Clustering of membrane-associated molecules has long been proposed to promote size-dependent interactions with the actomyosin cortical network, thereby allowing them to be transported by actin flow. Consistent with such a model, clustering of the conserved polarity protein PAR-3 in the C. elegans zygote is essential for its polarization by cortical actomyosin flows, and clusters of PAR-3 visibly move with flows. However, here we show that advection by cortical flow is independent of clustering. Clustered and non-clustered PAR-3 variants are advected equally well over short timescales by cortical actin flow as are other PAR proteins. Moreover, we see no strong links between advection and either cluster size or diffusivity that would be consistent with size-dependent interactions with the actin cortex. Instead, using a combination of experiment and theory, we find that efficient long range transport of PAR proteins by cortical flow is primarily tuned by the stability of membrane association, which is enhanced by clustering and determines the persistence of both transport by flow and the resulting asymmetries once flows cease. Consistent with this model, stabilizing membrane association was sufficient to induce segregation of a non-segregating PAR protein, effectively inverting its polarity. We conclude that the impact of advection on membrane-associated proteins is much broader than previously anticipated and thus cells must appropriately tune membrane association dynamics to achieve selectivity in the long range transport of membrane-associated molecules by cortical flows.