This paper links the quality (i.e., noise and bias) of productivity information and of performance measures to the choice of whether or not production should occur in teams. Agents have private access to a predecision, imprecise productivity signal and agents' performance is measured with both error and bias. With a private, perfect productivity signal, agents earn informational rents, and team production is more costly because teams are more productive and thus earn more rents than agents who work individually. Reducing the amount of the agents' information reduces the rents, making teams less costly, so that teams are preferred when the productivity signal is imprecise enough. More importantly, the threshold of the level of precision in the productivity signal for which teams are preferred depends inversely on the amount of measurement error in the performance measure. Increasing noise in the performance measure results in a decreased chance of reporting the high output, which reduces agents' rents. Therefore an imprecise performance measure can substitute for reducing agents' information and make teams less costly. Further, the impact of a biased performance measurement system depends on the precision of the biased system and the direction of the bias.