Continuous improvement through socio-technical mediation: human agency and informal practices in digitalized public organizations

Continuous improvement has long represented a foundational principle of quality management (Jung and Wang, 2006; Sanchez and Blanco, 2014), both in private organizations and in Public Administration (PA). Rooted in the philosophy of kaizen, it traditionally refers to a gradual, participatory, and learning-oriented process through which individuals and teams continuously refine work practices, improve performance, and enhance organizational outcomes (Iwao, 2017). At its core, continuous improvement has relied on human judgment, experiential learning, and the active engagement of workers in identifying problems and experimenting with solutions (Jurburg et al., 2017; Van Assen, 2021; Yang et al., 2025).

In recent years, however, the context in which continuous improvement unfolds has changed substantially. The diffusion of digital technologies – such as workflow automation, data analytics, artificial intelligence, and algorithmic decision-support systems – has profoundly reshaped organizational processes (Kumar and Shrivastava, 2025; Murire, 2024). In both private and public organizations, digitalization has promised greater efficiency, standardization, transparency, and real-time monitoring of performance (Chauhan et al., 2022). In this sense, digital transformation has often been presented as a natural accelerator of continuous improvement, capable of reducing waste, minimizing errors, and supporting evidence-based decision-making (Buonocore et al., 2026; Che et al., 2023; Von Leipzig et al., 2017).

At the same time, a growing body of literature has begun to question this optimistic narrative. Scholars have warned that the increasing reliance on digital systems may generate unintended consequences for learning, autonomy, and professional judgment (Abdulkareem et al., 2024; Parker and Grote, 2022). Concepts such as automation bias, technological deskilling, and algorithmic control highlight how poorly designed digital technologies can reshape the human foundations of continuous improvement, potentially constraining learning and professional judgment (Tiron-Tudor et al., 2025). Digital systems may, in some contexts, orient organizational behavior toward compliance, routinization, and adherence to predefined procedures (Natali et al., 2025). This tendency appears particularly relevant in public organizations, where digitalization is often shaped by accountability requirements, performance measurement systems, and regulatory constraints (Fleischer and Wanckel, 2024; Marienfeldt, 2024).

Within this debate, two partially overlapping but analytically distinct streams of research can be identified. A first stream examines digitalization primarily as a driver of efficiency and control, focusing on the implementation of digital tools, performance measurement systems, and their impact on organizational outcomes (e.g. Berente and Lee, 2014; Purnamasari et al., 2025). A second stream highlights the implications of digital transformation for professional discretion, work design, and organizational learning, often emphasizing tensions between automation and human agency (Androutsopoulou et al., 2025; Giest et al., 2025). While these studies provide important insights, they tend to focus either on technological infrastructures and measurable performance effects, or on broader questions of governance and discretion. Comparatively less attention has been devoted to how continuous improvement – as an everyday, learning-oriented, and practice-based phenomenon – actually unfolds within highly digitalized public organizations (Tinjan, 2025; Weichselberger, 2025).

In particular, existing research has rarely examined how improvement is enacted at the micro-level of daily work in contexts characterized by strong formalization, algorithmic monitoring, and institutional constraint (van Zoonen et al., 2025). We still know little about how public servants adapt, reinterpret, or sometimes circumvent digital systems in order to sustain meaningful forms of problem-solving and learning (Androutsopoulou et al., 2025; Giest et al., 2025). This gap is especially significant in PA, where digitalization is embedded in dense regulatory frameworks and accountability regimes that may amplify tensions between standardization and professional discretion (Buonocore et al., 2025; Del Barone et al., 2025).

Against this background, this study addresses the following research question: How does continuous improvement unfold in a highly digitalized public organization, and what role do human agency and informal practices play in sustaining learning processes under conditions of strong technological and institutional constraint?

To answer this question, the paper adopts a qualitative case study of a central PA characterized by extensive digitalization, standardized workflows, and performance monitoring systems. By focusing on a least-likely case – where human-centered and practice-based improvement would theoretically be constrained – the study explores whether and how improvement survives beyond formal programs and digital infrastructures.

