For any given Game Based Learning (GBL) project to be successful, the player must learn something. Designers may base their work on pedagogical research, but actual game design is still largely driven by intuition. People are famously poor at unsupported methodical thinking and relying so much on instinct is an obvious weak point in GBL design practice. Cognitive Walkthrough (CW) is a user-interface design technique that helps designers model how a type of user will understand an interface. The authors suggest that CW should be extended for use in any context where a designer must model a user's thinking. They present an extension of CW that is suitable for constructivist GBL and apply it to a previously evaluated game to understand why one section of the game was more successful than another. The CW extension explains hitherto puzzling results and suggests further development of CWs for designer support may be beneficial.