2.1 Tension between established and renewed technology usage

Because it replaces the digital platforms and infrastructures that organizations rely on for day-to-day operations, technology renewal involves changing not only technologies but also, and more importantly, their use. Implementing complex technologies inevitably impacts organizational structures, processes and cultures (Fichman & Kemerer, 1997; Gregory et al., 2015; Lyytinen & Damsgaard, 2001; Robey, Ross, & Boudreau, 2002); as a result, it may be thwarted by user resistance fuelled by a desire to continue using familiar structures and technologies (Polites & Karahanna, 2012). Knowledge barriers can also make implementing new technology difficult (Attewell, 1992), and users may address such barriers by constructing workarounds (Alter, 2014; Azad & King, 2008; Gasser, 1986), devising shadow systems (Behrens, 2009; Furstenau, Rothe, & Sandner, 2017; Strong & Volkoff, 2004), and engaging in reinvention (Boudreau & Robey, 2005) to preserve established use patterns. Polites and Karahanna (2012) further note that user resistance may continue despite their perception of a new technology's relative advantages, expressing a status quo bias in which they “disproportionally make decisions to continue an incumbent course of action, rather than switching to a new (potentially superior) course of action” (Polites & Karahanna, 2012, p. 23). User resistance may also arise due to a perception of high switching costs (Kim & Kankanhalli, 2009) and to the inertia generated by the installed base in digital infrastructures (Hanseth & Lyytinen, 2010). Such observations emphasize how existing technology use may adversely affect implementation of complex networked technologies, such as digital platforms and infrastructures (Gregory et al., 2015; Lyytinen & Damsgaard, 2001). Given the technology renewal paradox, we, therefore, expect an amplification of tensions between the established use patterns of existing technologies and the emerging requirements related to using new technologies across individual, group and organizational levels.

2.2 Tension between deliberate and emergent renewal practices

Enterprise systems research has shown that implementing digital platforms requires organizations to simultaneously redesign technology, structures, processes and skills (Robey et al., 2002). Consequently, these implementations are associated with long-term commitments and large economic costs; both factors make the implementations prone to failure and also make trajectories difficult to reverse once the projects are initiated (Barker & Frolick, 2003). Similarly, research has demonstrated that implementing new technology is often a slow and painful process (Lyytinen & Damsgaard, 2001; Wessel et al., 2020). Furthermore, because the technologies are subject to interpretive flexibility, the requisite intertwining of technologies with human agency and use contexts produces unpredictable discrepancies between intended and actual practices (Orlikowski, 1992). Successfully implementing complex digital technologies, therefore, requires dedicated implementation strategies, top management support and clear communication channels (Nah & Delgado, 2006). Managers are further advised to assume leadership of change by “orchestrating and coordinating the efforts of multiple change leaders throughout the organization” to effectively facilitate employees' behavioural change (Martin & Huq, 2007, p. 136). While undoubtedly important, such deliberate leadership must contend with and adapt to strongly confounding factors as organizational actors engage in local innovation, opportunistic structural shifts and emergent improvisations in response to unanticipated events (McGann & Lyytinen, 2005; Orlikowski, 1996, 2000). In a similar vein, researchers have argued that organizational changes do not always follow rational models based on clear goals, strategies and stable configurations of participants (Eisenhardt & Zbaracki, 1992; Gregory et al., 2015); instead, many organizations may follow ambiguous processes characterized by fluid participation, problematic preferences and unclear digital technology (Cohen, March, & Olsen, 1972). We draw on these insights to posit that technology renewal fosters tensions between deliberate and emergent renewal practices as actors engage to support or resist organizational efforts to replace existing digital platforms and infrastructures with new ones.

