We describe the utility of process models for summarizing the sequential actions of individuals. Such models describe why users did what they did, what information they used from the outside environment, and what knowledge they used to perform the task. These detailed explanations of users' thoughts and actions can enhance interface design by offering behavior summaries that are inspectable and transferable to new interfaces. Sequential data sets and models for human-computer interaction are often large and complex. We present a computer-supported methodology for developing these models as summaries of sequential data. We illustrate that this methodology can make building and using such models tractable by applying it to an existing model for using an on-line database.