OR/MS Today - June 2000 - Was It Something I Said?

June 2000

Was It Something I Said?

A New Way of Looking at Models

By Vijay Mehrotra

In his new book "Serious Play: How the World's Best Companies Simulate to Innovate," Michael Schrage emerges as a strong advocate for the use of software models in business. The book is an ambitious study that makes a powerful case for modeling, drawing on examples from a diverse set of companies.

However, Schrage's perspective on modeling — what it is and why it is valuable — is quite different from that of an operations researcher. Given our training in mathematical programming and probability, most of us OR types think of models as frameworks for optimization and/or for accurately representing the stochastic behavior of a system. We see models as valuable because they produce "better" results, where this goodness can be rigorously quantified in the context of the model.

In contrast, Schrage sees models primarily as communication tools that profoundly change the way people communicate within and across organizations. "The fundamental question," he writes, "isn't 'What kinds of models, prototypes, and simulations should we be building?' but 'What kind of interactions do we want to create?' " To Schrage, abstract models and the communications that are driven by them are an increasingly essential part of the innovation process. By enabling us to see how disparate variables and inputs interact with one another, we are forced to debate, confront and defend what we truly think is important — and quickly. In addition, models often cause our eyes to be opened to possibilities by providing the opportunity to try things we had previously only imagined and by suggesting new alternatives that can be iteratively tested and played with.

This is a whole new way of looking at things, and I believe that it is a very positive paradigm shift for those who make a living from modeling business problems. Schrage's views provide some powerful validation about the importance of what we do, if we can accept that WHY this is important is not exactly what we had thought.

Granted, Schrage's definitions of "model," "simulation" and "prototype" are more liberal than the mathematical representations with which we are comfortable. But his basic concepts really resonated with me, and not just because he cites examples from yield management and traffic theory that are familiar to us operations research professionals.

In 1989, I had the opportunity to work with a group of simulation professionals who were building models for analyzing semiconductor fabrication facilities. As a naive graduate student, I had anticipated that our discussions would center on model verification methods, probability distribution assumptions and variance reduction techniques.

There was almost none of that. We focused first on getting the right people into a room to discuss the manufacturing processes that were to be modeled — where people openly and heatedly debated the accuracy of the data collection systems, the necessity of different operations to the process recipe, and the possibility of different policies for preventive maintenance. Before I was able to shout "Irrelevant!", operational improvements were being made right before my eyes — a result of the modeling process that we were conducting.

We then conducted a standard capacity study, identifying bottlenecks and recommending additional equipment in some areas. This was met with polite applause. However, a few weeks later the director of the facility asked us to run several scenarios in which we introduced new "hot lots" into the system. His goal was to understand the impact of these high priority jobs on the rest of the traffic in the network. Once armed with the simulation results about manufacturing cycle times, he ran off to argue with the executive sponsor about the relative merits of loading the system down with these specials.

Years later, I found myself working as consultant to a technical support call center. There was a strong feeling that customer calls could be prevented if only the product development group would listen to the people in the call center (there was substantial amount of animosity between "us" and "them"). After an initial assessment period, we decided to borrow from our friends in manufacturing and finance: We developed a sampling and inspection process that generated lists of "Frequently Asked Questions" which we married to an Activity-Based Costing method to estimate the support costs associated with each of these specific issues.

Beyond the production of the lists themselves, which were intended to identify customer problems at an "actionable" level, this analysis model was most effective in getting people to the table once a week to discuss the symptoms, the root causes and the possible remedies for these specific problems. The result was a substantial savings in technical support costs, as well as the creation of continually self-correcting software products.

More recently, I worked on research and development of a software product. Our team was attempting to build a sophisticated scheduling solution. From a mathematical perspective, this was a really rich problem. In 1999, this product made it to market. Promoted by a skilled professional sales force, this product was a huge success with the targeted customer population, creating a nice opportunity for my team and I to deliver consulting and training services to a whole new set of companies.

Once we started working with these customers, we discovered something that totally stunned us: More than 60 percent of these organizations were not using the Optimal Scheduling functionality that was the primary mathematical capability of this software. Yet to this day product sales continue to escalate and customer success stories keep pouring in.

Vijay Mehrotra is the CEO of Onward Inc., in Mountain View, Calif. He can be reached via e-mail at vijay@onward-net.com.

Table of Contents

OR/MS Today Home Page

Lionheart Publishing, Inc.
506 Roswell Street, Suite 220, Marietta, GA 30060, USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com