It is still an open question when groups perform better than individuals in intellective tasks. We report that in an Acquiring a Company game, what prevailed when there was disagreement among group members was the median proposal and not the best proposal. This aggregation rule explains why groups underperformed with respect to a "truth wins" benchmark and why they performed better than individuals deciding in isolation in a simple version of the task but worse in the more difficult version. Implications are drawn on when to employ groups rather than individuals in decision making.