The contribution of this study is threefold and can be specified more precisely along empirical, theoretical, and conceptual lines.

First, the paper makes an empirical contribution by providing in-depth evidence of how continuous improvement actually happens in a highly digitalized and institutionally constrained public organization. While much of the existing literature discusses digital transformation at a macro or system level (e.g. Bodrožić and Adler, 2022), this study documents the micro-level practices, adaptations, and workarounds through which improvement is enacted in everyday work.

Second, the study advances a theoretical contribution by reframing continuous improvement as a practice-based and socio-technical phenomenon. Improvement develops through situated action, professional discretion, and informal coordination within everyday work practices, alongside the presence of formal systems such as dashboards or structured programs (e.g. Bell, 2005). In doing so, the paper extends existing work on continuous improvement and organizational learning by demonstrating how these processes persist – even in contexts dominated by digital standardization – through the ongoing interplay between technological infrastructures and human agency.

Third, the paper offers a conceptual contribution by bridging the literature on continuous improvement with emerging debates on human-centered digitalization, including Industry 5.0 and PA 5.0 paradigms (Barsekh-Onji et al., 2025; Troisi et al., 2024; Visvizi et al., 2025). It suggests that sustainable improvement in digital public organizations is associated with technological sophistication together with the presence of spaces that support reflection, discretion, and collective learning.

The concept of continuous improvement has traditionally been associated with the philosophy of kaizen, emphasizing incremental change, collective participation, and ongoing learning embedded in everyday work practices (Chung, 2018; Singh and Singh, 2015). Within quality management literature, continuous improvement has long been understood as a cultural and organizational orientation that fosters reflection, experimentation, and shared responsibility for performance enhancement (Nilsson-Witell et al., 2005). Early contributions highlighted the central role of employees' involvement, problem-solving capabilities, and experiential knowledge in sustaining improvement over time, positioning learning as the core mechanism through which quality evolves (Linderman et al., 2004; Murray and Chapman, 2003).

Over time, however, continuous improvement has been progressively formalized through structured methodologies, performance indicators, and standardized procedures (Medne and Lapina, 2019). While such formalization has enhanced consistency and scalability, it has also shifted attention toward measurable outputs and compliance-oriented logics (Bond, 1999). In many contexts, the learning-centered spirit of continuous improvement has become intertwined with, and at times overshadowed by, mechanisms of control and monitoring (Chang, 2005; Maletič et al., 2012). This tension between improvement as an emergent learning process and improvement as a standardized performance regime constitutes a central debate in contemporary quality management (Waardenburg et al., 2025).

A socio-technical perspective helps reconcile these dimensions. From this view, continuous improvement emerges from the interaction between technical systems and human actors (Appelbaum, 1997; Patnayakuni and Ruppel, 2010). Technologies can enable feedback, transparency, and coordination; yet they cannot replace the interpretive and sensemaking processes through which individuals identify problems, question routines, and develop context-sensitive solutions (Gattringer et al., 2021; Mesgari and Okoli, 2019). Continuous improvement extends beyond tools and procedures. It can be understood as a relational and socially embedded phenomenon shaped by professional judgment, shared meanings, and situated practices. (Savolainen and Haikonen, 2007).

The diffusion of digital technologies has profoundly altered the organizational conditions under which continuous improvement takes place (Choi, 1995). Enterprise systems, workflow automation, performance dashboards, and algorithmic decision-support systems promise greater efficiency, traceability, and real-time control. In public organizations, these technologies have been widely adopted to enhance transparency and accountability, often within broader reform agendas linked to digital government and public sector modernization (Cordella and Bonina, 2012; Kakouris and Meliou, 2011; Lapuente and Van de Walle, 2020; McIvor et al., 2002).

