Abstract
Research on ambidexterity has been focused primarily at the organisational level. There is little in the literature detailing ambidexterity in more complex forms of organising. While much is written on the ‘what’ of ambidexterity – its enactment has been shown to be beneficial – there is a gap in our understanding of the underlying mechanisms, architectures and dynamics by which organisations can achieve both exploration and exploitation. In this article, we use the lens of intellectual capital to extend current architectural theory. A case study of a project management office in an IT/IS multinational provides the context in which to qualitatively unpack the underpinning intellectual capital architecture, illustrating the interwoven application of human, social and organisational capital in a complicated organisational setting. We reveal the co-existence and mutual interdependence of multiple intellectual capital and process elements, further explaining the ‘how’ of this increasingly important area of study.
Introduction
Businesses face the persistent challenge of balancing continuous improvement with innovative problem solving. March (1991:71) defines these two dynamics in terms of requirements for exploitation and exploration, where ‘[e]xploitation includes such things as refinement, choice, production, efficiency, selection, implementation, execution’, while exploration ‘includes things captured by terms such as search, variation, risk taking, experimentation, play, flexibility, discovery, innovation’. The study of how both exploration and exploitation may be achieved, rather than being seen as mutually exclusive, is addressed within the increasing literature on ambidexterity.
The use of the word ambidexterity in scholarly debate has risen rapidly (Raisch et al., 2009) in multiple areas of research, including strategic management, innovation and technology management, organisational learning, and organisational behaviour (Simsek, 2009). Although the generic meaning of ambidexterity is the ability to pursue two different paths simultaneously, there is no consistent definition (Cao et al., 2009; Gupta et al., 2006). The term refers variously to adaptability and alignment (Gibson and Birkinshaw, 2004), controllability and responsiveness (Graetz and Smith, 2005), innovation and efficiency (Sarkees and Hulland, 2009) and incremental and revolutionary change (Tushman and O’Reilly, 1996). For the purposes of this article, we take ambidexterity to mean the ability to refine existing domain knowledge (exploitation) and create new knowledge to overcome knowledge deficiencies or absences identified within the execution of the work (exploration). This is consistent with March (1991) and Benner and Tushman (2003).
It is accepted that ambidextrous learning is beneficial but there is a lack of consensus about what this actually means (Cao et al., 2009; Gupta et al., 2006; Raisch et al., 2009). Indeed debate is ongoing regarding its enablers. Further, the major forms of ambidexterity – temporal (time-based, Tushman and O’Reilly, 1996), structural (separation of units, overseen by senior management, Jansen et al., 2008; Lubatkin et al., 2006; O’Reilly and Tushman, 2004) and contextual (behaviour-based, Gibson and Birkinshaw, 2004) – are primarily understood at the organisational level. There is limited theorisation and empirical evidence within the literature on how exploitation and exploration are achieved in practical, team-based operations, as reflected by O’Reilly and Tushman’s (2011: 8) observation that ‘what is needed is a greater insight into the specific micro-mechanisms required for a manager to implement and operate an ambidextrous strategy’. Recent work by Tushman and colleagues on how managers facilitate exploitation and exploration (O’Reilly and Tushman, 2011; Tushman et al., 2011) signals the transition to a new phase in ambidexterity research. Emphasis is shifting from organisational to agential processes. Hence, it is questionable if existing conceptualisations are still relevant, especially explaining ambidexterity occurrences in a complicated, evolving organisational setting. Using the lens of intellectual capital, we extend current theory and present a case study to illustrate the phenomenon. Specifically, we identify multiple forms of ambidexterity existing simultaneously and build on existing architectural theory by showing how intellectual capital elements enabling this can be understood as orthogonal, interrelated and mutually reinforcing. The aim is to engender further debate and research on managerial ambidexterity
Literature
A line of argument central to understanding ambidexterity is whether the concepts of exploitation and exploration represent a continuum (i.e. one is enacted in preference to the other) or whether they are in fact orthogonal (capable of co-existence without trade-off) (Gupta et al., 2006). In temporal ambidexterity, exploration and exploitation are considered separate in time, with the organisation moving from one dominant mode to the other. Structural ambidexterity also assumes that these modes are separate, with one organisational unit focusing on exploitation, another on exploration. When both are integrated at the senior management level, it produces an ambidextrous organisation. In this context, management occupies a key role in resolving the tension between the two. Raisch and Birkinshaw (2008: 396–397) elaborate:
[C]hoices about how to resolve the tension at one level of analysis are often resolved at the next level down. So for example, a business unit may become ambidextrous by creating two functions or subdivisions with different foci, a manufacturing plant may become ambidextrous by creating two different teams (one in charge of enhancements to flexibility and another in charge of efficiency improvements), and a single team may become ambidextrous by allocating different roles to each individual.