2.3 Tension between inner and outer renewal contexts

Research on implementing complex technologies demonstrates that external pressures can also shape agendas and motives for technological change (Liang et al., 2007; Teo, Wei, & Benbasat, 2003). Among these external pressures are technology standardizations (Lyytinen & King, 2006), digital platform ecosystems (Rolland et al., 2018), governmental mandates (Brown, Massey, Montoya-Weiss, & Burkman, 2002) and managerial fashions (Wang, 2010). At the same time, organizations are permeated by internal conflicts between organizational entities and levels (Bacharach, Bamberger, & Sonnenstuhl, 1996; Gregory et al., 2015), within which situated agendas and cultures may flourish, establishing their own rationales, practices and implications for technology renewal (Hoffman & Klepper, 2000). Such internal cultures may be at odds with external renewal agendas for rationalization and standardization (Hanseth, Jacucci, Grisot, & Aanestad, 2006). We, therefore, submit that the technology renewal paradox fosters tensions between inner and outer contexts, where inner contexts incorporate the culture of an organizational entity (including social norms, strategic objectives and management structures), while outer contexts incorporate the environment that the organization operates in (including social, competitive, economic and political factors) (Pettigrew, 1987).

As Table 1 summarizes, we have identified and conceptualized three distinct yet related tensions pertaining to technology renewal as a paradoxical digital transformation process. We base these tensions on our review of a broad range of extant research on IT-related change. The bulk of the reviewed literature addresses implementation of new technologies into organizations outside the context of renewal initiatives. So, while the insights still apply, we posit that technology renewal adds further complexity because it replaces an organization's foundational technologies, thus creating unavoidable functional and operational overlaps between legacy and new technologies. As foundational technologies are business critical, replacing them involves making sure that organizational processes and functions can continue to operate even if the new foundational technology would falter during its early stages. Furthermore, the identified tensions represent composite meta-level concepts that cannot be observed empirically. Instead, the tensions typically manifest in different situations during the renewal initiative, wherein specific organizational conditions and perceptions render them salient and provoke organizational responses. As such, we posit that it is possible to reveal the generative structure of how and why paradoxical tensions affect the trajectory of renewal initiatives by empirically identifying and analysing how the tensions manifest and how the resulting organizational responses impact the existing course of action.

3.1 Data collection

We collected data through two engagements and two follow-ups (Van de Ven, 2007) with HealthOrg. The first engagement took place from August 2005 to May 2006, the second engagement from May 2009 to December 2009, and the two follow-ups in 2011 and 2020.¹ During the first engagement, we used an exploratory approach grounded in problems related to the new digital platform as perceived by actors at HealthOrg Headquarters. Our second engagement was more focused and comprehensive, explicitly targeting the continued use of the two overlapping and largely incompatible digital platforms. Our understanding of how the tensions stemming from technology renewal's paradoxical nature affected the trajectory of the HealthOrg renewal initiative thus developed over several years. To facilitate triangulation, we acquired data from different sources as a way to develop “converging lines of inquiry” (Yin, 2013, p. 120) and corroborate evidence of tensions and consequential responses.² Although our primary data sources consisted of interviews and documents, we also used several other sources, including a qualitative survey (see Table 2).

All interviews were semi-structured, open-ended and relied on interview protocols (Kvale, 1997) to ensure a systematic data collection approach (Patton, 2002). For the first engagement, we developed two protocols: One for managers who oversaw the new digital platform's acquisition and implementation, and one for other Headquarters' employees. For the second engagement, we focused on the on-going use of two competing platforms in relation to HealthOrg management, Headquarters employees, and IT department staff. The first author conducted all interviews and tape-recorded, transcribed, and supplemented them with notes taken during the interviews. The selection of interviewees for the two engagements differed to some extent. In the first engagement, a project manager for the new platform helped us select respondents, supplying us with an extensive list of Headquarters employees. From this list, we chose interviewees from each Headquarters department. Given time constraints and the fact that some employees were unable or unwilling to schedule a meeting, we interviewed employees from only five of the seven departments, along with two project managers, in the first engagement. In the second engagement, we attempted to interview all individuals included in the first engagement, as well as representatives from the management board and IT department. However, as Table 3 shows, two interviewees from the first engagement could not participate, and some interviewees had changed roles between the first and second engagements.

Having unlimited access to documents helped us develop a broad contextual frame of reference for interpreting interview statements. For the second engagement, we also drew on an internal survey and extensive data from a qualitative survey we conducted during the first engagement. Although case studies typically do not include surveys (Yin, 2013), our qualitative survey provided a broad range of data on use and non-use of the two competing platforms, as well as the respondents' perceptions of the renewal initiative. Most survey questions were open-ended, and we distributed the survey electronically to all Headquarters' employees who had access to both platforms. Finally, the first author confirmed our findings by presenting and discussing our empirical analyses with HealthOrg representatives in two separate workshops and through a final report that we provided to the organization.