Digitalization introduces new technological tools into existing processes while also reshaping the underlying logics of improvement. Automation can generate forms of technological lock-in, where predefined procedures narrow the space for discretion and adaptation (Dolfsma and Leydesdorff, 2009; Witt, 1997). Algorithmic systems may shift attention from reflective judgment to reliance on system outputs, subtly redefining what counts as valid knowledge in decision-making (Parker and Grote, 2022). In such settings, improvement may take the form of optimizing predefined parameters, with less emphasis on processes of collective learning.

Concepts such as automation bias, technological deskilling, and algorithmic control further illuminate these dynamics (Downey, 2021). While digital systems increase efficiency and transparency, they may simultaneously weaken individuals' capacity to interpret complexity and exercise professional judgment (Howcroft and Taylor, 2023; Zhang et al., 2025). In PA, these effects are often amplified by high levels of formalization, legal constraint, and accountability pressures (de Gennaro et al., 2026; Del Barone et al., 2025; Riemma et al., 2025). As a result, digitalization may reinforce procedural compliance and reshape continuous improvement into a compliance-driven or technology-centered process.

A technology-centered model of improvement can thus be characterized by a strong reliance on standardized workflows, performance indicators, and automated feedback loops, where improvement is primarily equated with alignment to system-defined metrics. Such a model emphasizes efficiency, predictability, and measurable outputs, potentially narrowing the interpretive space available to organizational actors.

Against this backdrop, a growing body of research has reasserted the importance of human agency in organizational life (Poulis et al., 2021). From this perspective, improvement develops through how individuals enact, interpret, and occasionally reshape organizational rules within everyday practice (Faulconbridge et al., 2025). Professional discretion – the capacity to exercise context-sensitive judgment – becomes central to sustaining meaningful improvement, especially in complex and ambiguous work environments (Ranerup and Henriksen, 2022; Ranerup and Svensson, 2023).

Informal practices constitute a critical yet often under-recognized dimension of continuous improvement (Minbaeva et al., 2023). Small adjustments, improvisations, workarounds, and collaborative problem-solving activities often remain outside formal documentation while supporting the organization's capacity to cope with variability and unexpected challenges (Mamédio et al., 2025). These practices often constitute an integral part of how improvement takes shape in everyday organizational work.

From a practice-based perspective, continuous improvement emerges through everyday action (Visser and van Hulst, 2024). Learning unfolds through experience, interaction, and gradual refinement of routines (Zeivots et al., 2025). Improvement may therefore remain partially invisible to formal measurement systems, embedded instead in tacit knowledge and situated coordination.

In digitalized environments, such practices adapt to technological constraints. Employees reinterpret digital outputs, selectively use system functionalities, and develop complementary routines to address rigidities (Parker and Grote, 2022). Continuous improvement thus persists even under high automation, but it may become less visible and less formally recognized (Thomas, 2024).

A human-centered view of improvement builds on this insight. In contrast to technology-centered or compliance-driven models, a human-centered perspective conceptualizes improvement as grounded in human capabilities, interpretive agency, and collective learning. Technology can be understood as an enabling infrastructure that supports professional judgment and relational coordination within organizational work practices.

The emphasis on Industry 5.0 (Leng et al., 2024) and PA 5.0 (Barsekh-Onji et al., 2025) reflects an emerging recognition that technological progress must be aligned with human-centered values. Within this paradigm, technology functions as a means of enhancing human capabilities, well-being, and learning (Burzagli et al., 2022). Continuous improvement can therefore be understood as a socio-technical process in which data-driven optimization interacts with human agency and organizational learning (Sawaragi, 2020).

This perspective calls for a reconceptualization of continuous improvement in digital public organizations (Marin-Garcia et al., 2025). Attention can also be directed to how digital systems interact with professional practices and influence opportunities for learning and adaptation, alongside considerations of efficiency and performance indicators (Zia et al., 2025). In PA, where service quality depends on discretion and responsiveness to citizens' needs, such a shift is particularly significant (Zolak Poljašević et al., 2025).

