Recent educational research and agendas highlight the importance of cultivating data literacy among young students as a necessary skill to deal with the challenges of the AI era. This can be leveraged by the development of educational digital designs that would engage students with data literacy practices such as identifying data patterns, critically analysing data, and understanding how AI algorithms classify data and represent information. However, there is still a gap in educational tools designed for enhancing data literacy in K-12 education and in relevant empirical studies. In this paper we discuss the design and evaluation of an online game authoring system, called SorBET, that aims to engage young students with data literacy through the collaborative play and design of classification games. SorBET integrates three diverse computational affordances to allow user’s engagement with data as they play or design digital games. As players they collaborate to classify falling objects into predefined categories using hand gestures and voice commands. As designers, they access and modify the game data with an interactive database and the game rules with block-based programming. We evaluated the SorBET design and its impact on data literacy development with a 2-cycle design-based empirical study with secondary school students, who played and modified two classification games. We collected and analysed a set of data including a system usability scale questionnaire, semi-structured interviews, audio recordings and students’ game versions. The findings suggest that SorBET’s affordances for designing the game data have the potential to support data literacy practices by enabling students to analyze, interpret, and manipulate data in a dynamic and interactive environment. Additionally, the feature of embodied interaction with the game interface enhanced collaboration and urged important discussions around game data between players.