3.2 Data analysis

In qualitative process research, time provides a fundamental anchor in explaining why and how event sequences are structured and lead to certain outcomes, such as virtuous and vicious cycles of change (Langley, 2007; Pettigrew, 1992; Van de Ven, 2007). However, the empirical material that is subjected to process analysis is frequently limited to representing what Pentland (1999) refers to as surface structures. Such structures are represented by how the interviewed actors and other data sources order the issues, experiences and stories in time. Thus, Pentland (1999, p. 722) argues that “to describe a process, one needs event sequences. But to explain a process, one needs to identify the generative structures that enable and constrain it.” In a similar vein, Van de Ven and Poole (1995) stress the importance of moving beyond merely sequencing empirically observed events, arguing that process analysis should strive to uncover the generative mechanisms that, based on manifestation of deep structures at particular junctures, drive focal processes. Thus, based on extant research, we adopted a paradoxical perspective as our analytical framing with a priori and theoretically driven assumptions about both the technological renewal's paradoxical tensions (Table 1) and how the dynamics of renewal trajectories may be empirically revealed and subsequently theorized. In the following, we describe the three consequential steps through which we analysed and theorized the rich data.

In the first step, we coded the empirical material. To do this, we conducted what Pentland (1999) refers to as descriptive surface structure coding to capture and categorize significant issues, decisions, activities and statements, along with the time at which they occurred. We also applied temporal bracketing (Langley, 1999) to identify and explicate evidence of experienced tensions. We thus considered critical junctures in the temporal data as shifts in dominant tensions. To identify these tensions, we followed Cho, Mathiassen, and Robey (2007), focusing on observed conflicts (Bjerknes, 1991), specific interests, cultural assumptions and institutionalized values (Robey et al., 2002; Robey & Boudreau, 1999), as well as conflicts between different organizational units and levels (Bacharach et al., 1996). Through this process, we identified conflicting requirements to technology, conflicting approaches to change and opposing forces related to different contexts. We initially identified 16 tensions; we then compared and contrasted them to identify similarities and potential overlaps. This resulted in a final explication of six tensions that dominated and shaped the trajectory of HealthOrg's renewal initiative. Thus, although organizational change initiatives inevitably entail a multitude of micro-level activities, conflicts and tensions, our analytical efforts resulted in an aggregated view that revealed six tensions that were crucial in shaping the process. We categorized each tension in terms of the three paradoxical tensions in Table 1. We then mapped out the timeframe during which each tension was in play, identified the associated organizational responses, and traced the impacts of the responses on the overall renewal trajectory. Finally, we framed the duration of each tension as a stage in the overall technological renewal process.

In the second step, we created a process narrative of the technological renewal initiative and how it unfolded within HealthOrg. In addition to explicating the process dynamics within each initiative stage, this approach revealed how the organizational responses at each stage triggered subsequent tensions. This, in turn, illuminated the interplay between tensions and responses as a generative mechanism that, based on manifestations of paradoxical tensions, drove the renewal trajectory over the observed 9-year period. In accordance with Pentland's (1999) recommendations for process narratives, we included (a) a sequence of stages over time; (b) focal actors within HealthOrg; (c) a narrative voice; (d) a clear frame of reference to the impacts of tensions and responses; and (e) other relevant content and context indicators.

Our third and final step combined the empirical account (Table 4) with our upfront conceptualization (Table 1). This allowed us to further theorize the generative mechanisms that drove the change trajectory at HealthOrg Headquarters, as we describe in the “Theory Development” section below and summarize in Table 6. IS theories vary considerably in scope, level of generalization and degree of causality (Gregor, 2006). We focused on mid-range theorizing, as our ambition was to develop a clearly delineated theory of “bounded relevance and validity” (Avgerou, 2013, p. 401) that could explain how and why paradoxical tensions affect the trajectory of a technology renewal initiative. To this end, mid-range theory integrates general theory and empirical observations (Merton, 1967), and involves a limited and clearly delineated scope, moderate levels of abstraction and statements related to the particular phenomenon under investigation (Gregor, 2006).

