Developing countries are increasing their adoption of information systems at the country level now. One important aspect distinguishing the implementation of information systems in developing countries from that in developed countries is that developing countries usually lack the resources and capability for training and support, and the workers need to learn to use the system from their own experience. Thus, a better understanding of the workers’ learning-by-doing after the implementation of an information system in developing countries may have important theoretical and practical implications, but empirical evidence on this issue remains limited. This dissertation seeks to fill in the gap by investigating workers’ learning-by-doing after the implementation of an information system at two levels. First, it studies how an individual customs agent’ experience preparing and submitting customs documents influences her performance in document preparation and submission tasks. Second, it also examines how an agent-inspector dyad’s experience working together affects the performance of customs inspection tasks completed through the cooperation of the dyad. The first chapter provides an overview of the dissertation. The second chapter examines how the relatedness of workers’ prior experience affects their learning-by-doing and operational performance in service work. Prior research has viewed relatedness along a single dimension. However, tasks and the underlying knowledge required for task performance can vary along multiple attributes. This chapter extends prior conceptualizations of relatedness by defining it as a multi-dimensional construct and also accounting for the level of task relatedness between different categories in each task dimension. It separates the level of workers’ experience from the relatedness of their experience, and then link the two constructs to workers’ task performance, including their efficiency and quality. Analyzing data on the processing of 998,258 import customs declarations in Costa Rica from 2006-2010, the second chapter finds that customs agents, the major workers processing the customs declarations, learn from their experience to improve their time to complete the task but not their quality of completion. Moreover, it finds that the relatedness of customs agents’ experience to their current task is positively related to the quality of task completion but has a U-shape relationship with completion time, such that the completion time first decreases with and then increases with an increase in customs agents' experience relatedness. The chapter also finds that the impact of customs agents’ experience relatedness is enhanced when the agents have more experience. Overall, the results highlight the role of experience relatedness in workers’ performance in learning-by-doing service work, and help to identify ways for managers to improve different operational performance measures. Many service tasks are completed by dyads rather than by an individual worker. In this setting, the individuals in the dyad not only need to acquire knowledge about the task, but also have to learn to work with each other. Thus, individuals’ experience working together may have significant performance implications for dyads. However, this effect remains largely unexamined, especially when there are conflicts within the dyad. In the third chapter, it theorizes how a dyad’s experience working together influences the dyad’s task performance, and label it as a learning-by-working-together effect. The chapter further proposes that the impact of dyad experience can vary across tasks with different levels of complexity, goal conflict, and combinations of the two. It examines learning-by-working-together in a setting where there is goal conflict, but the dyad must work together to complete the task: customs inspections. Based on a field study on data of 323,520 customs inspections in Costa Rica, the third chapter shows that the number of prior interactions between a customs agent and a customs inspector is positively associated with the agent-inspector dyad’s efficiency in customs inspection. In addition, it demonstrates that the impact of an agent-inspector dyad’s experience working together is greater for high-complexity tasks than for low-complexity tasks, and weaker for high-conflict tasks than for low-conflict tasks. It also shows that due to a joint effect of task complexity and task-level goal conflict, dyad experience exhibits the largest impact on the performance of high-complexity, high-conflict tasks. The chapter discusses the implications of our results for the study of learning curves and for the practice.