All he's saying here is that you should test/push boundaries, not to continue once they've rejected the advances. But, he notes, that even if they reject, it achieves a desired effect which may eventually pay off.