Abstract:

Role-plays are a staple of business education because they provide a structured way for students to practice persuasion, relationship management, and professional communication under conditions of uncertainty and interpersonal pressure. Yet instructors routinely face implementation barriers with live peer role-plays, including absenteeism, uneven preparation, inconsistent role fidelity, and limited time for repeated practice. This Best Innovative Practice article describes a Custom-GPT grade-appeal role-play designed to preserve the experiential value of a difficult influence conversation while reducing logistical constraints and expanding opportunities for iterative skill development.

The activity is implemented in an undergraduate Organizational Behavior course using a three-step instructional sequence. First, students receive a brief in-class module introducing evidence-based persuasion strategies (e.g., clarifying objectives, supporting claims with specific evidence, framing, active listening, responding to objections, and maintaining a relationship-preserving tone). Second, students complete a take-home role-play in which they interact with a policy-driven “Virtual Professor” to request reconsideration of a course grade. The simulation is anchored in a standardized performance context to ensure consistent challenge and comparability across students. During the interaction, the Virtual Professor asks probing questions and responds differently to effective versus ineffective influence tactics, discouraging entitlement, pressure, or bargaining and emphasizing fairness and policy constraints. After the professor provides a closing decision statement with rationale and next steps, the Custom-GPT switches into an “Observer” mode that delivers rubric-based feedback and coaching, including scores on key persuasion dimensions and targeted suggestions for improvement. The assignment is graded complete/incomplete to encourage experimentation and reduce performance anxiety, and the number of attempts is recorded as an indicator of iterative engagement. Third, an instructor-led debrief consolidates learning by connecting students’ experience to course concepts and emphasizing transfer to workplace influence conversations.

Evaluation evidence includes student survey responses and qualitative feedback. Students completed a brief post-activity survey assessing perceived realism and engagement, perceived skill practice in communication and persuasion, and perceived usefulness of immediate feedback for reflection and improvement, along with an open-ended question about preferences for GPT-based versus live peer role-plays. Survey results (N = 43) indicated strong perceived value across all three Likert items, with mean ratings between 4.19 and 4.47 on a 5-point scale and high agreement rates (88.4%–93.0% selecting Agree or Strongly agree). Students completed an average of 1.4 attempts, suggesting some iterative practice. Open-ended responses highlighted tradeoffs between GPT role-plays (convenience, reliability, repeatability, immediate feedback) and live peer role-plays (human unpredictability, richer emotional cues, interpersonal accountability), suggesting that a blended approach may be most effective.

Keywords:

Business education, persuasive communication, role-play simulation, generative artificial intelligence.