Optimizing your enterprise through statistical analysis

Operational research is the science of making business work better

“Every company is trying to be more competitive with the same money and fewer people. The pressure is on. But what do you do when you run out of obvious things you can do with common sense? Turn to extras. This is high-tech stuff.”

Thus speaks Randy Robinson, a consultant with a Ph.D. in OR (operational research) from Massachusetts Institute of Technology’s Sloane School. But when Robinson talks about using high tech to gain an advantage, he means something beyond IT’s traditional definition of high tech.

Let’s face it: For the most part, IT is still focused on infrastructure and support of the business through the use of computers and communications. Typically, IT leaves it up to the in-house customer -- in other words, management -- to supply the data. Far too many IT organizations are order takers. They say, “We can put together a system for you. What do you want to do?”

Robinson says management may lack knowledge of advanced analytical methods that can make a process better than they ever thought it could be. Operational research fills that void. Rather than just taking the data the customer offers, it discovers what kind of data is helpful. When that’s done, you can build models, run simulations, and perform other calculations on the data to understand where performance gains can be introduced. Robinson calls it the engineering of the information content.

OR is one of those unheralded technologies that makes a real difference in the world. OR professionals gather all the right data, build models, and run predictive analytics and simulations in order to design more efficient operations.

The science started in the late 1930s. The British high command knew that the Germans would attack by air. The question was how best to deploy their limited resources -- radar and fighter pilots -- to give the RAF as much time as possible to scramble and meet the enemy head-on, without knowing the direction from which they would attack.

More recently, Disney used OR to study queues with simulators to cut down on the average length a customer would have to wait in line. Continental Airlines also used OR to reconfigure crew assignments and routes to minimize delays when air traffic resumed following Sept. 11, 2001.

Late last month, Informs, the leading professional OR organization, gave its prestigious Franz Edelman Award for Achievement in Operations Research and the Management Sciences to General Motors. In a nutshell, GM started an OR project in the 1980s that continues to this day. The goal of the project is to improve production at workstations by eliminating production bottlenecks through the use of analytics and simulations.

GM is a huge entity with thousands of workstations and a complex manufacturing system, so it should be no surprise that this project has been in continuous operation for 20 years. So far, GM has saved $2 billion by improving its capability to produce in-demand products on a timely basis without incurring additional overtime costs on the line. Jonathan Owen, staff research engineer at GM, tells me that the ROI was about 5,000 percent -- $50 saved for every dollar spent.

This is the business of OR. My point is that not all productivity improvement comes directly from IT. No one says IT can’t go outside its own discipline to team up with other professionals to improve the way things work.