Does Software Process Ambidexterity Lead To Better Software Project Performance?

Plan-based and agile software development processes seem diametrically opposed in their approaches, with the former emphasizing discipline and control and the latter promoting flexibility and improvisation. Similar tensions in organizational contexts where efficiency versus flexibility considerations simultaneously jostle for management attention has led to the recognition that ambidexterity or the ability to manage seemingly conflicting demands is an important precursor to organizational success. In this study, we extend the idea of ambidexterity to software development processes and empirically examine the performance implications of the ability of software project teams to pursue process designs that simultaneously facilitate both control and flexibility. Utilizing data from a quasi-experiment involving 424 large commercial software projects of a multinational software services firm, we employ a potential outcomes empirical methodology to examine the causal linkage between software process ambidexterity and project performance. Our results show that projects that encountered frequent requirement changes, larger and complex code-bases, new technologies, higher levels of enduser engagements, and smaller, inexperienced teams tend to choose ambidextrous software process designs over a pure plan-based approach. We find that ambidextrous process design positively contributes to better project performance, including on the average about 9% higher productivity, 50% reduction in delivered defects, 12% reduction in internal defects, and 3% improvement in overall profitability. Complementing the archival data analysis with an in-depth qualitative study of the projects pursuing ambidextrous process designs, we enumerate the different mechanisms employed by the project teams to balance control requirements with needs for realizing flexibility. We discuss the implications of our results and elucidate potential pathways to achieve and sustain ambidextrous process designs in software firms.