Organisational Structure Patterns in Agile Teams: An Industrial Empirical Study
This research addresses the problem of optimizing team structures for architects in agile software development, though it is incremental as it builds on existing mirroring strategies with empirical data.
The study investigated organizational structure patterns in 30 agile software teams, identifying 7 recurring patterns with one occurring over 37% of the time, which is linked to young communities and higher architecture issues, and found a negative correlation between an organizational measure and architecture issues.
Forming members of an organization into coherent groups or communities is an important issue in any large-scale software engineering endeavour, especially so in agile software development teams which rely heavily on self-organisation and organisational flexibility. To address this problem, many researchers and practitioners have advocated a strategy of mirroring system structure and organisational structure, to simplify communication and coordination of collaborative work. But what are the patterns of organisation found in practice in agile software communities and how effective are those patterns? We address these research questions using mixed-methods research in industry, that is, interview surveys, focus-groups, and delphi studies of agile teams. In our study of 30 agile software organisations we found that, out of 7 organisational structure patterns that recur across our dataset, a single organisational pattern occurs over 37% of the time. This pattern: (a) reflects young communities (1-12 months old); (b) disappears in established ones (13+ months); (c) reflects the highest number of architecture issues reported. Finally, we observe a negative correlation between a proposed organisational measure and architecture issues. These insights may serve to aid architects in designing not only their architectures but also their communities to best support their co-evolution.