Portfolio Management in Large Scale Agile Software Development

As organisations scale agile software development across concurrent teams, the governance problems shift upward. Team-level agility cannot resolve what is, at root, a portfolio coordination problem: who decides which work gets done, in what sequence, with what resources, and against what strategic objectives? Traditional portfolio management models were designed for sequential, plan-driven delivery. They were not built for the iterative prioritisation, continuous investment decisions, and cross-team dependencies that large-scale Agile environments generate. The practitioner community has responded with a range of frameworks and guidance. The academic literature has not kept up: the evidence base for Agile Portfolio Management (APM) governance remains fragmented and considerably thinner than what practitioners have produced independently. This thesis investigates that gap.

The study used a Multivocal Literature Review (MLR), combining 23 white literature sources retrieved from four academic databases (Scopus, ACM Digital Library, IEEE Xplore, and EBSCO) with 18 grey literature sources such as practitioner frameworks, enterprise consulting reports, and professional guidance documents. Reflexive thematic analysis following Braun and Clarke (2019, 2021) produced 923 coded extracts and 16 merged themes, organised across four research questions (Braun & Clarke, 2019, 2021b).

Five themes explain why organisations adopt APM (RQ1): strategic alignment, adaptive empirical governance, resource and flow management, behavioural enablers, and Project Management Office (PMO) infrastructure. Where APM is absent (RQ2), two failure patterns dominate namely governance breakdown with strategic drift, and structural fragmentation with coordination failure which account for 127 coded extracts. Three barrier types block the transition from traditional Project Portfolio Management (PPM) to APM (RQ3): political, cultural, and leadership resistance (the most prominently evidenced, at 48 extracts); structural and scalability constraints; and financial and capability gaps. The RQ4 evidence suggests that effective APM at scale draws on six governance components working in combination:: adaptive steering mechanisms, institutional architecture and PMO roles, decentralised team autonomy, portfolio visibility and financial governance, stakeholder engagement routines, and interdependency management.

The study makes three contributions to the academic literature. First, the analysis identifies an adoption paradox: organisations that most need APM governance are often the least positioned to adopt it. Second, the RQ2-to-RQ4 challenge-solution correspondence, surfaced through systematic thematic synthesis of 923 coded extracts from 41 sources, yields a diagnostic-prescriptive framework. Third, the evidence suggests that APM governance is better framed as an organisational capability than as a methodology selection. Portfolio managers can use the framework to sequence governance investments against their specific needs and organisational context. Keywords: Agile Portfolio Management, portfolio governance, large-scale software development, Multivocal Literature Review, scaled Agile, project portfolio management, adoption paradox.