Taken together, the literature reviewed above converge around three analytically interrelated pillars: continuous improvement as a learning-oriented socio-technical process; digitalization as a transformative force that restructures improvement logics; and human agency as a mediating and enabling mechanism through which improvement is enacted in practice.

These perspectives jointly inform the analytical lens of this study. First, we assume that continuous improvement emerges through situated action within everyday organizational practices. Second, we conceptualize digitalization as a structuring force that shapes how improvement processes unfold within organizations. Third, we posit that human agency – expressed through discretion, informal practices, and interpretive work – plays a central role in mediating the relationship between digital infrastructures and learning processes.

On this basis, the study adopts a human-centered and practice-based perspective to examine how continuous improvement unfolds in a highly digitalized public organization. By integrating insights from quality management, socio-technical theory, and research on digital governance, the framework provides a coherent conceptual foundation for analyzing how improvement is enacted, negotiated, and sustained under conditions of technological standardization and institutional constraint.

This study adopts a qualitative case study design aimed at exploring how continuous improvement unfolds within a highly digitalized public organization (Bitektine, 2008). A case study approach is particularly suitable when the research objective is to investigate complex organizational phenomena embedded in real-life contexts, especially when the boundaries between the phenomenon and its context are not clearly defined. Consistent with interpretive and practice-based research traditions, the study focuses on developing analytical insights into how continuous improvement is enacted in everyday organizational life.

The case was selected following a theoretical sampling logic grounded in the notion of a least-likely case. This approach is particularly appropriate when the aim is to generate analytical insight into the conditions under which a phenomenon is expected to be weak or unlikely to occur, while statistical representativeness is not the primary objective.

The selected organization is a central PA operating in a highly institutionalized environment, strongly affected by regulatory reforms, performance measurement requirements, and extensive digitalization initiatives linked to national and European policies, including the implementation of the National Recovery and Resilience Plan (PNRR). Such contexts are typically characterized by high levels of formalization, procedural rigidity, and reliance on digital systems for monitoring, control, and accountability.

From a theoretical perspective, this setting represents a least-likely case for the emergence of human-centered and practice-based forms of continuous improvement. Existing literature suggests that high levels of automation, standardization, and rule-based governance tend to limit professional discretion and reduce opportunities for learning-oriented improvement, favoring instead compliance-driven behaviors and procedural adherence.

Selecting this case therefore constitutes a conservative research strategy. Observing continuous improvement, understood as an emergent, learning-based, and human-driven process, within such a highly constrained and digitalized environment provides analytical support for the argument that improvement develops through the interaction between formal systems, technological infrastructures, and situated organizational practices. This highlights the role of human agency and informal practices in sustaining improvement under conditions characterized by high levels of formalization and digital standardization.

The selected case concerns a central PA operating at the national level, directly affected by recent public sector reforms and by the implementation of PNRR-related projects. For reasons of confidentiality and institutional authorization, the name of the organization is not disclosed. Over the past years, the organization has undergone a profound process of digitalization involving the introduction of digital workflows, standardized procedures, monitoring dashboards, and performance reporting systems aimed at increasing efficiency, traceability, and compliance with national and European regulations.

As is common in many PAs, digital transformation has been accompanied by strong pressures toward formalization, documentation, and accountability. At the same time, the organization continues to rely on the professional expertise of civil servants who are required to interpret regulations, manage complex cases, and adapt standardized procedures to concrete operational contexts. This combination of high formalization and high professional discretion makes the case particularly suitable for exploring how continuous improvement emerges at the intersection between digital systems and human agency.

The study is based on multiple qualitative data sources, allowing for triangulation and increased analytical robustness. Data were collected over an extended period of approximately twelve months, enabling sustained engagement with the organizational context and observation of recurring patterns in work practices.

First, a systematic analysis of organizational documents was conducted. In total, more than fifty documents were examined, including internal regulations, procedural guidelines, digital workflow descriptions, operational manuals, and project documentation related to digitalization initiatives and PNRR implementation. These documents provided insight into the formal design of processes, the intended role of digital tools, and the officially prescribed approaches to performance management and improvement.

