Performance testing asks candidates to prove skills by doing, but developing these assessments demands strategies that differ sharply from multiple-choice testing. This panel will deep dive into the critical development steps unique to performance formats, security approaches purpose-built for them (random task selection, sequential task playlists, and automated item generation), and the strategic use of conversational items when full tasks take too long. Attendees leave with concrete practices drawn from live VR and hands-on lab programs.