Understanding the relationship between organization design and innovation is a topic of significant interest to both scholars and practitioners. Two central strategic management theories imply a fundamental trade-off exists at the heart of this relationship. On the one hand, taking a knowledge-based view, more centralized designs are expected to be associated with enhanced intra-organizational knowledge flows, which can enhance innovation. On the other hand, taking an incentives-based perspective, more decentralized designs are associated with higher observability of effort and likely to facilitate the more effective use of incentives, which can increase innovation efforts. I offer a framework to understand the implications of this trade-off by viewing innovation as a process. I divide this process into upstream tasks around invention, and downstream tasks around product development and commercialization. I empirically examine this trade-off in the context of the pharmaceutical industry. I find that greater centralization is associated with the creation of more original inventions and more of these inventions progressing through the earlier stages of development. In contrast, greater decentralization is associated with the creation of more inventions and with more inventions progressing through the later stages of development. These results illustrate that knowledge flows play a more important role further upstream and incentives play a greater relative role further downstream. This study provides a theoretical framework for understanding both where (in the organization) and when (in the innovation process) design choices can facilitate or hinder innovation outcomes, thereby helping to reconcile some of the disparate findings in the extant literature.