We study the market partition between two distinct firms that deliver services to waiting time sensitive customers. In our model, the incoming customers select a firm on the basis of its posted price, the expected waiting time and its brand. More specifically, we quantify by a cost any departure from the ideal brand expected by each incoming customer. Considering that the two underlying queueing processes operate under high traffic regimes, we analyze the market sharing dynamics by using a diffusion process. As a function of control parameters, such as the waiting and brand departure costs or the incoming traffic intensity, we are able to analytically characterize a transition between an Hotelling-like regime (dominated by brand considerations) and a deadline type regime (dominated by waiting time considerations). The market sharing dynamics is described by the time evolution of a boundary point, which time evolution belongs to the class of noise-induced phase transitions, so far widely discussed in physics, chemistry and biology. Explicit illustrations for both symmetric (i.e. identical servers) and asymmetric cases are worked out.