Internet systems are a typical scenario where sequences of interactions arise. Modeling the factors that drive the dynamics of an online auction, for example, is complex, since successive interactions become a loop-feedback mechanism, that we call reactivity, that is, the user behavior affects the auction negotiation and vice-versa. In this paper we briefly describes our methodology for characterizing online auctions, considering reactivity. We present the reactive transitions, that is the approach we adopt to model reactivity in online auctions. The reactive transition models the reactivity function, providing a way to discover the bidding behavior's patterns. We also validate our model using actual bidding data from eBay. The results show rich details to understand bidding behavior, that can be used to design support-decision agents and simulate e-markets.