Maximizing value creation in agile sprints

Abstract

Agile software development principles prioritize the delivery of value through working software. Earlier value creation is preferred to reduce the time to market and get sooner feedback from customers. A challenge in planning agile sprints to achieve value upfront is the tension that exists between the value, resources, size of each feature or story deliverable, and the dependencies among them. While the role of effort and resource constraints in value creation has been studied extensively, the role of dependencies has not been fully addressed in the agile context. In this thesis, we propose a framework to improve value delivery in agile software development by decoupling cyclic dependencies to achieve more robust multi-sprint plans in a scaled agile environment. We analyze this novel approach using an arbitrary test dataset to demonstrate how different decoupling methods yield different value trajectories. We also suggest an optimization method to maximize such value creation through sequencing by simultaneously considering timing, dependencies, and resource allocation. We perform a brute-force optimization approach on the test dataset to demonstrate how more rapid value creation can be achieved over multiple sprints.