A near-death experience

Many of us have heard stories about near-death experiences. A person involved in a serious accident or a person gravely ill recovers and tells their account of leaving this Earth and walking toward the bright light, outstretched arms awaiting them. Miraculously, they come back to life – functioning, contributing and taking advantage of a second chance – and eager to tell their story.

A near-death experience is a good description of what happened to operations research (O.R.) – an advanced form of analytics sometimes referred to as “prescriptive analytics” – at UPS. In the mid-1990s, the O.R. group at UPS was walking toward the bright light, voices calling, ready to leave this world as we know it. At the time, UPS was looking at areas to reduce costs, and O.R. was, quite frankly, a cost bucket with no apparent return on investment despite being surrounded by plenty of problems ripe for O.R. approaches and algorithms.

Ultimately, O.R.’s potential to positively address the complex operational problems UPS faced outweighed the cost concerns, and operations research was given a second chance at life, but clearly some changes were needed in order to properly tap its potential.

Change No. 1: Organization. Many companies over the years have debated where O.R. should reside in an organization. Should O.R. report to IT? Should O.R. report to the business side? If on the business side, where?

Before changes were implemented, operations research at UPS reported to two groups: UPS Airline and UPS Transportation. The O.R. team associated with UPS Airlines was located in Louisville, Ky., worked almost exclusively on airline projects and reported to IT. The second O.R. group, located in Timonium, Md., worked on opportunities associated with the tractor-trailer and rail movements between UPS facilities. The two O.R. teams – separated geographically and by the nature of their projects – had little interaction.

In the first major change, the two O.R. groups were placed together in the Industrial Engineering department (IE). IE is responsible for optimizing UPS operations, from the vision to the execution of the plan. IE develops “roadmaps” for processes and technology and prioritizes projects to deliver against the roadmaps. As these roadmaps are developed, optimization components are included in the vision.

The move to IE proved vital to O.R.’s eventual success at UPS in part because O.R. is a component in many industrial engineering degree programs at universities. Although most IE’s can’t write the sophisticated algorithms associated with O.R., they understand the power and limitations of O.R.

Change No. 2: The approach. Along with consolidating the O.R. expertise and moving it to the IE department, UPS also changed its approach to O.R. Previously, the company’s approach went something like this: “O.R. is all about optimization, right? All the O.R. groups need to do is push a button and out pops the optimized answer.” This flawed approach was one of the major reasons that O.R. spent years with no tangible deliverables, resulting in frustration by both the O.R. groups and the business owners demanding the answer. Lessons learned: The managing group needs to understand the power of operations research. Just as important, it needs to understand the limitations of O.R.

The new approach brought to the surface three key issues: 1. data, 2. assistance vs. optimization, and 3. roadmaps.

A longstanding dilemma with O.R. can be summed up as follows: When a model produces a bad answer, is the problem with the data or the algorithm? As this question was being addressed at UPS, the assistance vs. the optimization opportunity became clearer. Once the data requirements were identified and obtained, in many cases the organized data (presented in a simple tool such as Excel or Access) became a valuable aid to the employee making planning and execution decisions. Although the human could not determine the optimum with the assistance, the organized data allowed him or her to make better decisions. While the algorithms were being developed and improved in research, the employee was making better decisions, bringing immediate benefit to UPS. And once the new algorithm was available, the optimization process benefited from the validated data.

The final lesson learned through the change in approach concerned the “roadmaps.” Many of the previous O.R. applications were standalone applications, creating a lack of continuous “clean” data, as well as the need for specialized runs (often conducted by the O.R. person). By having a vision for the processes and technology (the roadmap), the optimizations became “back box” O.R. applications that were part of the overall process. The systems and processes were integrated to work with or without the algorithms. The O.R. application simply made the solution optimum. This approach allows UPS to make improvements while the O.R. applications are developed.

These improvements, usually in the form of cost reductions, help fund additional research while bringing immediate benefit to the bottom line. Because the properly integrated systems and processes are not dependant on the O.R. “black box,” UPS does not have O.R. roadmaps; instead, UPS has roadmaps with O.R.

Change No. 3: Accountability. A near-death experience means you survived, but it does not guarantee how long you will live. Everyone involved with the O.R. groups within UPS knew this second chance was in all likelihood the last chance.

Not every research attempt will deliver. Research is not the same as a project. A project has tools that allow the measurement of the triple constraints of functionality, schedule and costs. Earned value and other measures can continually keep the project manager and senior level managers informed on the progress of the project. Corrective action can be taken to make sure a project stays on track.

Research progress is much harder to gauge. Seemingly large advances are often followed by long periods spent discovering what doesn’t work. At UPS, in many cases, if the research hits a plateau for an extended period of time, management determines a decision date. If significant progress is not made by the decision date, the research is considered a sunk cost, the research is halted and the O.R. group moves on to the next “black box” identified in the roadmap. Research progress is reviewed by mid-management at least every month and by senior management at least twice a year. Just because it’s research doesn’t mean it can’t be held accountable.

Today, UPS truly has a world-class operations research group. UPS received the INFORMS Prize in 2003 and placed second in the Franz Edelman Award competition that same year for its work on “Planning the UPS Air Network.” The INFORMS Prize recognizes organizations that “effectively and repeatedly integrate the principles of operations research and management science in pioneering, varied, novel and lasting ways.” The Edelman, considered the “Super Bowl” of O.R., is an annual competition that honors the best example of OR/MS and advanced analytics practice in the world. Both the INFORMS Prize and the Edelman Award are presented by the Institute for Operations Research and the Management Sciences (www.informs.org).

Along with its active engagement with INFORMS, including contributions to INFORMS journals, UPS O.R. collaborates (or has collaborated) with Georgia Tech, University of Iowa, University of Michigan, MIT, Northwestern, Princeton and University of Southern California.

The O.R. group has expanded its reach to many components of UPS operations, including pickup and delivery, air and ground transportation and facility utilization. Operations research has been recognized by UPS’ executive management for its contributions to the company, and C-level executives continue to challenge the O.R. group to deliver increased benefits in the area of service and cost.

O.R. faced a near-death experience at UPS and took full advantage of its second chance at life. Today, 15 years later, it’s hard to image UPS without O.R.