4.1 A new platform (November 2002-March 2004)

In this context, initiating a renewal initiative was an attempt to address this tension through a considerable investment in Beta, a comprehensive ECM platform. Beta was implemented in spring 2004, with extensive user training to facilitate assimilation. The platform was chosen because it focused on efficiency through increased standardization and, like Alpha, it also had generic capabilities to function as a development platform to cater to specific needs as necessary. It was, therefore, positioned as a viable and capable long-term alternative. Although Alpha was the legacy platform, IT managed it, while Beta was managed by the secretariat because of its responsibilities for record-keeping and process management. However, Beta was rolled-out without clearly communicating how it related to Alpha, which was the backbone of day-to-day Headquarters' operations. Alpha thus continued to function as the organization's main digital platform. Headquarters had attempted to decisively and constructively address diverging needs and requirements stemming from both within and outside the organization. Thus, the chosen approach was built on the expectation that staff members would gradually transition to the new platform because the legacy platform was deeply ingrained in the organization and involved a multitude of specific applications catering for specific needs. Thus, the gradual transition was deemed as the most pragmatic and viable way forward in order to maintain efficiency while simultaneously moving to the new platform.

4.2 Diverging user communities (April 2004-February 2006)

Although the goal was to solve Headquarters' pressing information management problems, implementing Beta without clearly articulating Alpha's future quickly created divergent user communities with different requirements and patterns of information management. Unsurprisingly, some individuals and units began transitioning to Beta, but others did not. This situation rapidly fuelled a tension between the established technology usage within most departments supported by Alpha and the renewed technology usage focused on information creation and sharing across all departments supported by Beta.

For its part, Beta was a highly adaptable, process-controlled ECM platform with elaborate functionalities for record-keeping, document management and process support; it also included administration and advanced search capabilities. Thus, if widely adopted, Beta was capable of serving as a common digital platform and providing better support than Alpha.

Faced with increasingly conflicting use patterns, the emergence of two distinct user communities and having insufficient power to directly change the status quo, in August 2005, the secretariat initiated a collaborative research project with the local university to analyse problems and propose potential remedies. The research project lasted 7 months, during which time Headquarters continued to accommodate the two diverging user communities based on functionally overlapping and partially competing digital platforms.

4.3 Ambiguous management mandate (March-December 2006)

During March and April 2006, the university researchers shared their findings in two workshops with representatives of the secretariat, the management board and the IT strategy group. Briefly, their findings were that: Beta's purpose was unclear to many employees; Beta (unsurprisingly) duplicated the functionality of several existing but incompatible technologies, especially Alpha; and it was generally impossible to exchange information between Alpha and Beta. Thus, the researchers concluded that the original information management problems had been exacerbated. These findings clearly showed that Headquarters faced a tension between emergent renewal practices, in which staff increasingly behaved as if Beta was dedicated to a subset of distinct activities involving record-keeping within the secretariat, and deliberate renewal practices, in which the secretariat continued to promote Beta as the new foundational digital platform for all departments.

The management board recognized that it needed to take action; it responded in early August 2006 by mandating the use of Beta for all information subject to record-keeping within Headquarters, which, in practice, involved most of the information produced within the organization. The decision was communicated to all staff in a written memo that clarified the distinct roles of the two platforms, and also explained the long-term strategy: A complete transition to Beta, and a gradual dismantling of Alpha. The management mandate was nonetheless ambiguous and failed to specify follow-up procedures to enforce it. As a result, the polarized perceptions of, and approaches to, the new digital platform persisted.

4.4 Strengthening the new platform (January-December 2007)

The application quickly became widely used and, therefore, strengthened Beta's overall strategic position. However, this organizational response did not address the tension between the traditional technology usage and the renewed technology usage based on Beta, nor did it alter the trajectory of relying on multiple overlapping digital platforms.

4.5 Upgrading the legacy platform (January 2008-June 2009)

During spring 2008, the slow but steady increase in Beta adoption continued. There was now widespread recognition within Headquarters that the simultaneous use of two platforms caused unsustainable inefficiencies in information management. Specifically, staff became increasingly aware of the tension between inner renewal contexts, where there was strong desire to continue idiosyncratic practices within departments, and outer renewal contexts, with increasing pressure to improve and standardize information creation and sharing across the organization.

