Stefan's dissertation explores how different interaction modes (parameter-based versus prompt-based toolkits) shape user experience in mass customization. Based on a controlled experiment with 178 participants, the study shows that emotional engagement (“valence”) follows distinct patterns depending on the interface, with important implications for product valuation and user behavior.

While traditional parameter-based systems replicate known value drivers such as preference fit, uniqueness, and self-design effects, prompt-based systems powered by generative AI fundamentally shift these dynamics. The findings highlight that interaction design is not neutral: it actively shapes how users experience, evaluate, and complete self-design processes.

These insights are particularly relevant as AI-driven interfaces become more prevalent, offering new opportunities to reduce abandonment rates and optimize user experience in customization environments.