AbstractThis paper presents an application of the maximum principle and the generalized maximum principle (a recent extension of that principle) to a discrete‐time model of consumer choice behaviour. The model includes two brands in a market and uses a non‐stationary first‐order Markov chain. The optimization is carried out by algorithms based on the above mentioned principles. These algorithms are described, and their performances are discussed and compared both with and without fulfilment of the directional convexity pro