How should social media platforms respond to harmful speech? For some content, platforms respond by removing it. But for other content, platforms allow the speech to be posted, but reduce its distribution throughout their networks to reduce its visibility. This second technique—demotion—is swiftly becoming one of the most significant ways in which our communications are governed. Yet it strikingly remains normatively undertheorized. This article offers a framework for when platforms should demote content instead of removing it. After explaining why demotion policies by platforms stand in need of normative justification, and reviewing the various human rights norms that constrain how demotion should operate, it argues that demotion should be pursued instead of removal in cases where removal would violate principles of necessity or proportionality, yet demotion would satisfy them. By explaining when and why this is so, the article seeks to illuminate how notions of necessity and proportionality—central to both political philosophy and human rights law—should be applied to the distinctive context of online content moderation.