Using gameplay metrics to articulate player interaction within game systems has received increased interest in game studies. The value of gameplay metrics comes from a desire to empirically validate over a decade of theorization of player experience and knowledge of games as ludic systems. Taking gameplay metrics beyond formalized user testing (i.e. with the aim of improving a product) allows researchers the freedom of examining any commercially available game without the need to have access to the game's source code. This paper offers a new methodology to obtain data on player behavior, achieved through analyzing video and audio streams. Game interface features are being analyzed automatically, which are indicative of player behavior and gameplay events. This paper outlines the development of this methodology and its application to research that seeks to understand the nature of engagement and player motivations.