Designing for Persistence: How Task Sequencing Affects Task Completion in Crowdsourcing Interfaces

Crowdsourcing systems often assume that task order is a neutral design choice. However, the sequence in which tasks are presented may substantially influence user behavior and persistence. This study examines the effect of difficulty-based task sequencing on user dropout and investigates whether adaptive sequencing can improve user persistence. We conducted controlled experiments on Amazon Mechanical Turk across two task domains, comparing fixed ordering strategies with a dynamic adaptive approach that adjusts task order based on user behavior. The results show that no single fixed ordering strategy is consistently optimal across users. In contrast, adaptive sequencing consistently produced lower abandonment rates than non-adaptive ordering strategies across both task domains, with a significant overall reduction observed in the combined analysis. These findings highlight task sequencing as a critical yet underexplored design dimension in crowdsourcing systems and demonstrate that even lightweight personalization at the interaction level can meaningfully improve user persistence.