The availability of social networks data seeks researchers' interest to develop algorithms and machine learning models to analyze user interactions and behaviors. These algorithms rely on network topology to represent structural changes and to detect remarkable precursors generally preceding major events. The approach presented in this article aims to study whether certain graphlets (specific patterns) can be considered as precursors of an event. We experiment the proposed method on three different sets of social networks data. We also study the role (position) of influential nodes in the graphlets, which have a central position in the global graph. After analyzing the results, we show that graphlets are considerable precursors of events.