Second, the study draws on the analysis of process-related artifacts generated through daily organizational activity. These include digital templates, reporting formats, workflow configurations, and system-generated outputs that reflect how work is structured and monitored in practice. Such materials were collected across multiple organizational units and allowed for a detailed understanding of how improvement logics are embedded in technological infrastructures.

Third, the research relies on systematic observation of work practices and organizational routines. Observations were conducted regularly throughout the data collection period, including attendance at internal meetings, review sessions, and routine operational activities. These observations focused on how digital tools were used in practice, how procedures were interpreted and adapted, and how informal coordination and problem-solving occurred alongside formal processes. Field notes and analytical memos were produced throughout the process in order to document emerging insights and critical incidents related to improvement practices.

Access to the organization was granted through formal authorization by senior management, which allowed the researcher to observe internal processes and consult relevant documentation. The researcher maintained a non-managerial and non-evaluative role, positioning herself/himself as an independent academic observer. This facilitated open access to everyday practices while minimizing the risk of influencing ongoing processes.

The analysis followed an interpretive and inductive logic, inspired by a bricolage approach to qualitative inquiry (Pratt et al., 2022). This perspective frames organizational phenomena such as continuous improvement as requiring a flexible and adaptive combination of theoretical sensitivity, empirical grounding, and iterative sensemaking, developed through the iterative engagement between data and theory.

Data analysis unfolded through an iterative process of moving back and forth between empirical materials, emerging interpretations, and relevant theoretical constructs. The analysis combined inductive coding with theoretically informed interpretation, allowing the study to remain open to unexpected patterns while maintaining coherence with the conceptual focus on continuous improvement, digitalization, and human agency.

A central element of the analytical process was the construction of a data structure (Table 1), which systematically organized empirical observations and progressively abstracted them into analytical dimensions. Following an abductive logic, first-order concepts were derived from recurrent practices, interactions, and situations related to improvement activities. These reflected the language and perspectives of organizational actors.

In a second step, these categories were grouped into more abstract second-order themes, interpreted through existing literature on continuous improvement, organizational learning, and digital transformation. This phase involved an iterative process of comparing empirical patterns with theoretical constructs, allowing emerging themes to be progressively refined and conceptually grounded while remaining closely connected to the data. Particular attention was paid to preserving the link between actors' experiences and the broader interpretive framework, avoiding premature abstraction and ensuring analytical coherence.

Finally, overarching aggregate dimensions were identified, articulating how continuous improvement is sustained in a highly digitalized public organization. These dimensions capture recurrent patterns across data sources and reflect higher-level mechanisms through which digital infrastructures and human agency interact in shaping improvement processes. The identification of aggregate dimensions was considered sufficiently stable when additional data did not lead to the emergence of substantively new themes and when consistent patterns were observed across documents, artifacts, and observations, indicating convergence and analytical saturation.

The data structure presented in Table 1 provides the conceptual backbone of the findings section. The four aggregate dimensions introduced in the Findings directly emerge from this analytical progression, ensuring transparency in how empirical observations were translated into theoretical insights.

To enhance rigor and credibility, several strategies were adopted. Triangulation across documents, artifacts, and observations strengthened the robustness of interpretations. The use of rich contextual description supports analytical generalization.

Given the interpretive nature of the study, reflexivity played a central role. The researcher continuously reflected on assumptions, interpretive choices, and potential biases, particularly given sustained access to the organization. The researcher did not occupy a managerial or consulting role within the organization, which reduced the risk of role conflict. At the same time, close engagement with daily practices required ongoing critical reflection to avoid over-familiarization. Reflexive memos were systematically used to document interpretive decisions and analytical shifts.

Finally, the analytical process was consistently anchored in established theoretical frameworks in continuous improvement, socio-technical systems, and digital governance, ensuring that empirical insights were systematically connected to broader scholarly debates.

