People collaborate to address tasks that can often be decomposed into smaller components. These components beg different forms of expertise and may differ in value to the group. We show that transactive memory structure is affected by the interaction between task component value, incentive framing (i.e., whether task incentives take the form of gains or losses), and the distribution of expertise with respect to task components. We find that when groups include a member with expertise in a highly valuable task component, that member is exceptionally motivated and puts forth high levels of effort. Only groups with such an expert successfully realized the potential performance benefit associated with high-value domains. Groups recalled more items under loss framing than under gains framing, but only when the potential losses were low and constant across components. We integrate our findings into the extant literatures on transactive memory and gains/loss framing.