The rationale of each unit is clear, since they represent two ends of a uni-dimensional continuum. Of note is Gupta et al.’s (2006) observation that exploratory research and development units can work effectively with more slow-moving and standardised manufacturing and sales groups within an organisation. This implies that structural ambidexterity can be conceptualised as spatially separated business units, and also as interdependent functional departments.
While structural ambidexterity and structural integration are not necessarily polar alternatives, the literature tends to portray them as end-points on a continuum and this may constrain our theorisation. Gibson and Birkinshaw (2004) introduce a third ambidexterity variant – their business-unit level contextual model emphasises the beneficial choices of individuals who can demonstrate ‘the behavioural capacity to simultaneously demonstrate alignment and adaptability’ (2004: 209). The former refers to coherent business activities working towards a common goal, while the latter refers to the capacity of the organisation to reconfigure those activities as required by the task environment. There is not necessarily a trade-off between the two.
Limited attention is given so far to the multiple, coupled sub-elements within an organisation (such as departments, functional groups or project teams). We suggest that this is an important gap in our understanding as most organisations involve vertical and horizontal structures with dynamic, evolving interfaces between multiple stakeholders (potentially including customers and elements of the supply chain). Such a context is difficult to reconcile with a ‘straightforward’ model of temporal or structural ambidexterity. Following Benner and Tushman’s (2003: 242) call for a lower-level analysis where ‘ambidextrous organizations are composed of multiple tightly coupled subunits that are themselves loosely coupled with each other’, we expect a micro-level analysis can reveal new insights on how exploitation and exploration can occur simultaneously at any point in time and within a given structural space.
These ideas provide a challenge to existing theory, as He and Wong (2004: 481) advise:
Exploration and exploitation require substantially different structures, processes, strategies, capabilities, and cultures to pursue and may have different impacts on firm adaptation and performance. In general, exploration is associated with organic structures, loosely coupled systems, path breaking, improvisation, autonomy and chaos, and emerging markets and technologies. Exploitation is associated with mechanistic structures, tightly coupled systems, path dependence, routinization, control and bureaucracy, and stable markets and technologies.
However, the more levels of hierarchy we attempt to consider within an organisation, the less clear it is that any single model of ambidexterity is satisfactory. Like Gupta et al. (2006) and Raisch and Birkinshaw (2008), we support the idea that the lack of multi-domain analysis of ambidexterity limits our understanding of the concept, and represents an inadequacy within existing theory. Similarly, although the role of the top management team has been investigated (e.g. Jansen et al., 2008; Lubatkin et al., 2006), the impact of the manager at the operational level has received less examination. We have evidence of the ‘what’ of ambidexterity – its enactment at the organisational level has been shown within the literature to be beneficial – but we lack a clear understanding of the underlying mechanisms, architectures and dynamics by which it can be achieved. In this work we also take a wider view, acknowledging such aspects as absorptive capacity (e.g. Easterby-Smith et al., 2008; Jansen et al., 2005) since organisations and individuals bring in knowledge from their broader networks.
The architectural framework of Kang and Snell (2009) is a useful first pass at understanding the mechanisms for interaction and learning at the individual and group level. It uses the concept of intellectual capital (IC) to model an organisation that supports ambidexterity. They argue that organisational knowledge may be broken into subcomponents of human capital (HC), social capital (SC) and organisational capital (OC) (see also Swart, 2006). Kang and Snell (2009) advocate that each of these has both exploitative and exploratory aspects, as shown in Figure 1. HC can be specialist (exploitative) or generalist (exploratory). A cooperative (exploitative) approach to SC (the knowledge embedded in relationships) uses dense social networks, for example established teams with strong ties, whereas an entrepreneurial (exploratory) approach uses weaker ties to seek new knowledge. OC represents organisational processes, and is mechanistic or organic in nature (Burns and Stalker, 1961). Kang and Snell (2009) propose that ambidexterity can be implemented at the organisational level by either a disciplined extrapolation (essentially, exerting greater control over potentially loose-knit generalists) or refined interpolation (allowing greater flexibility for more tight-knit specialist groups of staff) model. However, we contend that in complex environments, utilising one of two ambidextrous architectures may not be adequate to enable organisational success.

Model adapted from Kang and Snell (2009).
