It has been shown that indecisiveness is involved in many unwanted cognitive states, such as procrastination, distractibility, the lack of self-esteem, or even revenge. The purpose of this work is to propose a predictive model for the recognition of the indecisiveness class, from the analysis of the customer’s trajectory and his gripping. The movements are captured thanks to infra-red sensors. A structural behavioral architecture is built, based on eye-tracking methodology. Indeed, all the movements of a customer in a selling area can be assimilated to fixations and saccades. We show that the path in the selling area can be seen as a sequence of states. The final predictive classifier is built with a combination of Hidden Markov Models (HMM) through a logistic regression model (LRM) and leads to satisfying results, as it correctly predicts 88 % of the subjects’ indecisiveness classes.