Anecdotes abound suggesting that the use of predictive analytics boosts firm performance. However, large-scale representative data on this phenomenon have been lacking. Working with the U.S. Census Bureau, we surveyed over 30,000 manufacturing establishments on their use of predictive analytics and detailed workplace characteristics. We find that productivity is significantly higher among plants that use tools to automate prediction—up to $918,000 higher sales compared to non-adopting competitors using similar inputs. Furthermore, both instrumental variables estimates and timing of gains suggest a causal relationship. However, this productivity payoff only occurs when predictive analytics is combined with key workplace complements. Significant accumulation of IT capital, educated workers, or workplaces designed for flow-efficient production each enables non-trivial returns. Notably, managerial capacity – measured either as management practices or as manager headcount – also enables both predictive analytics adoption and performance gains. Our findings support claims that these fast-diffusing techniques can substantially improve productivity, while also explaining why some firms see no benefits at all. Further, they provide the first evidence of labor-augmenting automation in manufacturing, in contrast to the labor-substitution associated with robots.