This study describes how Protocol-Assisted Modeling, a technique adapted from industrial performance testing (Schryer 1992), can increase student writers’ audience awareness by helping them build more accurate mental models of how readers think and react to information. Reporting on preliminary results from an on-going study mapping the emergence of situated work-based literacy in French L1 technical communication Masters students, Dressen-Hammouda describes what happens when students are presented with a series of poorly designed documents: after predicting potential problems for readers, they revise their evaluation using a think-aloud protocol transcript showing how actual users reacted. She argues that the approach is effective in helping learners across disciplines and professions form deeper and more intuitive mental representations of the wide range of audiences students will be writing for.