Although offering a more comprehensive indication of the mechanisms for ambidextrous exchange, the model in its current form is still potentially inadequate in reflecting the ‘messy’ exchanges and activities within a complicated organisational and social environment. How can we understand this deficiency? Reus et al. (2009) provide a framework for considering and matching the complexity of an organisation’s knowledge investments with that of its knowledge requirements. Many practical organisations can represent complex knowledge fit, and under these circumstances, the refined interpolation or disciplined extrapolation model of Figure 1 appears insufficient. Potentially, in a less complex environment (for example, a functional department with relatively homogeneous staff, within a large and relatively stable organisation), this may represent a more simple knowledge fit, in which case a more defined architecture may be more suitable. We suggest that Kang and Snell’s model and its components require further unpacking to allow closer examination of ambidexterity at the ‘lower-level’ of operations. This is indicated in Figure 2, where each IC element can itself be considered as orthogonal, and not subject to strategic or resource trade-offs (in line with Cao et al. 2009; Gupta et al. 2006; Raisch et al. 2009). At the organisational, group, or department level, it may be unsurprising to find specialist and generalist staff, a range of diverse social contacts and relationships, and a balance between procedural rigour and flexible innovation. However, it may also be identifiable in individual managers in terms of their role. This is not an area that has been investigated within the literature.

Subcomponents of ambidexterity.
Project management and project management offices (PMOs)
We chose the context of projects and their management as an appropriate setting for this investigation. Project delivery is a combination of the learning and efforts of individual participants within a temporary organisational structure. The project itself should enable further learning (Arthur et al., 2001; Cacciatori et al., 2011). Project management is increasingly characterised by professional standards and organisational procedures with managers expected to apply the tools and techniques from a specific body of knowledge. The assumption is that structured planning and monitoring is an effective measure against the uniqueness of the project circumstance. However the failure to learn from past successes and failures continues to worry businesses and scholars (Williams, 2008). Each project has its own set of challenges, requiring strong absorptive and adaptive capacities of managers to select and develop their knowledge strategy for the project (Swart, 2011).
Our fieldwork was conducted at one of the project management offices (PMOs) within a large multinational. PMOs are used in project-based organisations to develop, maintain and institutionalise project management competences (Hurt and Thomas, 2009). An early author of the subject, Marsh (2001) conceived PMOs as providing important secondary added-value services to the project delivery team whereas the US Project Management Institute (PMI) portrays the PMO as the locus of project coordination and control. Its Body of Knowledge (PMI, 2008: 11) defines a PMO as:
an organizational body or entity assigned various responsibilities related to the centralized and coordinated management of those projects under its domain. The responsibilities of the PMO can range from providing project management support functions to actually being responsible for the direct management of a project.
This perspective attributes to the PMO strong status power as a central driver for project delivery and success. The UK Association for Project Management (APM), on the other hand, views the PMO as occupying a service function. Its focus is less about what the PMO is, and more on its relevance for achieving overall organisational benefit. Hence, a PMO can range from simple support functions for the project manager to being responsible for linking corporate strategy to project execution. A common thread across the different perspectives is the PMO’s coordination role and Julian (2008) shows that the role of the PMO manager is identified as a facilitator of cross-project learning within multiple communities of practice, and knowledge transfer between projects can be understood in terms of a constellation of socially situated practices (Erhardt, 2011; Kauppila et al., 2011; Swart et al., 2011). It also provides a context characterised by forms of complexity (Geraldi et al., 2011).
Regardless of their origin, purpose and structural configuration, PMOs are an ideal context for an empirical investigation addressing the question, ‘How is ambidexterity achieved?’ because it allows investigation at the individual, social and organisational level. The APM’s principle-based view of a PMO as a facilitator contributing to project success through its boundary-spanning activities was adopted for this study. As a centre of excellence, it can assume a narrow and localised focus or a broad and strategic agenda. Depending on its funding and support, it could simply be a repository of corporate project methodology, standards and project-specific information, or it might be responsible for transferring best practice to the projects, or alternatively, for project oversight and governance.
The case PMO
The case setting is a UK-based PMO within a large multinational IT-services provider. Given the diverse service catalogues and structures, it is not intended that this is a ‘typical’ PMO but its size, duration, services provided and contract status presented a rich context for examining the inherent tension between the requirements for stability and the realities of change (i.e. a complex setting for the study of orthogonal ambidexterity). It is presented as a case through which we can illustrate and examine aspects of sub-organisational ambidexterity, rather than to represent PMOs in general or organisational change. The level of analysis is the management team. We also interviewed senior managers on the client side and studied a large, recently completed, database upgrade programme supported by the PMO to understand the group’s relationship with these two types of stakeholder. This was intended to establish whether the extant ambidexterity literature was sufficient to characterise the activities or whether a more complex view could be established with empirical evidence from such a setting.
The purpose of the wider case programme has been to harmonise payroll and pension administration for a large UK public-sector organisation through a series of integration and upgrade projects and subsequent operational maintenance. The work requires both exploitative (overseeing the payment system, management reporting and administration, day-to-day support for project managers, together with scheduled quarterly releases for the IT system) and exploratory elements (including major system upgrades with a high degree of technical and socio-political complexity). The study PMO is long lived, having started in 1998, and the client stakeholders appeared to have grown reliant on the PMO managers and personnel. It is located at the client’s premises, and multiple programmes and projects are embedded within the service contract. In the first 11 years a strong and stable relationship developed between the PMO manager, who had served from the outset, and the client, supported by the ‘cost-plus’ nature of the contract.
