Award-winning analytics

When we think of analytics, we think of using data to make better decisions. But as any good analytics practitioner will tell you, that’s only part of the story. When analytics leads to fundamental changes in the way an organization approaches decision-making, the story moves from the back shelves to the bestseller list.

I recently attended a collection of inspiring presentations at the INFORMS Conference on Business Analytics and Operations Research, including all of the finalists in the Franz Edelman competition.

Edelman presentations are unique in a number of ways. They’re highly scrutinized. Teams of presenters must make the case that their work saved or made a lot of money for their organizations – typically tens or hundreds of millions of dollars. The Edelman evaluation committee begins by validating that the organization stands behind these figures, and then only a handful of presentations are chosen as finalists for the competition. The finalists are each assigned a coach who leads the team through the process of creating a carefully refined paper and presentation. High-ranking organizational officers routinely participate in the presentations, either in person or via video if they cannot attend the conference. The luminaries who took time to record personal messages for this year’s Edelman presentations included Paul Otellini and Andy Bryant, the CEO and vice chairman of the Board at Intel, respectively, and Meg Whitman, CEO of Hewlett-Packard.

The real gem, of course, is the content behind the presentations. One organization is chosen as the first-place recipient, but the Edelman committee emphasizes that all of the finalists are winners. And that’s as it should be. The finalists have already been through a demanding vetting process, and the application of good analytics deserves recognition as a win.

I learned something from all of the presentations I attended, but I was particularly taken by two of the Edelman presentations. In addition to impressive cost savings, both described activities that led to fundamental organizational change but in quite different ways. And both offered lessons for anyone seeking to build an organization where analytics serves as a cornerstone.

In the early 2000s, Intel faced a problem. A central reality of the semiconductor industry is ever-smaller transistors. Roughly every two years, companies such as Intel invest billions of dollars on new capital equipment to manufacture the latest generation of microchips.

For Intel, the problem was one of lead-time. Intel’s lead-time for ordering equipment from suppliers was roughly four to six quarters. However, forecasts that far in advance of demand for Intel’s finished products (which in turn dictated Intel’s equipment orders) were prone to error. If Intel could only improve its forecasts, the savings could run into the billions of dollars.

And here’s where the real beauty of Intel’s analytics efforts shined forth: The company didn’t choose to rely on finding better forecasts, recognizing that only so much could be done to improve them. Instead, Intel realized that while its four to six quarter forecasts were poor, its one to two quarter forecasts were reasonably good. Thus, if the company could reduce lead times on orders with its suppliers, the cost savings would be phenomenal.

Intel was able to better manage lead times by promoting a two-mode ordering arrangement. “Base” mode orders were placed with the usual lead times, while “flex” mode orders could be made with a much shorter turnaround. Sophisticated analytics models were used to develop the approach and to help make decisions on when to exercise the flex options (if at all). However, as the team pointed out when it undertook its first experimental evaluation of the approach with one supplier, “Our biggest challenge was convincing Intel management and our equipment supplier to depart from standard practice.”

The concept of a two-mode ordering structure isn’t in itself earth shattering, but it upended existing practices. Intel and its suppliers had to meet and negotiate until acceptable terms were reached for all parties involved – a process that has taken more than a decade and is still ongoing. Along the way, the many individuals involved in contracts and ordering have gained a much greater appreciation of the power of analytics. They’ve changed how they look at decision-making.

TNT Express also achieved organizational change as a result of its analytics initiatives but in a more direct fashion. TNT Express is a package delivery service headquartered in the Netherlands that operates air and road networks in Europe, China, South America, the Asia-Pacific region and the Middle East. At about 80,000 employees, it’s nearly identical in size to Intel.

In 2005, the company undertook its first analytics/operations research effort, taking aim at optimizing the use of the domestic road network in Italy. TNT Express was very late to the analytics game by the standards of companies such as Federal Express and UPS, where analytics has been a foundational tool for decades. What makes TNT Express’s story so compelling was how it recognized and responded to the lack of analytics. Based on its success in Italy, TNT Express took the major step of instituting the Global Optimization (GO) project.

Those involved with GO realized from the outset that analytics couldn’t be relegated to the sidelines, but that it “…should be at the core of our business…” To achieve this, the team undertook two expansive, people-focused initiatives known as the GO Communities of Practice and the GO-Academy. The GO Communities of Practice is composed of individuals with similar responsibilities in different parts of the world who meet three times a year. The Communities of Practice serve to facilitate knowledge sharing as analytics is rolled out globally.

The GO-Academy was developed in response to “…user resistance, which in some cases required a considerable effort to overcome.” The stated objective of the academy was “…to train employees in optimization principles and, at a high level, to acquaint them with the available optimization tools, without trying to turn them into mathematicians.” The academy’s curriculum consists of coursework and projects completed over a two-year period, and graduates are expected to promote and explain analytics throughout the organization. Even the company’s board members have participated in some level of training, and CEO Marie-Christine Lombard is an active, personal advocate of all aspects of the GO program. The effort has brought profound changes to the way TNT Express makes decisions, all precipitated by analytics.

Carefully reviewed papers prepared by all of the Edelman finalists are scheduled to appear in a forthcoming issue of the INFORMS journal Interfaces. I highly recommend taking a look at these papers, as well as the many other analytics success stories captured within the journal’s pages.