The analysis reveals that continuous improvement in the examined public organization unfolds through a complex interplay between digital infrastructures, formalized procedures, and human agency. Improvement takes shape through everyday practices in which employees continuously adapt, reinterpret, and at times work around digital systems in order to ensure the functioning of organizational processes. The findings are presented along four interrelated dimensions that capture how continuous improvement is enacted in a highly digitalized administrative context.

Figure 1 summarizes the process through which continuous improvement emerges in digitalized public organizations. Digital infrastructures structure work processes and introduce procedural constraints. In response, organizational actors exercise discretion and develop informal practices, which generate experimentation and situated learning. Through sensemaking and reinterpretation of digital systems, these practices progressively reframe continuous improvement beyond formal improvement programs.

The first set of findings highlights the ambivalent role played by digital technologies in shaping continuous improvement. Digital systems strongly structure organizational activity by defining mandatory workflows, predefined sequences of action, and standardized performance indicators. In practice, employees described how digital platforms “do not allow skipping steps,” requiring the completion of specific fields before moving forward in the process. For example, in one observed case, a workflow prevented the submission of a file because a non-applicable data field had not been filled, forcing the employee to select a default option simply to proceed.

This episode illustrates how improvement is formally embedded in digital systems through predefined standards and automated controls. Improvement becomes equated with procedural completeness and data consistency. However, such structuring mechanisms also reveal their constraining dimension.

The data show that digitally embedded improvement tends to prioritize compliance and traceability over experimentation. Dashboards and reporting tools continuously track outputs and timelines, reinforcing a logic in which improvement is defined as alignment with predefined indicators. During one internal meeting, for instance, discussion focused exclusively on whether targets had been “formally met,” while employees informally acknowledged that certain adjustments had been necessary to ensure practical feasibility.

Empirically, this suggests that digital systems redefine what counts as visible improvement. Theoretically, it points to a shift from learning-oriented improvement toward compliance-driven improvement. While digital tools enhance standardization and transparency, they simultaneously narrow discretionary space and limit bottom-up experimentation. Improvement becomes structured and stabilized, but less responsive to situational complexity.

Despite the strong structuring effect of digital systems, human agency remains central to the functioning of continuous improvement. Employees engage with digital prescriptions through processes of interpretation and adaptation in their everyday work practices.

For instance, in cases where standardized procedures did not adequately account for case-specific complexities, employees adjusted the sequencing of activities informally before re-entering data into the system. In another example, staff members maintained parallel spreadsheets to track exceptions that could not be fully represented in the official platform. These parallel practices functioned as pragmatic responses that supported the continuity and quality of service.

Such episodes demonstrate that professional discretion plays a key role in sustaining improvement. Adjustments are rarely formalized or recognized in performance indicators, yet they enable processes to function effectively.

From an analytical standpoint, these observations suggest that continuous improvement unfolds through situated adaptation within everyday work practices. The empirical material thus grounds the theoretical claim that human agency mediates the relationship between digital systems and practical outcomes. Improvement emerges through ongoing reinterpretation of formal rules in light of operational realities.

A further key finding concerns the nature of learning underlying continuous improvement. Learning often develops through repetition, interaction, and experience within everyday work practices, alongside more structured improvement initiatives and formal training programs.

For example, employees reported that handling recurring cases gradually enabled them to anticipate system constraints and adjust their actions proactively. Observations revealed moments in which experienced staff informally guided colleagues, explaining “how the system really works” beyond the official manual. These exchanges often emerged spontaneously within daily routines and were embedded in everyday work interactions.

Empirically, this suggests that learning accumulates through practice-based refinement of routines. Theoretically, it indicates that improvement is sustained by tacit and collective learning processes that remain partially invisible to formal measurement systems.

A persistent tension also emerges between digital rationality and practical knowledge. Digital platforms embody simplified representations of work processes, whereas actual cases involve ambiguity and contextual variation. In several observed instances, employees needed to reinterpret system outputs or manually reconcile discrepancies between system-generated data and operational reality.

