We introduce a new approach to tackle the mobile manipulator task sequencing problem. We leverage computational geometry, graph theory and combinatorial optimization to yield a principled method to segment the task-space targets into clusters, analytically determine reachable base pose for each cluster, and find task sequences that minimize the number of base movements and robot execution time. By clustering targets first and by doing so from first principles, our solution is more general and computationally efficient when compared to existing methods.