The Two Faces of Knowledge Search: New Solutions and Capability Development

Knowledge search as knowledge use in practice

In a narrow sense, knowledge search refers to the practices of ‘looking for and identifying’ knowledge (Hansen, 1999, p. 83) and enables its acquisition (Huber, 1991, p. 125; Wilkesmann, Wilkesmann & Virgillito, 2009). Previous findings demonstrate that knowledge acquisition – in the forms of individual sourcing, knowledge inflow, or the general gathering of knowledge – has a positive effect on individual employees’ performance and the attainment of organizational learning outcomes (Gray & Meister, 2004, 2006; Haas, 2006; Mom et al., 2007; Nag & Gioia, 2012; Teigland & Wasko, 2009; Tippmann, Sharkey Scott & Mangematin, 2012). Knowledge search can include browsing and mobilizing distant knowledge that is related to, yet different from, current knowledge and capabilities (Katila & Ahuja, 2002; Rosenkopf & Almeida, 2003; Rosenkopf & Nerkar, 2001), as well as knowledge more closely related to accumulated experience (Cyert & March, 1963).

In a broader sense, knowledge search not only refers to identifying knowledge, but also leads to exploring the recombination of existing (more or less) familiar knowledge into new solutions, a core activity for creating knowledge and capabilities (Fleming, 2001; Kogut & Zander, 1992, 1996; Singh & Fleming, 2010). Although knowledge search can be distant, in the sense of conscious efforts to search for knowledge that is dissimilar to the organization’s accumulated path-dependent knowledge, leveraging and reusing familiar, closer and ‘older’ knowledge can also offer benefits in terms of its legitimacy and credibility (Katila, 2002; Nerkar, 2003) and in reducing the uncertainty of recombinations (Fleming, 2001) while still producing new outcomes (Katila & Ahuja, 2002; Majchrzak, Cooper & Neece, 2004).

Compared with the literature’s typical aggregated and more general investigations, an important but less explored aspect of knowledge search is that it reveals how individuals engage with their organization’s knowledge system in practice (Collinson & Wilson, 2006) – defined as its global learning (Nonaka, Von Krogh & Voelpel, 2006). This draws attention to actual knowledge use: how managers actually locate and mobilize knowledge and then use it to develop solutions. We suggest knowledge search practices are vehicles for bringing together dispersed individual and organizational knowledge, and for connecting this knowledge to organizational problems for the creation of new knowledge and solutions, so facilitating the synthesis and reintegration of that new knowledge into the organization’s knowledge system. Due to its connecting capacity, we suggest that knowledge search is an activity that can help reduce a major challenge of organizational knowledge creation – the fact that ‘individual knowledge often fails to benefit others in the organization’ (Nonaka et al., 2006, p. 1183).

Middle managers: The potential of bottom-up and horizontal knowledge search

Some employees are disadvantaged in terms of their opportunities for knowledge search (Singh, Hansen & Podolny, 2010). Middle managers, however, play the most central role in realizing the generative potential of knowledge search, for various reasons. Mom et al. (2007) show specifically that both bottom-up and horizontal knowledge inflows to middle managers are positively related to the pursuit of explorative practices, as opposed to top-down knowledge inflows, which tend to serve organizations’ exploitation efforts. Nonaka (1994, p. 32) argues that middle managers are the ‘true knowledge engineers of the knowledge creating organizations’ because they have unique access to the knowledge of both frontline employees and top management, as well as to the expertise of their peers, and are able to convert this mix into new knowledge. Their advantaged position within the organization’s structure allows them to maintain social relationships with a range of actors, all potentially valuable sources for accessing diverse and novel knowledge for knowledge creation (Mors, 2010; Pappas & Wooldridge, 2007), for boundary spanning (Foss & Rodgers, 2011) and for cross-leveraging existing capabilities (Taylor & Helfat, 2009). King and Zeithaml (2001, p. 90) argue that ‘middle managers may be the most direct catalysts for factor mobility of key competencies within a firm’, leveraging and exploiting competences linked to success in specific business units and locations across the wider organization. Together, these studies on middle managers’ knowledge-related roles and practices suggest that they have what we can term a knowledge access advantage: i.e. a favourable organizational position from which they can seek out a broad spectrum of diverse knowledge and pursue both the bottom-up and horizontal knowledge searches that may be needed to achieve knowledge creation or exploration.