These observations illustrate how improvement develops through the reconciliation between formal digital logics and situated expertise. It unfolds through ongoing interactions between system constraints and human judgment.

Taken together, the findings indicate a broader reconfiguration of continuous improvement in digital public organizations. Improvement often takes shape as a distributed and frequently less visible process sustained by adaptive behavior, informal coordination, and professional judgment, alongside formal initiatives such as kaizen programs or structured performance cycles.

Digital technologies provide structure, transparency, and control, but they do not automatically generate learning. In several episodes, improvement was observed when employees navigated system rigidities through creative adjustments in their everyday work practices.

The empirical evidence thus supports a reframing of continuous improvement as a socio-technical and practice-based phenomenon. Continuous improvement persists through the ways human actors reinterpret, adjust, and supplement digital systems within everyday organizational practices. Recognizing these informal and often invisible dynamics is essential for understanding how improvement is sustained in contemporary public organizations.

This study set out to examine how continuous improvement unfolds in highly digitalized public organizations, with particular attention to the role of digital infrastructures and human agency in shaping improvement processes. The findings suggest that continuous improvement develops through the ongoing interaction between standardized digital infrastructures and situated organizational practices.

This study contributes to ongoing debates on continuous improvement and digital transformation by showing that improvement in highly digitalized public organizations develops through the interaction between technological systems and organizational practices. The findings extend existing practice-based perspectives by identifying a mechanism through which continuous improvement persists under conditions of high formalization and digital standardization.

By examining a least-likely case, the study shows that digital infrastructures shape where and how human-centered improvement becomes visible within the organization. Improvement becomes partially formalized through dashboards, workflows, and performance indicators, and unfolds within practice-based spaces where discretion, interpretation, and adaptive coordination operate. This dynamic shows how improvement is distributed across multiple organizational dimensions and is enacted through the interplay between digital tools and situated practices.

In doing so, the findings provide theoretical grounding for a human-centered understanding of improvement in digital public organizations. Human agency operates through processes that mediate, stabilize, and reinterpret digital systems within organizational practices. Continuous improvement emerges as a socio-technical accomplishment sustained through ongoing mediation between standardized digital logics and contextual professional judgment.

The theoretical implications of this study can be articulated along three interrelated contributions that refine and extend existing understandings of continuous improvement in digitalized public organizations.

A first theoretical contribution lies in refining existing socio-technical accounts of continuous improvement. Prior literature has acknowledged that improvement involves both formal systems and human participation (Linderman et al., 2004; Nilsson-Witell et al., 2005; Singh and Singh, 2015). However, much of this literature assumes that digitalization embeds improvement more deeply into organizational routines (Söderlund and Pemsel, 2022). The present findings suggest a more nuanced mechanism in which digitalization shapes where and how improvement processes take place within the organization. When improvement becomes tightly associated with standardized workflows and performance dashboards, learning-oriented experimentation may shift outside formal systems. Improvement continues, but it becomes less visible, more tacit, and more dependent on informal practices. This notion of relocation advances theory by specifying how improvement survives under digital constraint. The findings suggest that digitalization redistributes improvement processes across different organizational layers. Formal systems absorb measurable aspects of improvement, while interpretive and adaptive dimensions migrate into practice-based spaces. This conceptual refinement extends socio-technical theory by highlighting interaction dynamics together with shifts in where improvement processes unfold.

A second theoretical advancement concerns the nature of human agency in digitalized public organizations. Existing research often frames agency in relation to automation either as resistance or as diminished discretion (Parker and Grote, 2022; Ranerup and Henriksen, 2022). The findings here suggest a different configuration: agency operates as mediation. Digital systems introduce structured rigidity by codifying sequences, defining categories, and formalizing outputs. Within these arrangements, discretion continues to operate through zones where interpretive work becomes necessary. Public servants engage in reconciling system logic with case-specific realities, translating ambiguous situations into standardized inputs, and compensating for gaps between digital representations and operational complexity. In this sense, human agency plays a structurally significant role in how organizational processes unfold. Improvement becomes the outcome of continuous mediation between technological infrastructures and real-world variability. This reframes digital governance from a substitution logic (technology replacing human judgment) to a mediation logic (technology generating new forms of interpretive labor). Such a reframing contributes to debates on algorithmic governance and public-sector digitalization by foregrounding the relational character of discretion under digital regimes.

