Organizations succeed when they are capable of solving complex, non-routine problems. Often, these tasks are done by teams of individuals, usually after the individuals alone have had a chance to think through the issues and possibilities. The interplay of incentives and performance on complex choices is not well understood, neither theoretically nor empirically. In particular, an objective measure of performance is often not available, and thus less-studied incentives relying on subjective evaluation are needed. We study incentives for individual and group performance in a novel complex and non-routine task: guesstimations. In this task, first subjects work individually, then decide on the final answer in a group of three. First we study the group decision relative to individual inputs: Compared to the median quality individual guess, the group significantly improved, but compared to the best individual guess, the group performs significantly worse. Groups do outperform ``mechanical'' ways of aggregating individual answers and are especially valuable when individual answers do not ``straddle'' truth. Individual characteristics are not predictive of group outcomes in our setting. Second we study the effect of incentives on performance: while treatments affected group atmosphere, individual as well as group performance was not significantly affected.