Selecting R&D projects at BMW: a case study of adopting mathematical programming models

R&D project selection is a critical interface between the product development strategy of an organization and the process of managing projects day-to-day. This article describes the project selection problem faced by the transmission pre-development group of BMW. The group had to choose a portfolio of projects to set the foundation for the “best powertrain 2000.” This problem of project selection was structured as minimizing the gap between target performance and actual performance along chosen criteria. A mathematical programming model helped this organization to increase the transparency of their selection process, which previously had been based on experience coupled with evaluation of individual projects in isolation. The model was used to structure data collection, to apply to consistent criteria to the selection of pre-development projects, and to compare weighted project benefits. Implementation was a success in that the pre-development group continues to use it to make better decisions. However, the organization did not use the model for constrained optimization. We see two reasons for this partial implementation of the model. First, .an analytical model cannot capture all considerations relevant to optimization (e.g., uncertainty or strategic fit). Thus, constrained optimization promised only marginal further benefits while requiring substantial additional effort. Second, the diffusion of the analytical model from academia to industry faced the same problems as any technology transfer: Significant tacit knowledge must be transferred along with the codified knowledge of the analytical model. This required iterated problem solving, which itself required the limited time and resources of the diffusing agents (academia) as well as the adopting agents (industry). Thus, the organization adopted only those elements of the modeling method that could be transferred within the resource constraints, focusing on those elements offering the highest benefit per effort invested.