A third theoretical implication concerns learning. In digitalized environments, learning becomes embedded in micro-level adaptations and relational exchanges. Employees develop their understanding through repeated interaction with system constraints, case variability, and peer guidance within everyday work practices. This finding refines practice-based theories of learning (Visser and van Hulst, 2024; Yakhlef, 2010) by specifying how digital infrastructures reshape learning modalities. Learning becomes more tacit, incremental, and situated. It is less likely to be formally recognized yet remains central to organizational functioning. By integrating these three insights – relocation, mediation, and situated socio-technical learning – the study advances an integrative framework of continuous improvement in digital public organizations. This framework positions improvement as an emergent outcome of tensions between digital standardization and human interpretive capacity. Importantly, this integrative perspective contributes to debates on Industry 5.0 and PA 5.0 (Troisi et al., 2024; Visvizi et al., 2025) by empirically illustrating how human-centered digitalization operates as an organizational condition for learning and improvement. The findings show that the presence of interpretive space allows employees to engage with digital systems through judgment, adaptation, and experiential learning.

The findings offer implications that go beyond general calls for human-centered reform, pointing instead to the organizational conditions that shape how digital systems and professional practices interact in everyday work.

First, digital system design should incorporate structured flexibility. The observed reliance on informal spreadsheets, exception handling, and parallel routines indicates that rigid workflow architectures displace necessary adaptive practices outside formal systems. Designing platforms that allow contextual annotations, exception pathways, and iterative feedback could reintegrate adaptive improvement into formal infrastructures.

Second, managerial practice should recognize informal improvement as a legitimate organizational resource. In the studied case, much of the effective adaptation occurred through micro-adjustments that were invisible in performance metrics. Managers could institutionalize reflective forums in which such adaptations are shared, discussed, and collectively refined. This would transform isolated coping strategies into organizational learning mechanisms.

Third, public-sector reform initiatives – particularly those associated with large-scale digital programs – should balance standardization with adaptive capacity. Reform frameworks often emphasize harmonization and measurable outputs. However, the findings suggest that excessive procedural rigidity may unintentionally suppress learning-oriented improvement. Embedding qualitative indicators of adaptive practice alongside quantitative metrics could preserve interpretive space.

Finally, training programs for public servants should move beyond technical system use and address interpretive competence. Since improvement depends on mediation between digital logic and practical reality, cultivating judgment, contextual reasoning, and collaborative problem-solving becomes essential.

While the least-likely case logic strengthens analytical inference, studying a single central administration inevitably limits generalizability. Organizational cultures, degrees of digital maturity, and regulatory intensity vary across public institutions. Comparative research could examine whether similar relocation and mediation dynamics occur in decentralized or less formalized contexts.

Second, capturing informal practices empirically presents methodological challenges. Many adaptive routines operate at a tacit level and remain only partially articulated within everyday organizational practices. Although sustained observation mitigated this limitation, some micro-level dynamics may remain partially inaccessible. Future studies employing shadowing, ethnography, or process tracing could deepen understanding of interpretive mediation in digital environments.

Third, the interpretive and abductive approach means that findings are shaped by the chosen theoretical framing. Alternative theoretical lenses – such as institutional theory or critical data studies – might highlight different dimensions of digital governance and improvement. Future research could test and extend the proposed framework across theoretical paradigms.

Finally, longitudinal research is needed to explore whether the relocation of improvement into informal spaces is a stable configuration or a transitional phase in digital transformation. As AI-driven systems become more adaptive and context-sensitive, the relationship between digital infrastructures and human mediation may evolve in complex ways.