Three major events around the time when the 16-month study began created significant disruptions and pressures for the PMO team. First, there was a change of leadership with the retirement of the original PMO manager. Second, the long-standing ‘cost-plus’ support contract came to an end and a new ‘payment-for-service’ contract was put out for competitive tender. In addition, the parent company of the study PMO was purchased by a large multinational organisation during the case study period. The original organisation had a service-oriented culture, whereas the new owner was more product and efficiency driven. Organisational changes from this takeover resounded throughout the remainder of the study period, providing an extra level of complexity that the PMO had to deal with. Figure 3 gives a timeline showing events and contingent effects. Our focus is not about the changes per se or their management but on how, against a turbulent and complex background of personnel, contractual and organisational changes, the PMO management team continued to deliver their projects and programmes.

Research timeline.
This case is particularly interesting as three forms of ambidexterity (temporal, structural, contextual) can be identified (which could not have been envisaged at the outset), at a level for which there is a paucity of studies – that of a relatively small group within a larger structure. Temporal and structural effects can primarily be described at the sub-organisational level (i.e. the PMO and its environment, with the trends in Figure 3 and the changes to the organisational structures), although existing theorisations are inadequate to capture the dynamic realities. The contextual ambidexterity can best be understood at the individual level and is the primary focus of this study. However, they are interwoven and to attempt to disentangle them is to miss their complex interplay.
Data collection
The research team undertook six formal site visits, performing in-depth interviews with the newly appointed PMO manager, his predecessor, PMO team members, the delivery manager and the change manager (responsible for technical changes to the system) to gain a number of perspectives. There were additional informal telephone and email conversations to keep abreast of ongoing issues, and workshops were also held at the University, which some of the study participants attended. On separate occasions we also interviewed senior managers on the client side and the programme manager of a large system upgrade, key members of his staff and the PMO lead assigned to the work. The interviews included investigative questions about human, social and organisational capital, and exploitative and exploratory processes in the context of the PMO and its projects. Additional discussion uncovered the more general, unfolding story of the PMO with each visit and through the regular updates. A total of 16 respondents were interviewed, either individually or in groups, although the primary focus was the role of the manager in enabling ambidextrous attainments at the PMO and project level. Details are shown in the Appendix.
All on-site interviews were recorded and fully transcribed for analysis. The research team also had access to key documents (reports, key financial performance targets and metrics, procedural documentation and organisational reporting structures). For confidentiality reasons, documentation could not be removed from the site, but together with other staff interviews, this acted as a method of triangulation (Yin, 2009) within the case to enhance the internal validity of the study. The researchers met regularly to discuss the findings and interpretations and also shared these with colleagues performing parallel studies in different parts of the organisation to ensure that a consistent understanding was generated from the data.
Analysis and findings
We used a qualitative approach to gain a richer picture of the PMO operations, the role of the manager and the architecture enabling execution so that we might better identify the presence of ambidexterity, understand its accommodation and thereby address the research question in terms of the Kang and Snell (2009) model. The interview transcripts were analysed using an initial a-priori coding structure based on the intellectual capital elements (HC, SC and OC, each further categorised as exploitative or exploratory), together with exploration and exploitation (the combination accepted as ambidexterity). (However, as will be discussed shortly, emergent findings enabled new ideas to be developed from the data which moved beyond the original conception). From this, the first stage of analysis was to identify the nature of the architecture of the IC elements and their exploitative and exploratory forms, and the processes of exploitation and exploration that could be recognised. The coding was augmented through reviewing instruments previously used in the ambidexterity literature. Twenty-one instruments were published at the time of the analysis, and these aided in clarifying the coding detail.
At the start of the research period, the responses of a group of key team members emphasised that most of the work was relatively stable, as might be expected for a long-lived, mature IT system at the upgrade and maintenance phase of its lifecycle. This maturity was reinforced by in-depth interviews with project analysts, whose functional roles were stable and had remained relatively constant for several years within the PMO. However, with changes coming, it was accepted that this state would not remain constant. This made the context particularly interesting, as we had the opportunity to examine ambidexterity under these turbulent conditions. As the delivery manager remarked before the contract change:
Some of the requirements in the contract are significantly different, so it’s going to require a behavioural change on both sides. As [PMO manager] said earlier, the pricing mechanism is different; the way we charge for our services is effectively going to be completely different. We’ve been in this period for the last ten years. Everyone has got quite comfortable the way things are working and now in a very short space of time we’re going to have to try and transition into a very new behavioural model and that’s going to take some work and some focus in order to be able to achieve.