The new version also included functionality for project coordination, which was well addressed by Beta. While this organizational response removed the uncertainty related to Alpha's future within Headquarters, it did nothing to change the underlying tensions. Instead, it further propelled the failing course of action by deepening the original problems and further increasing the functional overlap between the two platforms.

4.6 Disconnected management practices (July 2009-October 2011)

Thus, from the IT strategy group's perspective, decisions regarding the production, storage, distribution and access of information were beyond the technologists' remit. As the head of the group put it: Someone else should assume responsibility for such strategic issues. Thus, following implementation of the new Alpha version, a tension emerged between emergent renewal practices that focused on coordination to manage information and technology, and deliberate renewal practices that continued to rely on management of information and technology as separate activities.

However, beyond the investigation, no explicit responses addressed the tension. Instead, the IT strategy group continued to regard information management as an important and separate strategic issue, while the management board expected the IT strategy group to deal with it as part of its technology management strategy. As a result, Headquarters continued to rely on multiple, overlapping digital platforms, and the disconnected management strategies offered little hope that the resulting tensions would be effectively addressed.

In October 2011, nearly 9 years after it had decided to launch a technology renewal initiative to address pressing information management problems, HealthOrg was continuing its failing course of action, despite broad awareness of its negative impacts. Although the six identified tensions that played out during the initiative all triggered organizational responses, none of these responses were substantive enough to steer the renewal process towards completion. Instead, each of the responses triggered new tensions and responses that further reinforced the failing trajectory.

4.7 Follow-up (June 2020)

5.1 Manifestation of paradoxical tensions

Drawing on paradoxical thinking (Cyert & March, 1963; Hargrave & Van de Ven, 2017), we view the paradoxical tensions (Table 1) as meta-level concepts that organizational actors may experience as salient tensions at specific junctures of a technological renewal initiative (Smith & Lewis, 2011; Wareham et al., 2014). That is, when these tensions manifest in practice, they assume specific material characteristics that present organizational actors with conflicting choice opportunities (Farjoun, 2010; Gregory et al., 2015). In our analysis of HealthOrg's technology renewal, we drew on paradoxical thinking to reveal evidence of how the tensions manifested as empirically observable salient tensions during the six stages of technology renewal (Table 4).

The tension between inner and outer renewal contexts manifested during two stages of the renewal process. During the first stage, HealthOrg's highly idiosyncratic information management practices based on the legacy platform clashed with the governmental mandate to improve and restructure the organization's digitalized services; this clash was intensified by a lack of internal coordination and control. Our analysis of this manifestation demonstrates how internal structures, processes and cultures may be incongruent with each other (Hoffman & Klepper, 2000), and how external pressures, such as strategic moves by competitors, or governmental mandates (Brown et al., 2002), can shape transformation agendas in ways that clash with internal conditions (Gregory et al., 2015; Liang et al., 2007). Moreover, during stage five, this tension resurfaced, albeit in a different manifestation. During this stage, Headquarters found itself deeply reliant on two separate digital platforms to perform its business activities. As such, the tension manifested as conflicting strategic objectives; in isolation, the objectives made sense, but from an organizational perspective, they were clearly at odds with each other. These findings suggest that managers of technology renewal initiatives will likely face incongruences in structures, processes and cultures across different internal contexts, while they simultaneously must accommodate external technological, competitive and political forces.

During the HealthOrg renewal initiative's second and fourth stages, the tension between established and renewed technology usage manifested as ambiguous choices between two digital platforms. Technology renewal unavoidably involves functional and operational overlaps. In this case, during the second stage, the new platform, which emphasized information sharing and standardization, had been implemented, was fully functional, and attracted some HealthOrg staff and departments. Meanwhile, the legacy platform continued to operate and cater to the idiosyncratic needs of most HealthOrg units and individuals. Thus, HealthOrg found itself using two functionally overlapping and partially incompatible platforms. During stage four, the tension resurfaced as two distinct use patterns within the daily operations at Headquarters. Our analysis of these two manifestations emphasizes how foundational legacy technologies contribute to deeply ingrained structures (Fichman & Kemerer, 1997) and processes that are difficult to change (Polites & Karahanna, 2012; Robey et al., 2002; Wessel et al., 2020), even in the face of strategic attempts to use new technology to shift organizational trajectories. To address this, managers of technology renewal initiatives must move beyond HealthOrg's approach and provide compelling incentives and support to transform legacy structures and processes to suit the new technologies.

