TRANSFERRING ROUTINES ACROSS MULTIPLE BOUNDARIES: A FLEXIBLE APPROACH

We currently know little about how transferring can be accomplished when sourceand target environments only have little in common. This chapter utilizes the case of EuroCo and AsiaCo to account for how a transfer of interrelated routines across multiple boundaries unfolds. A pragmatic and flexible approach to transferring, where coordinating actors attended to replication and adaptation as means rather than ends, is illuminated. Notably, coordinators split their work into smaller chunks by focusing on artifacts, people, and actions. As pressures to progress the transfer increased, they conceived of new ideas for performances and put the ideas to use along three trajectories focused on embedding, embodying, and enacting routines. Eventually, they blended performances from each trajectory back together into a new overarching notion of what was to be transferred. In elaborating on and discussing these findings, the chapter contributes to literature on routine transfer. Boundary conditions and avenues for future research are discussed.


INTRODUCTION
"Do you know how to do this?" The question was posed on the phone to one of the higher-level managers in EuroCo, a large food multinational with a home base in Europe. The manager's name appeared on the advertisement for a new position. I had called him on a hunch not to get the job, but to learn about a project mentioned in the work description. The project as well as the job, it turned out, concerned a "transfer." It was to be carried out together with another large food multinational, AsiaCo, in the home market of the latter. More specifically, the new hire would help export a quality control system, a set of production routines used in the raw milk supply chain of EuroCo in Europe, to AsiaCo and their supply chain in Asia. This was no small ambition, since the source-and target contexts differed in multiple ways. The organizational and industrial set-ups, for example, were poles apart. EuroCo, a cooperative, had for long been working closely with their raw milk suppliers. AsiaCo, a listed corporation, had mainly kept producers at arm's length. Complications could also be expected due to the large geographical span of the transfer and the differing languages and cultures involved. Hence, as he attended to my question ("do you know how to do this"), the EuroCo manager hesitated to spell out an unequivocal "yes." They believed they could do it, he clarified. After all, both organizations had committed to the transfer, evident from a signed project agreement. The idea was to set up a model of the system routines at one raw milk farm. Later, they would disseminate this model to other AsiaCo suppliers, although some changes would probably have to be made. Exactly how to progress (or "do this"), the manager alleged, "is what we will have to find out." Routines are essential for helping organizations reproduce knowledge and technologies across locations (Argote, 2012;Kostova, 1999;Winter & Szulanski, 2001). Yet, scholars have only recently started unpacking the full complexity involved as practices are targeted for transferring between highly diverse contexts (Bertels, Howard-Grenville, & Pek, 2016;D'Adderio, 2014D'Adderio, , 2017. At present, we still know little about how transferring can be accomplished when the source-and target environment only have little in common, such as when routines are subject to move across, rather than within, organizations (Friesl & Larty, 2013). This is unfortunate, since independent organizations are increasingly joining forces to export complex routines between locations across the world (Bouncken, Gast, Kraus, & Bogers, 2015;Dagnino, 2012;Luo, 2004). Thus, to better align transferring research with practice, there is a need to explore how routines can be recreated across multiple boundaries: organizational, geographical, as well as cultural.
Previous research allows for the anticipation that organizations aiming to transfer one or more routines are likely to encounter performative struggles, which denote a kind of confrontation-in-action between competing ordering systems or goals (Callon, 2007;D'Adderio, 2017;D'Adderio & Pollock, 2014). As part of performative struggles during transferring, some practitioners may gravitate toward mobilizing social and material features in support of routine replication, while others will try to promote routine adaptation (D'Adderio, 2014). The literature on transfer-as-replication, nevertheless, recommends organizations to suppress adaptation of routines until an exact copy of the source routines has been accomplished (Winter & Szulanski, 2001). At the same time, the literature recognizes that not all settings lend themselves to this approach (Szulanski & Winter, 2002;Winter, Szulanski, Ringov, & Jensen, 2012). Instead, then, organizations might enact a transfer-as-adaptation, where actors to a larger extent alter routines as they integrate them into a target environment (Bertels et al., 2016;Lazaric & Denis, 2005), or where actors redesign routines in a separate organizational process prior to replication (Gupta, Hoopes, & Knott, 2015). These differing alternatives illustrate that transfers and performative struggles may unfold in numerous ways, depending on circumstances (D'Adderio, 2014(D'Adderio, , p. 1348). More research is needed to unpack the plurality of possible transferring approaches and illuminate the conditions under which some performances, and struggles, are likely to take place (D'Adderio, 2017;Gupta et al., 2015;Winter et al., 2012).
Drawing on a process study of an inter-organizational routine transfer that took place between EuroCo and AsiaCo, the present chapter addresses both of the abovementioned needs: the need to account for the practice of routine transfer across multiple boundaries and the need to learn more about how specific circumstances may impact transferring. The case exemplifies transferring under extreme conditions, where multiple routines are to be recreated across fundamentally different settings. The access I gained to an inter-organizational team in charge of coordinating the transfer allowed me to trace processual dynamics over time. This enabled answering the following research question, with particular attention to the role of coordinators: How does transferring of routines across organizational structures, cultures, and geographical locations unfold, and what are the performative struggles at play?
The case analysis illuminated a flexible, creative and pragmatic approach to transferring. Coordinators abandoned their initial focus on routine replication due to a number of contextual constraints, such as key stakeholders' differing interests and priorities. As coordinators struggled to proceed, they enacted and further progressed a transfer-as-adaptation as they learned more about which routines were afforded by the target environment. Notably, coordinators first split their work into smaller chunks by focusing on artifacts, people and actions. As pressures to progress the transfer increased, they generated new ideas for performances and put these ideas to use along performative trajectories focused on embedding routines (in artifacts), embodying routines (in people) and enacting routines (in practice). Eventually, the coordinators blended performances focused on artifacts, people, and actions back together into a new overarching idea of what was to be transferred. Instead of emphasizing a system for control, they shifted their attention to creating structures for giving advice. The further discussion of these findings serve to expand and develop ideas related to performative struggles and selective performance (D'Adderio, 2014(D'Adderio, , 2017 and the changing nature of objectives (Dittrich & Seidl, 2018) during transferring.
