A virtual keyboard is a software component that replicates the functionality of a physical keyboard, allowing users to input characters without physical keys. Virtual keyboards are commonly used in scenarios where a physical keyboard is unavailable or impractical due to constraints such as security considerations. From a security perspective, virtual keyboards help mitigate keylogger data sniffing and shoulder-surfing attacks. Consequently, many web applications, especially those related to banking and finance, use virtual keyboards for password entry. In this work, we propose a graph-based usability analysis of virtual keyboards. We convert the input key sequence into a graph and then apply relevant graph-based metrics to identify usability issues. To assess the effectiveness of this approach, we conducted a lab study with 30 participants. Preliminary results indicate that a user’s discomfort and confusion while creating a password directly affect the properties of the graph.