Research and development (R&D) projects face significant organizational challenges, especially when the different units who run these projects compete among each other for resources. In such cases, information sharing among the different units is critical, but it cannot be taken for granted. Instead, individual units need to be incentivized to not only exert effort in evaluating their projects, but also to truthfully reveal their findings. The former requires an emphasis on individual performance, whereas the latter relies on the existence of a common goal across the organization. Motivated by this commonly observed tension, we address the following question: How should a firm balance individual and shared incentives, so that vital information is both acquired, and equally importantly, disseminated to the entire organization? Our model captures two key characteristics of R&D experimentation: information is imperfect and it is also costly. Our analysis yields several important implications for the design of such incentive schemes and the management of R&D portfolios.