The rest of the chapter proceeds as follows. First, the literature mobilized by the study is outlined. Second, the empirical case and methodological approach is introduced. Third, findings are presented in the form of narratives and analytical summaries. Fourth, contributions and questions for future research are discussed.

LITERATURE
Routines, defined as repetitive, recognizable patterns of interdependent actions carried out by multiple actors (Feldman & Pentland, 2003), can help organizations create value in ways that may be hard to copy (Grant, 1991;Teece, Pisano, & Shuen, 1997;Vogel & Hansen, 2011). Exactly for this reason, however, organizations may try to duplicate routines across locations to further benefit from their associated returns (Argote, 2012;D'Adderio, 2014;Winter & Szulanski, 2001). It is not strictly possible to separate routines from their original context, due to their embedded, situated, and socio-material nature (Howard-Grenville, 2005;Kremser & Schreyögg, 2016;Orlikowski, 2007). Yet, new routines can be created in a target context informed by routines in a source context, which we typically understand as a transfer.
A routine dynamics lens has proved to be useful for deepening our understanding of transferring because it insists on seeing routines as more than entities, or things (Feldman, Pentland, D'Adderio, & Lazaric, 2016). As opposed to considering a transfer as accomplished when routines' associated knowledge and artifacts have been transmitted from one setting to another, a routine dynamics approach pays explicit attention to performances, which involves actually assembling things and relationships in practice (D'Adderio, 2017). Transferring is thus, from a routine dynamics perspective, understood as a continuously emergent achievement, produced by situated combinations of actors, artifacts, and understandings. Transferring is bound to involve some degree of transformation, and the degree to which the transferred routines resemble a source template can vary from case to case (Friesl & Larty, 2013;Gherardi & Nicolini, 2000).
So far, the literature has predominantly illuminated routine transferring as it is performed by single organizations who integrate routines modeled from an external source (Bertels et al., 2016;Canato, Ravasi, & Phillips, 2013;Lazaric & Denis, 2005) or transferring between organizational units that share the same parent or owner (Ansari, Reinecke, & Spaan, 2014;D'Adderio, 2014;Gupta et al., 2015). Progressively, routines are traded between independent organizations (Bouncken et al., 2015;Dagnino, 2012;Luo, 2004). Routines are thus subject to transferring attempts spanning boundaries of ownership, nationalities, and cultures. We currently lack clarity regarding how such settings shape the prospects and process of transferring routines. This is unfortunate, since inter-and intra-organizational transferring processes are likely to be distinct (Easterby-Smith, Lyles, & Tsang, 2008). Despite this void in the literature, recent work has highlighted some key phenomena that may play out differently in different transferring setups: performative struggles and the patterned responses to such struggles, in particular the responses of coordinating actors.

Performative Struggles and Responses during Transferring
In her study of an outstanding organization attempting to copy an advanced computer server from one geographical location to another, D'Adderio (2014) directed attention to how organizations "enact routinized patterns that selectively perform one goal over another, both at specific points in time, and over time" (p. 1345). In essence, she unpacked how the transfer was influenced by the simultaneous presence of, and actors selective responses to, two contrasting organizational goals: one pushing for routine replication and the other pushing for routine adaptation. The presence of such competing goals, or programs (Callon, 2007), resulted in performative struggles that shaped the process and outcomes of transferring (D'Adderio, 2017;D'Adderio & Pollock, 2014).
The literature on transferring highlights two main solutions to, or rather enactments of, performative struggles. In the case studied by D'Adderio (2014), performances supporting replication were first foregrounded while performances supporting adaptation were suppressed. Post-transfer, the logic gradually shifted so that performances favoring adaptation dominated (D'Adderio, 2014). Due to the predominant idea of first prioritizing imitation of routines, nevertheless, this accomplishment can be seen as a transfer-as-replication. Another response to performative struggles entails more flexibly integrating routines into a target environment, to fit the new context (Ansari et al., 2014;Bertels et al., 2014;Canato et al., 2013). In some cases, this integration may be achieved by first engaging in a separate redesign process and later spreading the redesigned routines across different organizational units and locations (Gupta et al., 2015). Due to the predominant logic of first prioritizing routine adjustment in these cases, they exemplify transfer-as-adaptation.
There is a need for more research on when and why organizations may enact a transfer-as-replication versus a transfer-as-adaptation (D'Adderio, 2014; Gupta et al., 2015). Based on current literature on transferring, organizations seem to find replication well suited for gaining advantages relating to economies of scale and for achieving a high outcome predictability of operations across locations (Szulanski & Winter, 2002;Winter et al., 2012;Winter & Szulanski, 2001). This rationale rests on the insight that, due to the complexity of routines, there is causal ambiguity regarding how an established routine creates value (Axelrod & Cohen, 1999). To avoid missing out on its key components, copying all the elements associated with the routine and bringing them together (replicating) in the new context is recommended before adjustments are made (Ansari et al., 2010;Terwiesch & Xu, 2004;Winter et al., 2012). As D'Adderio (2014) showed, this does not mean that performances supporting adaptation are completely eliminated. Instead, they are largely, and temporarily, suppressed until a working model has been established.
The replication literature recognizes that performing a transfer-as-replication is inherently difficult (Szulanski & Winter, 2002). Bertels et al. (2016) observed such difficulties in their study of how one organization (OilCo) attempted to integrate a routine used by industry peers. For OilCo, the routine subject to copying was a cultural misfit, which led to unintended workarounds and hindered performances. In the words of the authors: integrating a routine can lead to the repeated performance of more complex patterns of action than those originally espoused, because additional work is done by employees as they try to accommodate the misalignment of their culture with the espoused routine. (Bertels et al., 2016, p. 587) Other cases, too, support the idea that some contexts and conditions may invite adaptation, rather than replication. When the practice of Six Sigma was implemented in 3M, for example, the routine to be integrated was changed along with the cultural repertoires of the people performing it (Canato et al., 2013), thus helping to avoid some of the inefficiencies experienced by OilCo. A case where a management practice was adapted within a multinational led Ansari et al. (2014) to argue that a template logic may prevent, rather than ease, diffusion in a target environment. In their study of a failed transfer-as-adaptation, Gupta et al. (2015) suggested that an initial redesign phase is needed when a routine is to be replicated to a context vitally different from the source.

