Gender gaps in earnings persist even among high-skilled workers, partly because men and women often perform different tasks within and across jobs. We study a rare setting in which high-skilled men and women perform the same tasks under comparable conditions, allowing us to assess gender differences in productivity and pay without confounding from task or client allocation. Using administrative data from the Swedish Public Employment Service, we exploit a rotation scheme that quasi-randomly assigns job seekers to employment caseworkers. We find that productivity differences are small: job seekers assigned to female and male caseworkers exit unemployment at similar rates, and hourly wages—conditional on productivity—are nearly identical across genders. Despite this, female caseworkers earn about 8 percent less per year, entirely due to differences in contracted and actual hours worked. We also find suggestive evidence that male caseworkers are more likely to be promoted than equally productive female colleagues. When tasks are standardized and performance is measured objectively, gender differences in productivity and hourly pay are minimal, while gaps in annual earnings and career progression persist.