The managerial role was investigated using the framework of Figure 1 to see if this categorisation could be applied to managerial work, and if one mode of ambidexterity was dominant. However, it rapidly became apparent that all six IC elements were indeed evident, contrary to the ‘either/or’ approach of the Kang and Snell (2009) model. For example, in the category of human capital, neither solely specialist nor generalist applied for the PMO manager’s role:
Because it’s quite a wide range of areas that I cover, I would say that there are areas I have a lot of experience in specifically, so in those areas I tick the specialist box, but because of the wide scope of things I’m expected to cover, I am certainly not a specialist in a lot of them. So I couldn’t pigeon hole myself in either, to be honest. A generalist with a specialism in certain areas.
This was also evident from the programme manager, who relied upon his technical expertise in the role, but also needed an appreciation of the wider business issues. This came through similarly from the client senior manager, who embodied a range of specialist and generalist skills that were hard to emulate:
The model is to outsource more … but they couldn’t find anybody to replace me… somebody who has IS and design information skills, programme management skills and is aware of the business. (Client 1)
In terms of social capital, team relationships were the key to success, and aspects of leadership were based on these.
Very important I would say, if you want to get the best out of people. People need to share your vision and your goals and in order for people working on your project to share your vision and your goals, you have to have a relationship with them. (Programme manager)
The communication used by the respondents balanced formality (exploitative) and informality (exploratory). Wider relationships, built up over time, were also important, enabling information to be accessed based on a history of working together:
There are a lot of occasions where a new name will come into the frame in a new role, and the account manager will say, we’ve got so-and-so just taken this role, don’t know who he is. And from the experience I have built up, relationships in previous roles and over the last 12 years with the company or within the business unit, it means it gives you a head-start. If you’ve got a pre-established relationship, even if it’s on a different level … you’ve got mutual ground to talk about, it helps and it makes you more valuable to your bosses. (PMO manager)
The PMO lead also used his network to ensure his planning was based on others’ experience as well, which also enabled flexible choices to be built in:
It was a complex proje
However, the new contractual requirements limited some of the informal communication processes that had grown up between staff. As the PMO Change Manager commented:
“The [client] people sat on one floor and the analysts sat further up and the team are sat further up. Then they just wandered up and talked and that used to be acceptable but it’s not now. It was good; now they have to charge for it. Previously in the core contract you could do that because the time would get charged and we would recover the time. We won’t recover the time, so if they spend two hours talking to somebody about a plan, they need a piece of paper in their hand that says I can charge you two hours for doing this. That’s going to make it for me quite difficult.”
One of the managers on the client side also recognised instances where the communication channels were clearly no longer working effectively and these limitations were identified as factors hindering project performance.
at the lower level, people have stopped talking to one another, it permeates up into a difficult discussion … They are worried about our behaviour and they have given us a matrix of people we should speak to on each level when there’s an issue … this is unacceptable as we are on this account together. (Client 2)
Finally, the management of organisational capital also exhibited both the mechanistic processes (procedures, meetings, reporting and so forth) necessary for smooth operation, and an organic, flexible approach to accommodate situation-specific issues.
At a high level it’s structured, so I’d say it’s more of a framework to operate in. Whereas you’ve got very fluid local processes fit within that framework, so you have a boundary that is established for you, and you know the limitations of that boundary, but how you operate within it is very fluid. (PMO manager) There’s process, and there’s process overkill … My focus was day-to-day hands-on stuff, following at the high level these quality gates I found very, very useful. So it really gives you and everybody else that I call the key stakeholders, where we are going, how we are moving that along. (Programme manager)
As we theorised, the coding and analysis of the client and supplier managers’ responses showed that they used both exploitative and exploratory elements of HC, SC and OC (i.e. all six of the elements in Figure 1), which helps our understanding of the significance of managerial resources in enabling ambidexterity at the level of the project or organisation. This remained the case throughout the study, and for the managerial role the ambidextrous approach (via the exploitative and exploratory HC, SC and OC elements) was evident despite the temporal emphasis on exploitation or exploration at the higher organisational level. Issues such as limitations on the communication channels were identified as practical barriers, but the desire to retain both strong and weak ties to enable work progression was apparent throughout.
In contrast, as might be expected, we found a more exploitative response in each category among the project analysts. The wider scope and complexity of the PMO and project/programme management role may therefore be the context in which intellectual capital ambidexterity is more beneficial and the multiple facets more important in work execution. It was also apparent that the managerial role was critical in supplying the leadership that enabled and encouraged contextual ambidexterity in others. Although ours was clearly a limited sample, our findings go towards providing a greater practical understanding of contextual ambidexterity in such an environment. We note that this contextual ambidexterity, the basis of our initial theorisation, remains evident alongside (broadly) structural and temporal ambidexterity and although these are conceptually distinct, the complexity of the case allows all three to be identified as co-existent phenomena. This approach has not been taken previously within the literature.