The Role of Coordinators in Responding To and Shaping Performative Struggles
The notion of performative struggles, and the potential plurality of how a transfer can be performed, begs the question of what, or who, decides whether one goal or program will be foregrounded and another suppressed during transferring. D'Adderio (2017) illustrates that combinations of situations, actors, and artifacts afford different responses to programs at different points in time. This situational agencement, rather than simply agency, may eventually develop into a pattern where one performative program, and goal, is favored over another (Callon, 2007;D'Adderio, 2017).
Notwithstanding the importance of paying attention to agencement, past studies illustrate that some actors may play a particularly prominent role in responding to performative struggles during transferring, namely coordinators. Work focused on knowledge-and technology transfer, for instance, highlights how coordinators may act as gatekeepers, brokers, trainers, or testers and thus be instrumental for progressing a transfer over time (Battistella, De Toni, & Pillon, 2016;Van Burg, Berends, & Raaij, 2014;Howells, 2006). In Lazaric and Denis (2005) study of an organization that changed routines to comply with an ISO norm, coordinators were important for motivating participation and facilitating communication. In the studies by D'Adderio (2014) and Bertels et al. (2016), coordinating actors mobilized artifacts and communities to influence routines toward alignment or improvement.
As they attend to contrasting pressures during a transfer, moreover, coordinators might end up revising ideas about what the transfer should accomplish. In Dittrich and Seidls (2018) study of emerging intentionality in CellCo, actors gradually updated their understanding of what an assembly routine should achieve as they shifted attention between the initial purpose of the routine (ad hoc production), the means to pursue this purpose (assembling whenever there was a need) and emerging new goals, referred to as ends-in-view (preserving lab space during the day time). During this process, new means (working nights and weekends) came to be foregrounded while yet other emerging ends-in-view (enabling workers to autonomously organize their work) further influenced which actions were taken. Over time, the understanding of what the routine should accomplish (production outside of normal working hours) was revised (Dittrich & Seidl, 2018).
Although it does not concern a transfer as such, Dittrich and Seidls (2018) study is important because it highlights the possibility that initial contrasting goals of a transfer, such as replication or adaptation, might not necessarily remain in conflict as a transfer unfolds. Rather than being foregrounded and suppressed at different points in time, such goals may come to be reinvented or outcompeted by other performative programs and goals, as coordinators make sense of how to proceed. The telecom giant Eriksson's transition to an agile development routine supports this idea. In this case, exemplifying a deliberate routine change, the initial goal was related to efficiency, but managers in Ericsson discovered a new purpose and performative trajectory along the way, namely that "the new mode of operation could also further small-scale innovation" (Lindkvist, Bengtsson, Svensson, & Wahlstedt, 2016, p. 587).
Despite these pointers from the literatures on transferring and routine dynamics, no studies seem to have unpacked the nature of performative struggles when routines are transferred across organizations and highly dissimilar contexts. It remains to be discovered how coordinators and other actors influence which goals are attended to, and the means through which they are accomplished, in such cases.

CASE AND METHOD
To unpack the dynamics of routine transfer across highly dissimilar contexts, a process study (Langley, 1999;Locke, 2001) is utilized. The study started in 2013, broadly focused on tracking the collaborative activities of two multinationals. Processual approaches are particularly suited to uncover the interplay of the actions and motives of agents and the evolving contexts surrounding them (Van Burg et al., 2014;Berends & Deken, 2019;Giddens, 1984). The findings relevant for this chapter came about after zooming in on how coordinators from the two organizations progressed a specific transfer project.
As revealed by the vignette introducing this chapter, the two organizations of interest (EuroCo and AsiaCo) had agreed to transfer a quality control system from EuroCo in Europe to AsiaCo in Asia. The organizations were similar in a few ways: they were among the world's top producers within their trade; they operated across borders and were thus present in partly overlapping markets. Importantly, they originated from, and had their home bases in, different parts of the world. Moreover, as further elaborated in the findings, they differed considerably in terms of organizational structure and culture. Due to these profound differences compared to other known examples, the transfer between EuroCo and AsiaCo can be seen as informative for illuminating transferring under extreme conditions (Flyvbjerg, 2006).

How the Data Were Collected and Analyzed
The main methods of data collection consisted of ethnographic procedures such as non-participant observation, shadowing, informal interviews, and documentary reviews (McDonald, 2005;O'Reilly, 2012). Data collection took place in Asia and Europe during two years, starting in the fall of 2013. For example, I visited a number of production sites in the milieu of AsiaCo where I talked informally with shop floor employees as well as managers. I visited similar sites of EuroCo in Europe together with the inter-organizational team of coordinators. I observed a large-scale training of inspectors in Asia and interviewed trainers and coordinators about other trainings. I spent time in the office with coordinators and participated in their meetings. When I did not personally observe important events, I inquired about them retrospectively through calls or email. I followed quality inspectors in AsiaCo at work and interviewed them informally to understand their current practices. At one point, it was decided to send a practitioner from EuroCo to work with producers in Asia for six months. I captured the process leading up to this decision in my interviews and informally interviewed the practitioner after his return. The latter data were compared with data captured during informal interviews with production site managers in Asia. I further triangulated with data from other sources (Jick, 1979;Strauss, 1987) such as reports, meeting minutes, annual reports, and copies of internal presentations. I also conducted independent searches for news articles and industry briefings.
