Exploratory search has an unclear and open-ended definition. The complexity of the task and the difficulty of defining this activity are reflected in the limits of existing evaluation methods for exploratory search systems. In order to improve them, we intend to design an evaluation method based on a user-centered model of exploratory search. In this work, we identified and defined the characteristics of exploratory search and used them as an information seeking model evaluation grid. We tested this analytic grid on two information seeking models: Ellis' and Marchionini's models. The results show that Marchonini's model does not match our evaluation method's requirements whereas on the other hand Ellis' model could be adapted to better suit exploratory search.