USA
“We call it discovery learning,” Catherine Rezak, president of Paradigm Learning, Tampa, Fla., USA, told the Tampa Bay Review. “If you have complicated information, it's better if you can find a way to get people involved.”
Jerald Smith, professor emeritus at Florida Atlantic University, Boca Raton, Fla., USA, and author of several simulation programs, adds the element of competition to his group simulations. Independent teams work through problems. Afterward, the winners discuss the strategy decisions they made, while the also-rans share their mistakes. “The losing team usually just throws money at the problem, and the winning team is very consistent and organized. They have a flowchart at least in their mind of what they want to do,” Smith says.
Under the Hood
How well these simulations work depends on the imagination and programming abilities of their creators. “What we do is study an industry and find out what makes it tick, and then we build the algorithm to describe what drives sales,” Smith says.
As for the algorithms themselves, developers generally are tight-lipped on specifics. Some algorithms may be similar to those used in economic modeling. Still other simulations add Monte Carlo analysis, which can analyze thousands of potential outcomes to identify those most likely to occur.
That computing power can be essential to electric utilities that need to know, for example, precisely what to charge for power, how much to generate and when, based on changing market conditions. RiskAdvisory, a Calgary, Alberta, Canada-based division of SAS serving the utility industry, used a Monte Carlo simulation-building kit from analytical software firm Palisade Corp., Newfield, N.Y., USA, to identify the risks utilities face. The software can pour through thousands of calculations in less than three hours and provide an answer critical to its clients' decision-making process. The software enables planners to focus on operations at an individual plant and create a distribution of probable outcomes based upon 10,000 iterations per month throughout a specified year.
WHAT'S YOUR GAME?
“If you want to understand simulations, the only way to do it is to become familiar with today's computer games,” says Clark Aldrich, author of Simulations and the Future of Learning. Indeed, using business simulations to plan strategy would seem natural for those of us raised on computer games, such as SimCity. As with computer games, simulations fall into several categories—all potentially useful, depending on a company's particular goals.
Designers sometimes create game-like simulations based on real-life or imagined business case studies. Participants take on top management roles and attack a specific challenge. How can a small hotel chain expand nationally? Should a computer firm merge with its rival? Some generic simulations are geared toward individuals, while others foster communication and teambuilding through groups.
Game designers can take their generic models and adapt them to a specific company problem. Though prices can run from $5,500 to $50,000 or more, an ill-conceived plan could cost much more. Customized simulations allow management to test a strategy before committing resources.
That's especially true of process-oriented simulations aimed at optimizing factory-floor or information-flow operations. Managers can try out solutions before actually tinkering with a company's production lines.
Some group simulations seek to intensify the learning experience by putting everyone in the same location. Web simulations are gaining ground on this approach because getting high-level participants in one place is always difficult, and far-flung enterprises with teams from different national and cultural backgrounds need to develop collaborative methods via the Web.
Neural networks and other varieties of artificial intelligence software are another component of some simulation programs. Often, they're best at modeling ongoing processes. The software works by continually searching for an optimal way to predict or model an action—how crowds move through an airport, for example. More important, because systems, especially those involving people, rarely remain static, artificial intelligence programs typically are able to adjust their answers as new information becomes available.
Charlotte, N.C., USA-based NuTech Solutions Inc. helped Southwest Airlines, Dallas, Texas, USA, improve its cargo handling procedures through an analysis of routing decisions made by Southwest personnel. The software was able to delve into the many complex interrelationships that existed throughout the airline's cargo distribution system, enabling planners to see how bottlenecks could be avoided. This systemwide approach to speeding cargo flow proved superior to efforts aimed at improving efficiencies at individual facilities.
As a result, Southwest was able to boost cargo shipping revenue by $10 million, cut package handling by personnel on the ground by 20 percent, while cutting the number of packages that needed to be handled more than once by 75 percent, NuTech reports.
Care and Feeding
These hugely successful outcomes aren't possible unless the simulation is well-tuned to the task at hand. The program's construction should provide a data-entry interface that makes it easy to supply the information that's needed.
When constructing the model, restricting the number of variables to a minimum often works best. “You could never put all the variables into any model or else it would be as complicated as the real world,” Warren says.
To handle these tasks, companies often tap advisory professionals in companies or internal consultants, Warren says. “Larger packages can be intimidating to average people, and they might be built by the people within companies that build spreadsheets now.”
Increasingly, programs are being designed that anyone can use and understand. “If people can draw a diagram depicting the process flow of their organization, they should be able to create a simulation for it,” Adra says.
Do They Really Work?
When applied to specialized problems, simulations seem uncannily accurate at finding solutions. Take the problem of downsizing, an unfortunate fact of life in many organizations today. Companies may employ an ad hoc approach to cutting staff, delegating the task to managers down the line. But the Pentagon used a simulation program to see if it could use fewer staffers in a key department at the Redstone Arsenal, Huntsville, Ala., USA. The software from Lanner Group Ltd., Reddtich, Worchestershire, U.K., enabled planners to see what specific department tasks were the most time-consuming, thus allowing them to focus efforts—where feasible—on automation. The program enabled the Department of Defense to cut six positions at an annual savings of $300,000, and the cost of the labor-saving solutions could be made up in one year.
When simulations attack larger problems, their ultimate validity comes into question. The problem is compounded because companies often avoid sharing information about their successes. “If they think they have something good going, they don't tell each other about it,” Smith says.
Warren says only the most elaborate, painstakingly constructed simulations can ever hope to provide strategic direction on their own. In general, the solution proposed by a simulation program should be just one among several indicators top management considers before jumping in headfirst.
In the future, simulations increasingly may run parallel to an organization's actual operations, serving as a comparison between the two. The results of such a comparison would allow the firm to track performance based on the simulation's metrics as well as react quickly to new situations.
Simulations can produce equally exacting results when they're
on process-oriented problems.
—HOSNI ADRA, PRODUCT
MANAGER AND PARTNER,
NAPERVILLE, ILL., USA
“Previously, people have developed a model, learned what they wanted from it and then put it aside,” Adra says. “We say, ‘You have a model you've built. Why don't you use it to forecast how things will run next week?’ So we've created links to databases and Excel that dynamically load data. You can ask what-if questions such as, ‘Given what I have on the floor, if I need to produce this many parts by the end of the week, can I do that?’”
The next logical step might be real-time decision support. “That evolution of a people/information-centered organization would be analogous to the evolution that's already taken place in other industries,” Warren says. “We don't manage chemical plants anymore by having people run around and turn valves. We've got systems that can do those things better than we ever could.” PM
Mark Ingebretsen writes a daily column about health care and biotech for the Online Wall Street Journal.
PM NETWORK | AUGUST 2004 | WWW.PMI.ORG
AUGUST 2004 | PM NETWORK