To guide my analysis I developed a visual data structure (Gioia, Corley, & Hamilton, 2013) by means of open coding (Corbin & Strauss, 1990). For this chapter, I went back to the raw data to zoom in on the dimensions and themes of interest. My initial analysis had captured the flexible nature of how the routine transfer was enacted over time and identified three performative trajectories focused on artifacts (embedding), people (embodying), and actions (enacting). Moreover, I had noted that the learning taking place during these differing trajectories led to a rethinking of the system subject to transfer. My initial analysis did not elaborate on why the specific performances were chosen by coordinators or the dynamics through which adjustments were made. I therefore re-read field notes and re-examined timelines. During this process, I went back to the literature on transferring, which provided the language of performative struggles (Callon, 2007;D'Adderio, 2017) and ends-in-view (Dittrich & Seidl, 2018) to talk about some of the findings. During the final analytical stage, I drew up an empirical model as well as a simplified theoretical model of the dynamics at play. As for the purpose of presenting the findings, I went back to field notes in search for representative observations and quotes. To ease readability, I organized the observations and quotes into two narratives.

FINDINGS
To effectively communicate the findings I provide two elaborate narratives followed by analytical summaries. Together, the narratives illustrate transferring as it unfolded across highly dissimilar contexts and the nature of the performative struggles at play. Narrative 1 reflects initial conditions and struggles experienced by coordinators. Narrative 2, to a larger extent, reflects responses to the struggles over time.

Narrative 1: Initial Conditions and Performative Struggles
It was already several months since I had first called up EuroCo to learn about their transferring project with AsiaCo. "We will have to find out" had the upper level manager answered when I asked if they knew how to proceed. Now, in a car moving across the Asian country side, a newly hired EuroCo manager (the one who got the job from the advertisement) was getting close to finding out that their initial plans had to be revised. Exactly what to do instead was yet to be discovered.
After a long day at work, the drive back had started out in silence, except for the presence of two tapping sounds. The first sound came from small rocks projecting from underneath the wheels of our car; gravel roads are hard to avoid when paying visits to raw milk farms. The other sound was produced by the EuroCo manager herself, as she typed up e-mails on her tablet. She paused only for phone calls or, if probed, to clarify what she was thinking or doing. "Our plans are always changing. But that's just how things work here, I am getting more used to it now," she explained after a while. "You are a fast learner, very good," chuckled her new colleague, a manager in AsiaCo. He was in the front passenger seat, next to the driver. Turning to look at us, he added, seemingly amused by the idea: "You know, she's [becoming] more and more like us." The two managers in the car had established a good rapport by now. While they technically worked for different employers, they were both part of an interorganizational center of expertise. During the past decade or so, many European producers (such as EuroCo) had been interested in the Asian market, due to its size and its growing segment of consumers willing to pay a premium for foreign brands. Concurrently, Asian producers (such as AsiaCo) had paid increasing attention to how foreign manufacturers organized production in the hope of achieving similar levels of quality and efficiency. Against this background, EuroCo and AsiaCo had co-founded the center of expertise in Asia as part of a series of mutually beneficial exchanges of know-how. EuroCo wanted to learn about marketing their products in Asia while AsiaCo was interested in European management practices. The exchanges were negotiated on a project-by-project basis. Among the largest so far was a transfer, coordinated by the center.
According to the formal project agreement, the plan of the transfer was to copy a quality control system and implement it in the home environment of AsiaCo. The system was used by EuroCo and their suppliers of raw milk, which was the main ingredient in dairy products offered to consumers. The system consisted of multiple and interrelated routines, whereof some routines governed how other routines were performed. For instance, there were routines of quality inspection. Other routines accomplished cleaning of equipment, milking of the cows or trimming of the cows' hoofs. The most tangible aspect of the control system was its written procedures, designed to regulate the handling of animals as well as machines at production sites (large-scale farms). Additionally, the system was associated with more or less tacit knowledge about why to carry out certain activities, and how.
Transferring the quality control system required spanning several boundaries, such as overcoming large geographical distances and differing working languages.
The organizational forms of EuroCo and AsiaCo, and the industrial systems in which they were embedded, differed as well. EuroCo, a cooperative, had for many decades worked closely with their suppliers, who effectively owned their business. AsiaCo, a listed corporation, had been dealing with suppliers mainly through contractual negotiations and arm's length ownership. While EuroCo had a long history of slow and steady development, AsiaCo had a short history of rapid entrepreneurial growth. As was soon discovered by the coordinating managers in the center of expertise, the working cultures were different. "Once they start work, they don't want to change anymore, but here, we can go to the East today, and tomorrow we can go West," said a coordinator from AsiaCo. This perception was largely echoed by people from EuroCo: They signed the contract, and it was stated very precisely there -from our side -how we saw it. But to them, a deal is a deal only until you make a new deal. It is just a frame. It does not mean that we cannot do other things.
The EuroCo manager admitted to struggling with understanding how to mobilize people in Asia. "It's difficult," she said, as we were driving further away from one of AsiaCo's best performing supply sites; a newly built farm with glossy tiled floors and decorative plants in the foyer. Not a cow had been seen during the first part of our visit. That is, not before we caught a glimpse of the large ranch facilities through wall-sized windows, on our way back from the meeting room where the site manager had welcomed us. He sat down with us in formal office wear. Shop floor employees passed by, wearing coveralls that appeared to be brand new. As opposed to other and smaller production sites we had visited, the manager said they were able to attract younger workers and retain them. Partly, he believed, because of the farm's investment in modern equipment and their ambitious plans for the future.
It was on this farm EuroCo and AsiaCo had planned to model the new routines associated with the control system. Although they anticipated a need to adjust the routines to the new context, the idea was to start out by mimicking a small-scale version of the control system on Asian ground. Employees in AsiaCo were to be trained to perform the quality control part. After the system had been scaled up and disseminated to other raw milk production sites, controllers would help maintain it by carrying out regular inspections. Completing the training in Asia, however, proved to be more difficult than anticipated. For instance, they struggled to understand which individuals, among the large stock of AsiaCo employees, best resembled the type of quality controllers working in Europe. While one dairy production training for Asian workers had been achieved as part of a different project, the center had so far not managed to organize the first training of future quality control inspectors in connection with the system transfer. In retrospect, they attributed some of these mobilization struggles to incentive problems. "People in AsiaCo are busy. They have many other things they must do" explained a manager from AsiaCo.