We interpret the managerial role (based primarily on the responses of the managers, as detailed in the Appendix) as the key influencer of the project-level exploitative, exploratory or ambidextrous orientation according to the specific requirements at the time (points a, b and c respectively in Figure 4).

Ambiguity in the ambidexterity process.
However, the analysis of the exploitation and exploration process elements within the PMO proved to be more complex. Two specific findings emerged from examining the transcripts. First, although the theoretical concepts of exploitation and exploration are well established (March, 1991), and have been used differentially in many survey instruments, they can be considered as end-points. Examples from this case would be the lesson-learned procedure (exploitation), and the uncertainly in planning a novel technical solution to a fuzzy customer requirement (exploration). Much of the discussion within the interviews, though, covered aspects of operational improvement that were hard to categorise (and therefore code) as either exploration or exploitation:
So it’s developing internal resources, general improvements, changing to make a better outcome. (Manager1) It’s evolving as well as reacting to different organisational changes and changes in direction. (Manager2) And basically we allowed a programme that if it failed on the first weekend, that wasn’t complete failure. It was – we shift. We move, we work it out. And you can see nearly 4 contingency weekends to allow us, and at the same time still being able to pay everybody and over a particularly difficult period because it’s the Christmas payroll when they have to compress everything and everyone goes on leave. (Programme manager)
We argue that while high levels of exploitation, exploration or both are conceptually straightforward, there is a region of ambiguity (Figure 4) that acknowledges the orthogonality of ambidexterity but incorporates the difficulty of assigning a dimension to identified practices (supported by Lavie et al., 2010). This is in contrast to the quantitative studies whose instruments look for evidence of each form, but in line with accepted definitional issues within the study of ambidexterity (Cao et al., 2009; Gupta et al., 2006), and is a practical finding originating in the qualitative nature of the study. Exploitation and exploration are broad, ‘umbrella’ constructs (Suddaby, 2010), yet the interpretation of the data revealed more. We concur with Farjoun (2010) that exploitation and exploration can be mutually reinforcing, and considered as a duality rather than a dualism. In this light, we took the double-helix structure as a metaphor for their interaction over time as separate yet temporally interwoven and coupled dynamics. This interpretation helps to clarify our understanding, and admits the impossibility of classifying activities as ‘either/or’.
It soon became clear that a far richer picture could be obtained regarding the operationalisation of ambidexterity than had been anticipated at the outset. The initial analysis and coding were based on the intellectual capital architectural framework and forms of ambidexterity (understood as exploitation/exploration), and the data supported the initial theory and the purpose of the interviews. As part of this analysis, conflicting findings were sought to better understand the theory and evidence (Eisenhardt, 1989). None were found, although we did identify that the IC elements could be further sub-categorised to provide context-specific extensions to the model of Figure 1, increasing its analytic value. However, in performing the coding, further understanding surfaced (e.g. Leidner et al., 2009), and it was apparent that more benefit could be derived from the data. Specifically, many of the examples from the interviewees actually fitted multiple codes, which gave far more scope for analysis. In aiming to code for the primary essence of the text, we had difficulty disentangling the constructs. Attempting to use a hierarchical coding structure was not satisfactory in this regard and so a second round of coding analysis was performed on the data with the purpose of parallel-coding (King, 2004), also known as simultaneous coding (Saldaña, 2009), to highlight the emergent interaction between the coding elements. Given the difficulty of exploitation/exploration above, it was evident that there were strong inter-linkages between the elements under investigation, and to conceive of them in isolation was inadequate. Saldaña (2009: 62) advises caution, as this may appear to indicate indecisiveness on the part of the researchers. However, in seeking to specifically identify the co-occurrence and interaction of constructs, we argue that the applicability here is advocated due to the purpose of its application. We coded for the interactions between each of the elements of Figure 1, and processes of exploitation and exploration (i.e. to identify interactions between these eight categories). This coding scheme used HC1 for exploitative human capital, HC2 for its exploratory equivalent, and similarly for SC1, SC2, OC1 and OC2, together with X1 (exploitation) and X2 (exploration) to enable clear analysis and a simple nomenclature.
A multitude of complicated coding interactions were identified using this method, yielding further theoretical insight. For example, the PMO manager’s human capital was closely enmeshed with the choice of exploitative or exploratory operational processes (coded as HC1-HC2-X1-X2):
When I came back into this role from a previous role within [customer] account, there were several areas where I tried to repeat what I had done previously and quickly found out that wasn’t going to work. So the experience I had there was valuable, but could have led me entirely down the wrong path if I didn’t have the attitude of, OK, let’s look at this as a separate problem, apply the pieces from that previous experience that work in this case, but don’t blindly follow everything that worked last time.
