The goal of many current process improvement efforts is to become more agile by adopting an agile process. However, the results of several recent projects suggest that when attempting to become more agile, adopting an agile process is exactly the wrong thing to do! In this experience report, I discuss my failures with wholesale process adoption and my successes using an incremental adoption strategy based on metric- and retrospection-driven feedback. Similar to refactoring practices for design and code, this strategy identifies "process smells," and targets the worst of them with specific agile practices drawn from several popular agile processes.