As we stood at the targeted model farm and peeked into the production from behind the large glass windows, other challenges became apparent as well. In essence, the employees on the farm were already busy improving, implementing the advice of other foreign experts. In consequence, little time could be spared to consider the quality control system in the short term. This situation was particularly challenging for coordinators from AsiaCo. From the outset, they had perceived it to be critical to get hold of tangible versions of the routines so that, as one of them said, they could "understand everything." Translating the written standard operating procedures (SOPs) associated with quality controls -both in terms of replacing the language and in terms of adjusting to local conditionswould help fulfill this purpose. Notably, AsiaCo could not do the work alone. By the time we visited the targeted model farm together, the AsiaCo manager had been talking with them for months about finding time to discuss the new procedures. "It is difficult (…) They are such an important stakeholder for us, but they have so many other projects going on. I am actually not sure how to motivate them," said the EuroCo manager, back in the car.
In a later conversation, the AsiaCo manager confirmed the situation: On the one hand, AsiaCo clearly expected all their suppliers to improve on quality. On the other hand, due to farms' high bargaining power and their relatively loose ties to AsiaCo, it was deemed unproductive to directly instruct them. Still, the manager said he often found himself explaining to his superiors why patience was needed. For example, opinions differed regarding the importance of waiting for the so-called gap analysis. European experts, hired by the center, were preparing to tour several of AsiaCo's production sites to conduct this analysis. Some EuroCo experts had already visited other sites and produced short reports with recommendations to farm managers which, according to the experts, had largely been neglected. Still lacking at the time, nonetheless, was a more elaborate report listing the exact differences -the gaps -between the current setup in Asia and the setup of the control system in Europe. While many people in EuroCo felt it was important to wait for this report and expressed frustration about the lack of available data from the farms, people AsiaCo appeared to expect a more fast-paced implementation. A slow plan A, it seemed, meant it was time for plan B. "They [EuroCo] want it to be steady and procedural, while we [AsiaCo] want speed and want to get the effect instantly." Later, in another car, we were making our way from the airport to the main office facilities of the center of expertise. The car was progressing steadily on paved roads, and the EuroCo manager had taken a break from her multitasking. By this point, she had grown impatient too. Well out of the vehicle and into the elevator, she mentioned an idea she had begun considering, regarding how to proceed. It seemed essential now, she expounded, to "create a win."

Analytical Summary of Narrative 1
This narrative illustrates the challenging conditions of the transfer between EuroCo and AsiaCo and how coordinators worked on several fronts to make it happen. Notable in this respect, is how they grappled with complexity by splitting the transferring work into smaller tasks, focused on artifacts (creating the Asian versions of the SOPs), people (training quality controllers), and actions (analyzing and influencing actual performances at supply sites, as part of the gap analysis). In my analysis, I coded these efforts as concerned with embedding routines in artifacts, embodying routines in people and enacting routines in practice. While all aspects (artifacts, people and actions) will always come together during actual performances of routines, it seemed that the complexity of the system to be transferred led coordinators to conceptually bracket routines as they organized their work.
On all fronts, coordinators in EuroCo and AsiaCo experienced difficulties with progressing with their initial replication logic of setting up a mini-model of the system. This was due to a number of constraints encountered in the target context, such as the targeted model farm's limited availability, the difficulty of recruiting quality control trainees and the tendency of farm managers to overlook advice provided via formal reports. EuroCo, more than AsiaCo, seemed to favor sticking to the original project agreement, although the EuroCo coordinator directly situated in the target environment (the recent hire outplaced in Asia) adjusted to the local culture and began to mimic the impatience expressed by AsiaCo managers. Notable in this respect, is that coordinators lacked managerial control over the complex context they were navigating -a natural consequence of working across organizational boundaries. As a result, most stakeholders (such as the targeted model farm) had to be convinced, more than simply instructed. On top of this, data that would have allowed more direct comparisons between the source-and target context were not always available, which seemed to further hamper the pace of the progress.
Rather than keeping on grappling with performative struggles caused by contrasting goals of replication and adaption, coordinators showed signs of shifting their attention to the increasingly pressing need (or end-in-view) of demonstrating any kind of progress. Nonetheless, other contrasting programs, such as alternative improvement projects attended to by the targeted model farm and competing commitments of trainees, hindered advancement. As evident from the idea mentioned by the EuroCo coordinator, these difficulties eventually inspired a more creative and flexible approach. This approach is further illustrated in Narrative 2, and in Fig. 1, which elaborates how the transfer unfolded over time.

Narrative 2: Responses to Performative Struggles
"It was a definitive low-point" said the EuroCo manager in retrospect, about the time when she travelled back to Europe for the end-of-year holiday. The gap analysis was in, with a preliminary recommendation. Essentially, it had concluded what they already suspected: Due to the fundamentally different industrial and organizational conditions in Asia, compared to in Europe, the initial ambitions for how to transfer the control system were not realistic, at least not for the time being. For starters, the variability regarding the modernity and size of raw milk supply sites in Asia would make it difficult to maintain one system to fit them all. Due to a current undersupply of raw milk in Asia, moreover, most suppliers were focused on maximizing quantity instead of quality. Some non-negotiable quality standards still applied, but pressuring suppliers too much on raising the bar could lead them to walk out of contracts and instead start delivering to AsiaCo's competitors. To overcome these difficulties, the report suggested to focus only on a few supply sites, for instance the targeted model farm, which was owned by AsiaCo.
