We study how a profit-maximizing platform designs and prices certificates for workers who subsequently enter a competitive labor market. The platform designs a test, which generates a (possibly random) score based on each worker’s productivity, and sets a uniform fee that it charges each worker who wishes to disclose her score to firms. After observing the worker’s disclosure (if any), the labor market pays a wage that equals the worker’s expected productivity given all available information. We show that if the platform can select equilibria, it can extract all of the surplus by using two scores. In contrast, if the workers choose their most preferred equilibrium, then even with binary types the platform optimally uses a continuum of scores and higher scores appear exponentially more than the lower scores.