Uncovering the failure of Agile framework implementation using SSM-based action research

Organizational learning is crucial for knowledge acquisition, analysis, storage, and dissemination. It integrates individual knowledge into a shared organizational space. The literature extensively explores learning at individual, team, and organizational levels, with individual learning considered a crucial factor in organizational learning (Fiol & Lyles, 1985; Kim, 1993; Hayes & Allinson, 1998). A cycle of experience and knowledge sharing among individuals leads to team learning (Berg, 1993; Hurley & Cunningham, 1993; Bain, 1998; Chan, 2003), and team learning contributes to organizational learning (Senge, 1990; Bennett & O’Brien, 1994; Argyris, 1999; Edmondson, 1999). Knowledge management in software development involves handling complex, time-sensitive issues requiring high-quality knowledge (Tiwana, 2004; Ozer & Vogel, 2015). Agile SD aims to meet business needs, increase customer satisfaction, reduce bugs, shorten development cycles, and adapt quickly to changing requirements (Boehm & Turner, 2003; Parrish et al., 2004; Miller & Larson, 2005; Joey Cho, 2010). Despite these advantages, Agile SD implementation frequently fails. This study investigates existing frameworks supporting Agile SD implementation, particularly focusing on the learning process and human factors. The research questions are: (1) Can current frameworks uncover the actual issues in Agile SD implementation? (2) How do frameworks support the exploration of the learning process during Agile SD implementation?

The study conducted a systematic literature review of 1359 studies on Agile SD implementation, focusing on knowledge management and learning process approaches. The review used six criteria: study population (teams/groups), intervention (frameworks supporting Agile SD), outcome variables (learning process performance), time period (1990-2018), language (English), and methodological quality (top-tier journals). The initial search yielded 87 papers, which were narrowed to 22 discussing relevant frameworks and then 14 employing knowledge management or learning processes. The analysis of these 14 studies revealed that none fully addressed the research objectives. Some studies focused on frameworks incorporating learning processes (Qumer & Henderson-Sellers, 2008; McAvoy & Butler, 2009; Hoda & Noble, 2017; Pries-Heje & Baskerville, 2017), but they lacked detailed explanations of their framework development and practical application. Existing frameworks, while offering theoretical insights into knowledge management and learning, often fell short in empirically demonstrating their effectiveness in resolving practical Agile SD implementation challenges. The lack of an interpretivist approach to understanding the complexities of Agile SD implementation further motivated this study to explore an alternative methodology.

Given the challenges in capturing tacit knowledge, which is crucial in software development and often difficult to articulate and share, the study employed Soft Systems Methodology (SSM)-based Action Research (AR). SSM-based AR is an experience-based approach that considers both 'reality' and 'actuality', accommodating diverse viewpoints and promoting systemic learning. It is suitable for addressing complex organizational issues where tacit knowledge plays a significant role. The methodology involved several steps: (1) entering the problematic situation, gathering data through observation and interviews; (2) expressing the problem systematically using rich pictures; (3) formulating root definitions of relevant systems using CATWOE analysis (Customers, Actors, Transformation, Weltanschauung, Owner, Environmental constraints); (4) building conceptual models of these systems, considering efficacy, efficiency, and effectiveness; (5) comparing the conceptual models with the real-world situation to facilitate a structured discussion and identification of potential improvements; (6) defining desirable and feasible changes; and (7) taking action and reflecting on the results. The study applied this methodology to a case study involving a corporation that had experienced difficulties with Agile SD implementation. Six root definitions were developed, each addressing a specific aspect of the problematic situation. Conceptual models were created for each root definition, visually depicting the activity systems involved and potential areas for improvement. The comparison between the conceptual models and the real-world situation revealed gaps in knowledge and understanding, highlighting the need for improved communication, collaboration, and knowledge management practices.

The application of SSM-based AR to the case study revealed several key issues in the corporation's Agile SD implementation: (1) a lack of understanding of Agile SD principles and practices; (2) poor communication and knowledge sharing between the IT team and other business units; (3) unclear roles and responsibilities, leading to confusion and inefficiency; (4) inadequate documentation of requirements, resulting in misunderstandings and delays; (5) difficulties in capturing and sharing tacit knowledge within the team; and (6) insufficient mechanisms for storing and retrieving information. These findings underscore the importance of considering both explicit and tacit knowledge in Agile SD implementation. The comparison between the conceptual models (representing idealized Agile SD practices) and the real-world situation (the corporation's actual implementation) exposed the gaps between the intended practices and the actual implementation. Uchiyama's learning model, which identifies two stages of learning within the SSM-based AR framework, was used to analyze the learning process. The first learning stage involved converting tacit knowledge into explicit knowledge through the structured discussion prompted by the model comparison, while the second stage, which would be explored in future research, involves internalizing tacit knowledge obtained from practical implementation. The study identified potential actions that could be taken to address each of the identified issues. These actions ranged from defining and enforcing clear roles and responsibilities to implementing improved communication protocols and knowledge-sharing mechanisms. The feasibility of each proposed action was assessed, considering the organizational context and available resources.

This study's findings directly address the research questions. The results demonstrate that current frameworks for supporting Agile SD implementation often fail to adequately address the specific challenges related to tacit knowledge and the learning process. SSM-based AR offers a more effective approach by providing a structured framework for uncovering tacit knowledge and facilitating a systematic discussion of potential improvements. The key contribution of this study lies in its demonstration of how SSM-based AR can be used to empirically explore the learning process in Agile SD implementation. The case study highlights the importance of explicitly acknowledging and addressing the challenges associated with tacit knowledge conversion, communication, and collaboration within Agile teams. The study's insights are relevant to both practitioners and academics in the field of software development and knowledge management. The findings provide a practical framework for improving Agile SD implementation, emphasizing the human element and the crucial role of knowledge sharing. This methodology can provide a more nuanced understanding of Agile SD failures, allowing organizations to identify and overcome the challenges encountered during its implementation.

This study makes several key contributions. It empirically demonstrates the limitations of existing frameworks in capturing the learning process within Agile SD implementation. It proposes SSM-based AR as a more effective approach, emphasizing the crucial role of tacit knowledge and providing a structured framework for addressing challenges. The case study highlights several practical issues related to Agile SD implementation and offers recommendations for improvement. Future research should focus on completing all stages of the SSM-based AR framework, examining the second learning stage and further evaluating the effectiveness of the proposed improvements. A larger-scale study involving multiple organizations would also strengthen the generalizability of the findings.

The study's reliance on a single case study limits the generalizability of the findings. The qualitative nature of the data limits the ability to make quantitative generalizations. The study primarily focused on the first stage of Uchiyama's learning model, with the second stage to be investigated in future work. Further research would benefit from exploring the longitudinal impact of the proposed improvements, as well as examining the long-term effects of the changes implemented. The specific organizational context of the case study may limit the direct applicability of the findings to other contexts.