The EuroCo manager was not thrilled, as she felt they could "make the greatest difference" on many of the other supply sites. Seemingly blocking the possibility of creating wide-scale progress, though, was the difficulty of establishing a dialogue with farmers about how things could be improved. One gap analysis expert, for instance, described how a farm manager "just laughed" during a demonstration of how to properly trim the cow's hooves. For the expert, taking the extra time made sense out of consideration for animal welfare, which was in the end strongly related to raw milk quality: If cows could walk properly, he told me, they would get easier hold of their food and thus also eat better. If they ate better, they would produce milk of a higher quality. In fact, the quantity would also likely go up, due to their improved nutrition. From the manager's perspective, however, incorporating the new hoof trimming routine seemed less productive, simply "too much work." The EuroCo manager extended her end-of-year break to make time for discussions with her superiors back in Europe. It was time to consider seriously the idea she had mentioned before. When she returned to Asia, it was with the following suggestion in mind: to transfer a practitioner from Europe and make him work closely with a number of supply sites for six months. This, she imagined, would help creating actual results in places where it was really needed. Moreover, it could help demonstrate progress (a "win") and thus secure continued managerial buy-in to the transferring project. Decision makers in AsiaCo approved of the idea, and the practitioner started packing for Asia before the project contract had been formally signed. "It was not typical of us to act before all the formalities were in place, but we had started to allow ourselves to be more flexible" said the EuroCo manager in retrospect.
Meanwhile, regarding the struggles with getting the targeted model farm involved in translating SOPs, coordinators from AsiaCo gradually found a new way forward. This time they shifted the focus to translating procedures describing good farm practices, rather than focusing on control manuals for assessing those practices. After once again finding it difficult to schedule time with the targeted model farm, coordinators eventually progressed by involving experts from local universities in decisions about which SOPs to keep, omit, or alter. In collaboration with experts from EuroCo, a workshop was offered to the targeted model farm, "to show how we could help them create value." Because one of the decision makers at the site had a particular interest in animal health, the workshop activities focused on routines related to this topic. The workshop was considered a success, but it remained difficult to move forward with implementing the SOPs at the farm. Coordinators from AsiaCo proceeded anyway, by printing some of the translated SOPs as booklets. The plan was to distribute them to a larger group of production sites, including the targeted model farm.
Right after the practitioner arrived in Asia, coordinators in AsiaCo and EuroCo finally identified and targeted a set of employees from AsiaCo and conducted a quality control training. As the trainees went back to their offices, however, it turned out they would hardly get the opportunity to apply their new skills, since they rarely visited production sites. "We have actually spent quite some time training the wrong people," the EuroCo manager told me, looking back. She felt they needed to get hold of a new group of people for a new round of training. These people, she said, had to be comfortable with "putting on their dirty boots" instead of "walking around in the office with their high heels on." The center eventually recruited new people, from a different department in AsiaCo. Several changes were then made to the training content over time. For instance, some of the general classroom-based teaching was substituted with more specific instruction at production sites. This was deemed necessary to bring the trainees, who were still mainly recruited from office based jobs, closer to practice. To motivate participation, diplomas were designed and offered to participants at the end of the training. An exam was developed to certify that learning had taken place. "It makes it easier to document our progress. Plus, the diplomas are appreciated by the trainees. They are very motivated by these things," said the EuroCo coordinator. An AsiaCo employee explained that the diplomas symbolized qualification advancement, which would increase the status of quality controllers and help sustain their commitment.
During his stay in Asia, the practitioner from Europe put to use effective ways of communicating with farm sites. The formal report he wrote reflected that the purpose of his work was still to improve raw milk quality, but his focus as he spoke with raw milk farmers was more on efficiency and opportunity recognition. For example, informed by what one of the previous experts had passed on to him, he framed the need for hoof trimming as related to "money lying around on the farms, ready to be picked up." By the end of his stay, a systematic change of hoof trimming practices was documented and celebrated as a success. Other EuroCo experts changed their communication, too: At first we focused more on what was wrong, but we discovered that it was hard to get people at the supply sites to listen. We then changed our communication; "if you do it differently, it can be valuable for you." That helped a lot.
Coordinators in AsiaCo said they had been aware of the communication challenge out on the production sites for quite some time. One AsiaCo coordinator expressed that it was "easier for foreigners" to instruct farm site managers about what to do, than it would be for the locals trained to perform quality controls. Another AsiaCo coordinator had from an early stage started thinking about whether AsiaCo employees should act more as advisors to the farms, rather than putting forward demands and performing controls. "We do not have a separate service center, like the farmers have access to in your setting," he said to the EuroCo manager during a meeting. "I want to talk about if we should make such a service center." Eventually, the learning outcomes from the practitioner's stay were incorporated more widely, along with the new ideas about what roles trainees should take. During the latest version of the training, much emphasis was put on how to actually talk with farm managers and disseminate knowledge about routines. "What do you say when you see this," asked the trainer as he displayed photos of less-than-ideal production practices, probing the trainees to reflect on how to frame their messages to effectively motivate changes.
Eventually, coordinators started talking about putting in place an "advisory system" in the Asian context. The use of the term advisory system could be traced back to the written gap analysis. The analysis, in turn, was based on several informal talks with coordinators, in addition to farm site observations. When I asked coordinators about which point in time they had recognized the need to change, I got conflicting answers. "We have realized everything we planned for," said a higher-level AsiaCo manager. She pointed to the fact that the transfer project had received praise from top managers in AsiaCo as well as in EuroCo. The center, by then, enjoyed a good standing. The EuroCo manager more clearly articulated that "we decided at some point to focus on [transferring] an advisory system, rather than a control system." Another coordinator from AsiaCo put it as follows, during a discussion of what they were actually doing in the transfer project: "We are creating an Asian version. It cannot be the same." Analytical Summary of Narrative 2 Narrative 2 illustrates how, over time, coordinators from AsiaCo and EuroCo progressed their work as they adjusted to a more flexible approach to their initial plan of copying the control system. Coordinators invented shortcuts (such as involving university experts rather than people at the targeted model farm) workarounds (such as spending extended time at the farm production sites) and revised performances (such as the redesigned trainings) in response to the struggles they experienced. Because they lacked the authority to directly instruct important stakeholders, and because they had started out with a general awareness that some adjustments might have to be made, they seemed to transition quite easily into generating new ideas for how to proceed. The coordinator displaying the highest level of frustration with the resistance they faced was the EuroCo manager. As the pressure to demonstrate progress increased, she also championed the most radical idea (of transferring a practitioner to Asia).
