Serendipity — the process of unexpected discovery and innovation — represents a major opportunity for observational crowdsourcing. We show how observational crowdsourcing platforms may be inadvertently designed to disrupt the serendipity process. We then examine how design decisions affecting crowdsourcing projects and platforms may promote (instead of prevent) serendipity. We introduce the concept of data item pluripotency: the capacity for a data item to hold unexpected uses for a data consumer. We then develop a novel data design framework showing how project and platform design decisions influence the quality dimensions of a project's conceptual model, data items, dataset, and data use. The data design framework presents a powerful way to understand the relationship between design decisions, the different components of a crowdsourcing project, and the lower- and higher-order data quality dimensions we aim to cultivate. We conclude by discussing the implications for observational crowdsourcing and beyond.