During the negotiations of the Sustainable Development Goals, the United Nations consulted worldwide nearly ten million people for their views. Such proliferating megaconsultations are often uncritically accepted as a remedy for an assumed democratic deficit of intergovernmental institutions. We argue, however, that the potential of civil society consultations to democratize global governance is constrained by the limited legitimacy of these consultations in the first place. Global consultations regularly fail to include civil society actors from developing countries, or show other sociodemographic biases. Also, they often fail to strengthen accountability between citizens, international organizations and governments. In this article, we investigate the causes of this phenomenon by exploring the relationship between the design of consultations and their democratic legitimacy. The basis for our argument is an in-depth empirical study of three consultations carried out during the negotiations of the Sustainable Development Goals. We find that design is an important variable to explain the overall legitimacy of consultations. Yet its exact role is sometimes unexpected. Extensive material resources and open access conditions do not systematically enhance the legitimacy of the studied consultations. Instead, developing clear objectives, allocating sufficient time to participants, and formally binding the consultation to the negotiations hold considerably more promise.