Once the Stars Aligned: Serendipity and Innovation in New Entrepreneurial Ventures

Serendipity is a powerful driver of innovation, yet it is often dismissed as blind luck or attributed to an innate disposition. Consider serendipitous discoveries such as Fleming’s discovery of penicillin or James Dyson’s vacuum cleaner; while the outcome is clear, it remains unclear how such experiences can be systematically understood, transferred, or replicated. This doctoral thesis investigates the behavioral, relational, and search mechanisms and dynamics through which the process of serendipity can be systematically understood, namely: the learning process behind serendipity, curious search strategies in serendipity, and ego-network dynamics for unexpected matching pairs. The first study explores how individuals learn from unexpected events, how serendipity develops over time, and how unique learning trajectories shape its formation. This qualitative inquiry conceptualizes serendipity as a "garbage-can" learning process based on 62 interviews with scientists, entrepreneurs, and innovators. The second study examines how entrepreneurs’ networking processes give rise to serendipitous outcomes, and how such outcomes, in turn, reconfigure their ego-networks. Drawing from a longitudinal qualitative study of 59 entrepreneurs participating in a technology-focused accelerator, this study advances a bidirectional account of network dynamics in entrepreneurial journeys. The third study analyzes how curious search enables learning from surprising events and sustains experimentation in a non-stationary environment. This computational model formalizes curious search using a hierarchical multi-armed bandit model.