In this conceptual paper, we propose that insights from conversation analysis (CA) may provide a useful approach for scholars interested in online learning by focusing first on how learners themselves orient to performing social actions online. We further propose that gaining a better understanding of what conversational moves are actually doing in online discussion can help researchers and educators better assess the learning that takes place there. Through worked examples, we demonstrate the ways that three conversational features (agreements, personal experiences and stories, and cognition verbs) tend to function in online institutional talk. We illustrate how these conversational features can accomplish a variety of actions in online academic discussions, including social affiliation, displaying expertise, and distancing oneself from claims. Recognizing how these features are interactionally deployed can help scholars better understand the functions of learners’ talk. CA is a practical methodology for educators who may be reluctant to impose interactional frameworks on student discussions for learning.