With the software installed, Knabe and a project team set about validating the solution by creating a baseline. The project team was led by a representative from the company's distribution group, with Knabe acting as a supporting analyst. This team structure ensured that those functional specialists who intimately know the business came onboard with the solution early on and had the opportunity to see that the software was capable of producing accurate results.
Using data representing Gillette's current supply network, the project team ran the application to estimate what the company's as-is supply chain costs should be, based on those data. They then compared the software's results with the actuals coming out of Gillette's financial system. "You're not going to get it down to the penny," Knabe says. "But the financials and the software should come out pretty close, within a few percentage points. That gives the software some credibility." Where the variances were significant, the project team would drill down into the results to determine what needed to be fixed in the data they were loading into the system, then re-run the results and repeat the process until the software accurately reflected Gillette's business. "That process of working toward intuitive results with the project team is a major step in building confidence in the results," Knabe explains.
Sometimes, however, even when the team was certain that the data were correct, the system still came out with counterintuitive results. "That's where it gets really interesting," Knabe says, "because you start to gain an understanding of the drivers of your own network and what's causing the counterintuitive results." For example, corporate intuition might suggest that Massachusetts would be a good site for a warehouse because shipping freight out of the Northeast is relatively inexpensive since few manufacturers have production facilities in the region. But the system might suggest that high labor costs make the Northeast a less attractive site for a facility after all. By gaining an understanding of these types of insights, the project team gains both new corporate intuition and increased confidence in the solution's ability to produce accurate results.
That comfort level with the system became vitally important, Knabe says, when the project team started actually applying the solution to run what-if scenarios, planning the future location of facilities and affecting, potentially, millions of dollars in future costs. "You have to be open to learning from those counterintuitive results," Knabe explains. "If you know that you want a warehouse in New Jersey — and if you're going to say that every solution that suggests you don't have a warehouse in New Jersey is wrong — then you don't need to go through this process. But if you're open to the idea that maybe your warehouse should be in Cleveland — which is counterintuitive — you might find that you can actually get to the East Coast pretty fast from Eastern Ohio, and it's a lot less expensive. Then, because the project team has representatives from the Distribution group, once they are confident in the system's results, they can communicate the results and the recommendation effectively to upper management."
Gillette's first project using the software delivered initial results in March 2003 and final results in May of that year. Since then, Gillette has continued to use the software to support analyses around the globe, completing a total of eight projects that, Knabe says, have helped the company improve its customer service, reduce distribution costs or both. Although she was unable to share precise return on investment figures, she did note that the results of one North American project showed decisively that Gillette was able to achieve much higher on-time rates when it shipped using truckload versus less-than-truckload (LTL), indicating that the company should continue to build its supply network in a way that increases truckload frequency. (At the time of the interview for this article, Knabe was working on an analysis to understand the impact that the modeling software has had on a key metric for Gillette: percent of deliveries on time.)