Many decision‐makers struggle to reduce commitment to failing endeavors. While de‐escalation mechanisms have been documented and tested in information technology, accounting, psychology, and organizational behavior, little work has addressed de‐escalation in innovation. This study identifies and examines the applicability of de‐escalation mechanisms specifically in new product development (NPD) projects. Initially, we conducted an extensive literature review to identify de‐escalation mechanisms found in research across different academic disciplines. Subsequently, we conducted a qualitative study using semistructured interviews to gather primary data. In total, using a sample of organizations that compete in high‐technology industries, 31 managers and engineers in 15 NPD projects that were terminated prior to completion were interviewed regarding mechanisms used to de‐escalate commitment to failing new product innovation projects. Several mechanisms reportedly used for innovation projects parallel those identified in the previous literature. More importantly, the authors uncover novel and previously undocumented ways to discontinue or redirect poorly performing projects. Specifically, internal competitions, feature‐level de‐escalation, continual assumption assessment, benchmarking, roadmaps, and comprehensive product testing are particularly well suited for innovation projects. These findings extend the body of extant literature on de‐escalation of commitment to include innovation, notably in high‐technology settings. The study also highlights that organizations may want to accept or even encourage failure, thereby increasing the efficacy of resource use. Implications for further research and practice are offered.