Abstract
Many workers are evaluated on their ability to engage with customers. We measure the impact of gender-based customer discrimination on the productivity of online sales agents in sub-Saharan Africa. Using a novel framework that randomly varies the gender of names presented to customers without changing worker behavior, we find the assignment of a female-sounding name leads to 50 percent fewer purchases. Customers also lag in responding, are less expressive, and avoid discussing purchases. We show similar results for customers around the world and across workers. Removing customer bias, we find women would be more productive than their male coworkers.