As a result of their creative efforts, coordinators became more knowledgeable about the goals and programs that conflicted with their initial aims. They then opportunistically revised their approach informed by these goals and programs. For instance, the quantity-based aims and agendas of supply site managers were taken into account when communicating about new routines. Moreover, a separate workshop was designed to fit the interest of a decision maker at the targeted model farm. While the performative trajectories focused on artifacts, people, and actions prevailed (see Fig. 1), it is notable that aspects from each trajectory were increasingly blended back together as the transferring progressed. For example, the performance trajectory focused on embedding routines in artifacts (the SOPs) came to include embodying routines in people (the local university experts) and enacting routines in practice (the workshop at the targeted model farm). Similarly, the trajectory focused on embodying routines in people (through the training) came to include artifacts (such as diplomas and course materials) and enacting routines in practice (by moving the training from the classroom to the farms). The trajectory focused on enacting routines in practice (such as demonstrating hoof trimming during visits to raw milk production sites) produced artifacts (the written gap analysis and the practitioner's report) and embodied routines in people (by working closely with the farm managers, and other workers). In turn, coordinators fused back communication techniques, developed as they learned what worked to motivate farmers, into the training of AsiaCo employees.
Eventually, a new way of understanding and talking about the transfer (as concerning an advisory system, rather than a control system) materialized. This revised understanding seemed to more accurately reflect what coordinators were about to achieve, namely a kind of transfer-as-adaptation instead of the initial idea to set up a replication of the European system. It is notable that coordinators to a varying degree recognized the transition. Their reactions suggest that, even though they started out with a replication logic, enacting a replication had not been a salient end in itself, but rather a temporary idea for how to proceed. As they gained a deeper understanding of the routines afforded by the target context, coordinators revised this idea and enacted a transfer-as-adaptation, which proved to be a more effective means for achieving their increasingly pressing end-in-view: making progress.

DISCUSSION
The transfer literature has called for explorations of "how the replication dilemma unfolds for different organizations in less and differently demanding settings" (D'Adderio, 2014(D'Adderio, , p. 1348 as well as how alternatives to a replication approach may look like (Gupta et al., 2015). Based on its account of how transferring can be progressed across fundamentally different settings, this chapter offers several important extensions to the current knowledge base relevant for these questions.
First, the case of EuroCo and AsiaCo illuminates that actors involved in a transfer may start out with a replication logic, as a tentative idea of how to proceed, but then gradually shift to an adaptation logic as they learn about the constraints and opportunities in the target setting. As illustrated in Fig. 1, the initial project goal (point A) informed a preliminary plan based on a transferas-replication logic (point B). In practice, coordinators used the plan as an idea rather than as a template, driven by the emerging end-in-view of progressing the transfer (point C). Initial transferring performances (point D1, 2, 3) were revised and the replication logic further abandoned as coordinators learned about clashes with stakeholders' interests and other key differences between the source-and target context (point E). During the process of making creative adjustments (point F1, 2, 3), coordinators became further informed about stakeholders' interests, contextual constraints, as well as opportunities for revisions (point G). This led to a gradual revision of the purpose of the transferring (point H) which in turn informed further revisions (point I). This pragmatic approach contrasts other known approaches in the literature. For instance, D'Adderio (2014) showed how a transfer-as-replication can be achieved by suppressing pressures to adapt routines and foreground replication until a working model has been established in the target context. In the case of EuroCo and AsiaCo, adaptation was not suppressed, but flexibly used as a substitute when replication did not work. Gupta et al. (2015), in contrast, showed how actors redesigned routines in a separate process before replicating them. In the case of EuroCo and AsiaCo, coordinators did not know before the transfer which adaptations were needed. Instead, they learned about this, and invented new versions of the routines, during the process of performing the transfer itself. In contrast to Bertels et al. (2016), where adaptations appeared to be perceived as less-than-ideal deviations from an initial plan to replicate, actors in EuroCo and AsiaCo approached adaptations as necessary means to their end-in-view of actually achieving progress. This insight is in line with the idea that when contexts differ profoundly, a template logic might hinder diffusion (Ansari et al., 2014;Gupta et al., 2015).
Second, the case expands the current notion of how, and which, performative struggles might influence transferring of routines. In the study of D'Adderio (2014D'Adderio ( , 2017, the main performative struggles seemed to be caused by contrasting pressures to replicate versus adapt routines. Coordinators in EuroCo and AsiaCo dropped the replication logic at an early stage of the transfer and proceeded by pragmatically revising their ideas of what to transfer, and how. More prominent in their case, it appeared, were the contrasting pressures between goals and programs attended to by other important stakeholders (such as the improvement projects of the targeted model farm) and the transfer project's ambition of influencing current routines. This underscores how replication or adaptation may not always influence a transfer as contrasting goals (D'Adderio, 2014) but instead be flexibly mobilized in response to other performative struggles. More practically, the case suggests how paying attention to the goals of multiple stakeholders and adjusting the transfer to fit those goals may help coordinators in overcoming incentive problems. Other cases have highlighted paying attention to incentives as essential for succeeding with a transfer (Gupta et al., 2015). Adding to the notion of selective performance during performative struggles (D'Adderio, 2014), the case of EuroCo and AsiaCo exemplifies the creative aspects of such performance when actors perform a transfer-as-adaptation.
Third, bridging insights from literature on routine dynamics (Dittrich & Seidl, 2018) with transferring work, the case illuminates the interplay between goals, means and ends-in-view as a transfer unfolds over time. In essence, the case of EuroCo and AsiaCo suggests that actors coordinating a transfer may first conceive of preliminary means (here: replicating) in order to achieve a goal of recreating routines in a target setting. In response to enacting these preliminary means, new and unanticipated outcomes may be created (here: experiences of stakeholders' lack of interest and commitment) that coordinators then respond to by inventing yet other means (here: adapting routines during transferring), informed by increasingly prominent and emerging ends-in-view (here: achieving progress). The coordinators' reformulation of what was actually subject to transferring (an advisory system, rather than a control system) directs attention to what Dittrich and Seidl (2018) described as emerging intentionality: that goals and intentions might be produced during enactment rather than be specified a priori and then enacted. In essence, this also underlines the importance of paying attention to agencement -the interplay of people, artifacts, and situations (D'Adderio, 2017)for understanding how transferring unfolds.
