How Golgi glycosyltransferases and glycosidases (hereafter glycosyltransferases) localize to the Golgi is still unclear. Here, we first investigated the post-Golgi trafficking of glycosyltransferases. We found that glycosyltransferases can escape the Golgi to the plasma membrane, where they are subsequently endocytosed to the endolysosome. Post-Golgi glycosyltransferases are probably degraded by the ectodomain shedding. We discovered that most glycosyltransferases are not retrieved from post-Golgi sites, indicating that retention but not retrieval should be the primary mechanism for their Golgi localization. We proposed to use the Golgi residence time to quantitatively and systematically study Golgi retention of glycosyltransferases. Various swapping chimeras between ST6GAL1 and either transferrin receptor or tumor necrosis factor α quantitatively revealed the contributions of three regions of ST6GAL1, namely the N-terminal cytosolic tail, transmembrane domain, and ectodomain, to Golgi retention. We found that each of the three regions is sufficient to produce retention in an additive manner. The N-terminal cytosolic tail length negatively affects the Golgi retention of ST6GAL1, similar to the effect of the transmembrane domain. Therefore, the long N-terminal cytosolic tail and transmembrane domain can be a Golgi export signal for transmembrane secretory cargos.