Ancient human migrations provide the critical genetic background to historical and contemporary human demographic patterns. However, our ability to infer past human migration events, especially those that occurred over rapid timescales, is often limited. A key example is the peopling of Polynesia, where the timing is relatively well defined, but the exact routes taken during the final stages and the source populations are not. Here, we discuss the technical limitations of current methods for inferring rapid human migration events, using the final stages of Polynesian migration as an example. We also introduce a promising new proxy method to infer human migrations—patterns of bacterial evolution within ancient dental calculus (calcified plaque). While we focus on Polynesia, this method should be applicable to other past migrations, enhancing our understanding of human prehistory and revealing the crucial events that shaped it.