Exploitative HC was also dependent on SC to be effective in the organisation:
I would say that myself and some colleagues at my level have a good overall picture. We have a clear view of what the end game is, we know what the underlying key risks are to the programme but even within areas that I am leading on, I would say the detailed specialist knowledge, on some occasions I don’t have that, I rely on the experts within the tasks. But again, I try to have regular meetings with all the managers that lead those areas, to make sure that any key issues or progress I have a good handle on. I don’t know the specialist detail, but I know where they are and what the key issues are they are working through. (PMO manager)
HC, SC and OC can also combine, and to conceive of them as independent misses their interaction in enabling practical solutions. As described by the database programme manager:
So in terms of going off-piste, so to speak, as a project manager, I’ve got a framework that I follow, but I don’t follow it rigidly. I take advice and direction from the likes of my boss when I need to … we had a number of occasions where we had to shift the deadlines of the project because of the availability of weekends, and because of changes in the operational schedule … It meant that the implementation plan I had, had to be shifted so I had to do some out-of-the-box-thinking on that. I’m not a renegade by any means, but I’m a bit of a pragmatist, not in the world of dogma.
Analysis of the interviews showed that conceiving intellectual capital resources as separate inputs to the process of contextual ambidexterity was insufficient; instead, the evidence from the coding analysis performed on the case data showed that not only were the IC elements strongly interrelated, but that attempting to extricate the ‘input’ and ‘process’ elements missed the insight that could be obtained by considering their interrelationship. This was especially prevalent for organisational capital, where disentangling the mechanistic processes and organic flexibility from the actual project delivery activities was difficult. This was the strength of the parallel-coding activity, in that these multiple interactions could be revealed. Indicative concepts are shown in Figure 5.

Interaction of constructs.
Discussion
At the outset of this article, we highlighted the issue of ambidexterity in complex working structures, and sought to understand better the architectures in such a context. Studies to date have primarily focused on ambidexterity as an organisational-level phenomenon, and there has been a lack of theorisation and empirical work at the sub-organisational or operations level of analysis. We drew upon the work of Kang and Snell (2009) and built on this through theory development and an illustrative empirical case. We theorised that orthogonal elements of intellectual capital could be used to explain the underlying architecture and operating mechanisms. The concepts of human, social and organisational capital were used, and we showed that these subcomponents of IC can be considered as co-existing and interdependent resources, each with orthogonal exploitative and exploratory facets at the managerial level. This is a significant extension to existing theory, in that it reveals the multiplicity of resources enabling ambidexterity, and the complexities of their interactions not only with each other but also with the processes of exploitation and exploration. We focused here on contextual ambidexterity, but noted earlier that the case could also be understood in terms of temporal and structural approaches.
At the start of the research period, the PMO’s role was at the account level, supporting the project managers (hardware and applications) and the business process outsourcing function for the customer. In the central stages, once the new contractual conditions had been implemented, the role of the PMO manager widened, with functions expanded to cover business support, finance and formal ongoing changes. Subsequently, a new delivery model whereby functions could cover multiple sets of work within the wider account structure (i.e. greater engagement with the client organisation in more locations) had to be implemented. Analysis of the organisational structure at these three stages of change showed complex, dynamic, network relationships. This corresponds to Benner and Tushman’s (2003) call for an analysis of tightly coupled subunits. Aspects of the organisational structure do indeed correspond to structural partitioning, including not only the functional elements of the PMO reporting to the manager, but also the interrelated projects, operations groups and additional outsourced technical functions. Structural ambidexterity is therefore present and identifiable, but a ‘simple’ partitioning model is inadequate, since the reality was a complex matrix of functions changing over time, requiring contextual responses.
At the organisational level, the trend was broadly from a more exploitative approach, to a more exploratory one, and back again, and can be interpreted (potentially superficially) as an indication of temporal ambidexterity. However, the observed temporal and structural ambidexterity require further analysis at the lower PMO level where the effects of strategic decisions are felt, but where the PMO manager and his team had the capability to (and did) emphasise the opposite; namely, to seek to preserve the positive customer-contractor relationship created under the ‘old’ regime by remaining focused on the tasks at hand, delivering quality service and an ordered, systematic implementation of the post-merger organisational improvements. At the same time, cognisant of the need for improved efficiency and transparency under the revised contract, the new PMO manager was keen to find ways to improve operations performance. Hence, the evolving environment (especially the behavioural change required for the new contractual conditions) also necessitated the demonstration of contextual ambidexterity from the PMO manager and his staff, to respond to the demands of both client and parent organisation. This involved balancing efficiency and effectiveness to meet margin targets whilst delivering services in line with the new contract and offers the possibility of extending current theory by showing how contextual ambidexterity may in fact support structural and temporal forms.
The complexity of the environment means that our conception of ambidexterity should not necessarily be limited to a single model. Indeed, this work has highlighted that ‘simple’ models of temporal and structural ambidexterity are also insufficient at the sub-organisational, working level, and require further refinement if this theorisation is to move forward.