Finally, the tension between deliberate and emergent renewal practices manifested during stages three and six of HealthOrg's renewal initiative. During stage three, the new platform's internal champions pushed hard to promote a full transition to meet explicitly stated objectives and the governmental mandate. Their attempts to form and execute a broad, deliberate implementation strategy (Nah & Delgado, 2006) and gain top management support (Martin & Huq, 2007) nevertheless clashed with the widespread perception that the new platform was simply a specialized technology dedicated to record-keeping. Such misperceptions flourished at the micro-level within most organizational units, illustrating the challenges involved in managing socio-psychological barriers (Orlikowski, 1992) and unanticipated events (McGann & Lyytinen, 2005) during complex digital transformations. In a similar vein, this tension's manifestation in stage six involved an emerging realization of the need to coordinate the management of information and technology, whereas the deliberate renewal practices included a continued separation of concerns and responsibilities between different departments. Together, these insights highlight the need for organizational actors to recognize the existence of and manage situated perceptions, ambiguous processes and emergent improvisations as they pursue technology renewal through rationally driven agendas and approaches.

Because we developed the three paradoxical tensions theoretically, based on insights from different research streams, they may also manifest outside the technology renewal context—such as in the implementation of new applications, transformation of digital infrastructures and management of enterprise system platforms. However, as our empirical analysis shows, HealthOrg's renewal initiative was particularly challenging because combinations of salient tensions related to all three tensions arose across the initiative's six stages (Table 4). Our analysis further revealed that all manifestations of the tensions had important impacts on HealthOrg's renewal trajectory. These findings support our claim that technology renewal is a distinct and paradoxical digital transformation process that is driven by three mutually constituent paradoxical tensions (Farjoun, 2010).

5.2 Organizational responses to paradoxical tensions

Paradox research (eg, Lewis, 2000; Lüsher & Lewis, 2008; Smith & Lewis, 2011) suggests that the effects of tensions depend on how organizations respond to them (Jarzabkowski & Lê, 2017), and that these responses may reinforce both positive and negative change cycles (Lewis, 2000; Smith & Lewis, 2011). For example, cognitive and behavioural forces tend to push for consistency (Cialdini, Trost, & Newsom, 1995), which can make organizational actors ignore a salient tension or engage to resolve it; similarly, organizational inertia in existing structures, processes and cultures (Gilbert, 2005; Gregory et al., 2015; Henderson & Clark, 1990) may cause actors to respond defensively and reluctantly, or challenge them to develop radical resolutions.

Furthermore, Smith and Lewis (2011) note that organizational actors may reject or avoid tensions, or accept and address them, by either integrating or splitting their opposite poles. Here, integrating entails deliberately accommodating both opposite poles, whereas in splitting, the actors actively choose between them (Smith & Lewis, 2011). To generalize the different responses during HealthOrg's technology renewal, we build on these concepts and also develop new ones. Table 5 summarizes the results of our theorization, generalizing the six organizational responses during HealthOrg's renewal initiative (Table 4) into four response types—integrating, splitting, pretending and avoiding.

In an integrating response, the organization accepts the tension it faces and addresses it by accommodating both poles (Smith & Lewis, 2011). During the first stage at HealthOrg, the decision to acquire a new ECM platform and simultaneously allow continued use of the legacy platform accommodated the conflicting requirements (ie, standardized vs customized approaches to information management). As such, management embraced a paradoxical situation, deliberately pursuing two parallel platform paths, while intending to gradually switch its operations to the new digital platform.

The splitting response is also resolution-oriented: the organization accepts a tension and addresses it by prioritizing one pole over the other. At HealthOrg, this response is illustrated in stage two, when the secretariat initiated a joint research project with the local university to develop knowledge about how to fully transition to the new digital platform. Although the response did not have an immediate effect, it was motivated by the recognition that having two distinct but overlapping digital platforms and user groups was unsustainable in the long term.