Fourth, by uncovering how complex routines subject to transferring can temporarily be split into different tasks (focused on artifacts, people, and actions, see points D1, 2, 3 and F1, 2, 3 in Fig. 1) that can later be blended back together again, the case unpacked a new variant of decomposing and attending to routines over time. In the case studied by D'Adderio (2014), actors tried to split up their work by observing and meticulously codifying single routines. Similarly, Gupta et al. (2015) found that single routines were disintegrated into "elements" which were then treated in isolation (p. 866). What differed in the case of EuroCo and AsiaCo was the logic of the splitting. Rather than duplicating or revising routines one-by-one, coordination work was broadly focused on transferring "parts" associated with multiple interrelated routines (for instance, transferring written descriptions of all the routines associated with the system and, at other times and places, transferring tacit knowledge about how to perform those same routines). This observation serves to open up new questions regarding how transferring actors should deal with the interrelatedness and multiplicity of routines. For example, Gupta et al. (2015) explained how a lack of attention to interdependencies between routines contributed to economic failure. While the economic success of the transfer between EuroCo and AsiaCo is not known, the case arguably showcases a way to maintain attention to interdependencies. Fig. 2 summarizes the core insights in a simplified model. Looking first at the starting conditions (upper left hand side), the source and target context in this case differed along multiple dimensions: geographical location, language, culture, organizational structure, and ownership. Moreover, multiple stakeholders both inside and outside of the two organizations striving to accomplish the transfer had to be dealt with to recreate the routines subject to transferring. Because stakeholders had differing priorities (for instance, raw milk farms paid more attention to increasing raw milk quantity than quality) there were contrasting interests at play. Informed by these conditions, coordinating agents enacted the transferring process (middle of figure) in a flexible way, where they creatively tried to align their actions with the constraints they faced. Over time (right-hand side), this helped them to learn not only about the limitations of the target context, but also about opportunities for revising the routines subject to transferring. Another set of starting conditions (bottom right hand side) further complicated the transfer and shaped the revision process; the routines targeted for transferring were multiple in number and highly interrelated. Navigating this complexity, coordinators selectively focused their performances on different aspects of the routines: artifacts, people, and actions. This conceptual bracketing of routines seemed to inform coordinators' as they made adjustments (new artifacts, new people, new actions) based on what they learned over time.
The main point illustrated by the model is that the conditions of highly diverse settings, stakeholders with different interests and multiple interrelated routines targeted for transferring informs a process of flexible and pragmatic enactment where coordinators need to learn by doing rather than commit fully to a predesigned plan. The basic idea of what is to be transferred, as well as how, may then be revised as a result of enacting the transfer and responding to the constraints and opportunities at play.

Boundary Conditions and Ideas for Future Research
The case attended to in this chapter has pointed to several conditions that may impact how a transfer unfolds. Naturally, these conditions should be taken into account before applying the insights to other cases, in line with the notion that the presence of "different goals, at different times, in different organizations can lead to very different outcomes" (D'Adderio, 2014(D'Adderio, , p. 1348. For instance, the case illuminates potential boundary conditions of pursuing replication, which has been called for (D'Adderio, 2014;Gupta et al., 2015). Replication theory essentially argues for the benefit of establishing a working model before adjusting or improving routines (D'Adderio, 2017;Winter & Szulanski, 2001;Winter et al., 2012), but this strategy has traditionally been associated with large organizations pursuing economies of scale. Moreover, it has been assumed that the source-and target units are interested in unifying their structures, cultures, and even their identities over time (Winter & Szulanski, 2001). The case of EuroCo and AsiaCo deviates from these assumptions, in that routines were selected for transferring across organizations that would remain different in most respects. Notably, the team of actors put in charge of realizing the transfer (the coordinators) lacked direct managerial control, which could otherwise have helped them shape the target environment to comply with a template. In cases where the source and target contexts of a transfer are highly diverse and the routines to be transferred are highly interrelated, thus, organizations might come to prefer a pragmatic approach unless coordinators enjoy unrestricted authority.
The case of EuroCo and AsiaCo opens up three main avenues for future research. First, there is still much to learn about the role of coordinators when routines are recreated across contexts. For example, in this case as in others (Lazaric & Denis, 2005), coordinators were important for motivating other actors to participate. Coordinators might also play other roles, under other conditions, which could be more closely accounted for. More generally, future studies could inform how different solutions to the coordination problem influence the process and outcomes of transferring. Connecting research in this area with a practice perspective of coordination (Jarzabkowski, Lê, & Feldman, 2012;Spee, Jarzabkowski, & Smets, 2016) would be fruitful. Second, the case of EuroCo and AsiaCo leaves important questions regarding the role of complexity for transferring. In the replication case studied by D'Adderio (2014), a high technical complexity of the routines seemed to inspire the idea to copy routines exactly. In the case of EuroCo and AsiaCo, in which a replication approach was abandoned, there was complexity in terms of multiple inter-related routines. As suggested by D'Adderio (2014): Organizations of lower complexity operating in slower technological and/or innovation environments, for instance, might perceive innovation and replication as being at least partial complements, not substitutes. This might also be the case for organizations in high-innovation but lower-complexity or lower-reliability contexts. These might decide to introduce innovations and improvements at earlier stages of transfer, despite clear incentives to withhold changes to the template until it has been successfully replicated. (p. 1348) More studies focused on the different ways in which transferring unfolds over time, and why, may help further unravel the role of complexity for the process and outcomes of transferring. Third, the case of EuroCo and AsiaCo is in several ways an extreme example because the two organizations had so little in common. One way to summarize their differences is to point to their largely separate histories. Future studies could further illuminate how organizational history impacts transferring of routines as well as inter-organizational coordination work more generally.