The theoretical implications of our findings are important. The use of an intellectual capital perspective is established within the literature, yet the subcomponents are generally understood as (broadly) separate, although co-existing, concepts (Subramaniam and Youndt, 2005; Youndt et al., 2004). Indeed, Kang and Snell (2009: 86) write: ‘[w]hile we have argued that human, social and organizational capital are conceptually distinct, the three may be related in practice – one may affect the others’. This work has illustrated the complex interrelationship not only between these elements, but also with the processes of exploitation and exploration.
The use of a qualitative approach begins to build a clearer understanding of the intricate architectures and mechanisms of ambidexterity through both theory building and empirical supporting evidence, and the insight informs our understanding of management learning. This case has shown how work execution (both planned tasks and responses to change) is entwined with the resources underpinning it. Managerial human capital shapes the operational choices, social capital underpins communication and group cohesion, organisational capital creates the framework through which the work is structured, and these can be mutually reinforcing. Our results offer qualitative evidence of their interdependence, and these intractable interconnections highlight the complexity inherent in organisational or project-level ambidexterity. The exploitative/exploratory learning orientation at the project level may be orchestrated by the manager, but this orchestration is both subtle and complicated, drawing upon IC, especially social and organisational processes. Through this intricate mix the diverse skills of the participants are leveraged in pursuit of the project goals. This is echoed in the understanding of the explore/exploit dimensions, where qualitative conceptual and definitional ambiguity arise, posing a challenge to the accepted demarcation between them. We proffer the response that the conception of duality enhances our knowledge in this respect. The qualitative approach has revealed the rich interactions of the elements, building on previous quantitative research which has largely focused on them as separate constructs. This work not only contributes to our theoretical understanding of ambidexterity, but also enables practitioners to understand better how the elements in Figure 1 work in their context, what sub-elements may be suitable in their role, and where any deficiencies or weaknesses may be.
Conclusions
This work has built on previous ambidexterity research to theorise and empirically identify the orthogonality of intellectual capital subcomponents, together with their mutual interdependence, that support contextual ambidexterity, extending existing work. The qualitative approach has also shown the significant interactions between these resources and also with the processes of exploitation and exploration, highlighting the complexity of this interplay in a way not revealed by previous quantitative research and thereby opening up new avenues for investigation.
A limitation of the work is that it is based on a single case study; broader generalisations are necessarily unclear. However, we believe that this study provides a platform from which to understand better how ambidexterity is effectively enabled within organisations. Subsequent research might examine specific managerial practices that facilitate exploitation and exploration at the project or organisation level, and it may be illuminating to further understand qualitatively how ambidexterity may be achieved via the interactions and dynamics of multiple managers.
These concepts have implications for practice in that they are powerful as frameworks for managers to reflect upon their organisations and behaviours. Creating an awareness of the balance between exploitation and exploration offers the ability to differentiate between these applications, to make conscious choices and to review context-specific actions in light of previous experience. The intellectual capital model enables further examination of the managerial role by investigating specific subcategories of each component to determine their applicability and value in a range of scenarios (such as which HC attributes, SC relationships and OC processes are most beneficial in the execution of the work – and which may be lacking and therefore require attention). This may also aid in understanding the wider question of operating modes and mechanisms by which managers may orchestrate ambidexterity effectively to enable successful outcomes, of interest and benefit to both scholars and practitioners. Our contribution has been to acknowledge the complexity of the knowledge requirements at the working level of complicated organisational structures and reflect that complexity in our understanding of ambidextrous architectures. We believe this work opens up a wide range of further investigation.
Footnotes
Appendix: Data collection detail
| Meeting | Format | Participants | Duration |
|---|---|---|---|
| 1 | Individual interviews (6) | New PMO manager; Retiring PMO manager, Programme manager 1, Change manager, Analyst 1 and Analyst 2 | 6 hrs |
| 2 | Group interview (7) | PMO manager; Programme manager 2, Change manager, Delivery manager, Analyst 1, Analyst 3, Project coordinator | 2 hrs |
| Individual interview | PMO manager | 1.5 hr | |
| 3 | Individual interview | PMO manager | 4 hrs |
| 4 | Individual interview | PMO manager (inc. contribution from Analyst 4) | 3 hrs |
| 5 | Individual interviews | Two senior managers on the client side. | 2.5 hrs |
| 6 | Individual interviews | [Study of a large database programme implementation] |
4.5 hrs |
Acknowledgements
The authors would like to thank Juani Swart, Harvey Maylor, John Ward and the anonymous reviewers for their invaluable help in developing this paper. We would also like to take this opportunity to express our gratitude to the participants in the case organization for their time and assistance over the course of this study.
Funding
This research received no specific grant from any funding agency in the public, commercial of not-for-profit sectors.