In the pretending response, the organization articulates a solution to a perceived tension without committing to consequential actions to realize that solution. As such, this response represents a form of window dressing. Stage three at HealthOrg offers an example of this response type: The board mandated use of the new digital platform, a decisive and likely productive response, but offered neither incentives nor sanctions that might impact staff behaviour and thereby enforce that mandate.

Finally, in the avoiding response, the organization accepts a tension without addressing it. That is, actors experience manifestations of tensions and may contemplate how to respond, but they take no actual actions to resolve the tension. At HealthOrg, this response type was illustrated in stage four, when the organization experienced a tension between established and renewed technology usage. While organizational actors engaged during this stage, for example, the secretariat strengthened the new platform's strategic position, none of its actions addressed the dominating tension.

As these conceptualizations show, technology renewal is a complex undertaking in which tensions at particular junctures present the involved actors with a variety of choice opportunities. It is the interplay between the tensions that actors experience and, in turn, the responses they enact that drives technology renewal by impacting trajectories and outcomes.

5.3 Impacts on technology renewal

Having detailed both how paradoxical tensions manifest in renewal initiatives and the various ways in which actors may respond to these tensions, we now examine how generative mechanisms that are based on underlying deep structures (Pentland, 1999; Van de Ven & Poole, 1995) shape renewal trajectories and outcomes. At a general level, the paradoxical tensions are the deep structures that, when triggered, activate a generative mechanism of on-going interplay between experienced tensions and organizational responses. This unfolding of a technological renewal initiative is a recursive pattern in which a paradoxical tension manifests as an experienced tension and a related response, which, in turn, impacts the renewal configuration and trajectory, which then triggers the manifestation of a paradoxical tension, and so on.

This recursive pattern is illustrated in the unfolding of HealthOrg's renewal initiative. During the first stage, the organization faced conflicting requirements for its information management practices, which expressed the tension between inner and outer contexts. The experienced tension was triggered by an internal investigation that revealed two critical issues: the serious problems in HealthOrg's information management practices, and a government mandate to standardize and improve information management across public organizations. HealthOrg responded to the experienced tension by investing in a new digital platform that would allow it to meet the government mandate, improve current practices and continue to serve idiosyncratic needs. However, while this integrating response acknowledged the tension and provided a short-term solution to a pressing problem, the response's fully transformative impact depended on a successful long-term switch to the new platform. Because this long-term strategy was not explicitly communicated to staff members, however, the response's actual impact was to make the organization increasingly dependent on two functionally overlapping and incompatible digital platforms. This triggered the manifestation of the tension between established and renewed technology usage: The legacy platform met the requirements of most employees and departments, while the new platform supported information creation and sharing across departments. This, in turn, produced a splitting response from the secretariat; it prioritized one pole (renewed technology usage) over the other (established technology usage) by investigating how it could realize a transition to the new digital platform. Over the subsequent stages, the technological renewal initiative continued to unfold through this recursive pattern, albeit with different tensions, responses and renewal impacts.

Interestingly, paradox studies have demonstrated how organizational responses to salient tensions can generate both virtuous and vicious cycles. Smith and Lewis (2011) argue that virtuous cycles require organizational actors to accept rather than reject the tensions they face, and to constructively address them by either integrating or splitting their opposite poles. In the HealthOrg case, we saw signs of such a virtuous cycle in the renewal initiative's early stages. However, from stage three onward, HealthOrg actors responded to the tensions they faced by pretending or avoiding responses; this eventually led to a failing initiative in which the organization continued to invest in and rely on two overlapping and partially incompatible digital platforms. As such, the HealthOrg case eventually came to represent a vicious cycle in which a persistent pattern of pretending and avoiding responses reinforced a trajectory towards renewal failure.

Drawing on Smith and Lewis (2011) and our HealthOrg findings, we suggest that, to reinforce a virtuous cycle and increase the likelihood of a successful technology renewal, managers must pursue a persistent pattern of integrating and splitting responses. In contrast, persistent patterns of pretending and avoiding will likely reinforce a vicious cycle and lead to renewal failure.