We propose the ‘problemistic search’ concept as being valuable for examining middle managers’ knowledge search practices. Cyert and March (1963) define a problemistic search as one which concerns an organization’s responses to non-routine events – situations for which its current practices and processes offer no adequate predetermined response (Nelson & Winter, 1982). Various studies have noted that middle managers’ responses to such problems can be extremely valuable in triggering organizational learning processes (Beck & Plowman, 2009; Kim, 1998; Lampel, Shamsie & Shapira, 2009). They regularly encounter puzzles, problems and challenges as part of their everyday work – indeed, it is typically their responsibility to deal with exceptions that interrupt the stable pattern of their units’ routines (Delmestri & Walgenbach, 2005; Mintzberg, 1980). Non-routine problems reflect an organization’s current renewal challenges, and so are usually experienced by middle managers before they come to the attention of top management. Such problems are often significant enough not only to have the potential for pointing the way towards the creation of new solutions to organizational problems, but also to frame the contributions to bottom-up capability development – defined as improvements in the functioning of a capability (Helfat & Peteraf, 2003) where such solutions modify or create new routines – the fundamental units of higher-order organizational capabilities (Dosi, Faillo & Marengo, 2008; Winter, 2000, 2003).

Towards understanding knowledge search practices and their implications

Although middle managers are recognized as drivers of such knowledge creation outputs as capability development, and knowledge search is understood as a way of mobilizing distributed knowledge for recombination into new solutions, organization studies have paid only limited attention to the nature and implications of middle managers’ actual knowledge search practices. This paper explores two questions: how do middle managers actually search for knowledge to solve problems? and how do their knowledge search practices affect capability development in organizations? Investigating these two key issues allows us to address the following three theoretical shortcomings.

First, the notion of middle managers’ knowledge access advantage does not explicitly distinguish between knowledge search and knowledge access. Although middle managers may have a superior potential to access knowledge, the knowledge search concept relates to the realization of that potential – the taking up of search opportunities in practice, and the actual use of knowledge made available or supplied by individuals and/or the wider organization. It has been demonstrated, for example, that the value of social capital – ‘resources embedded within, available through, and derived from the network of relationships possessed by an individual’ (Nahapiet & Ghoshal, 1998, p. 243) – only materializes when such relationships are actually used to mobilize knowledge (Maurer, Bartsch & Ebers, 2011). But managers often under-invest in searching (Lechner & Floyd, 2007), ceasing their efforts prematurely and/or searching only close to their existing experience (e.g. Simon, 1982). In addition, conflating search and access implicitly assumes that managers are aware of the precise location of knowledge. But this may not be the case, especially in large, diversified and geographically scattered organizations where knowledge distribution is particularly wide, making the search for knowledge more uncertain and difficult (Hansen, 1999). So it remains unclear whether, and how, the difference between knowledge search (as a proactive effort to locate knowledge) and access (with its focus on more ‘general’ knowledge supply) matters in organizations, in terms of shaping search paths and subsequent solution development.

Second, discussions of organizational memory – the ‘storage bins’ of organizational knowledge (Huber, 1991; Walsh & Ungson, 1991) – have tended to focus on the mechanisms organizations use for accumulating and retaining knowledge for later and broader access (Rowlinson, Booth, Clark, Delahaye & Procter, 2010). Knowledge about ostensive aspects of routines – the schematic, abstract and generalized idea of the routine (Feldman & Pentland, 2003) – is often codified and stored centrally in an organization’s knowledge repository and many organizations also deploy expert systems to allow knowledge workers to codify and share their knowledge. This bias in the current literature towards investigating knowledge accumulation has caused a critical oversight: ‘by focusing on how organizations build stores of knowledge [research] pays little attention to how that knowledge might (and sometimes must) be used in unintended and innovative ways’ (Hargadon & Fanelli, 2002, p. 292). Consequently, there are few theoretical insights into how individuals actually search and retrieve knowledge from their organizations (Fiedler & Welpe, 2010, p. 402; Olivera, 2000, p. 861). Building theory about knowledge searching can help develop our understanding of how individuals (such as middle managers) engage with their organization’s knowledge system – what knowledge they mobilize, what challenges they encounter and how they achieve new solutions that then become part of the evolving organizational knowledge system: our study relates to such questions of knowledge use in practice.

Last but not least, knowledge search has major implications for knowledge creation in organizations. Given the potential of middle managers to contribute to capability development and renewal (Floyd & Lane, 2000; Floyd & Wooldridge, 1999, 2000), we need to investigate both the capacity of knowledge search practices to develop new solutions and the implications of those solutions for the wider organization. In order to appreciate fully the knowledge search practices of middle managers at the micro-level it is necessary to unpick their links to such strategic implications as capability development.

Research design and setting

We adopted a qualitative case study research design as being particularly suited to the exploratory nature of our research, and used multiple case studies to obtain more robust findings, as emerging theory is better grounded, more accurate and theoretically transferable than theory induced from a single case study (Eisenhardt, 1989, 1991; Yin, 2009). This study used an embedded design (Yin, 2009) comprised of two units of analysis: middle managers’ knowledge searches and their organizations’ mechanisms for accumulating and circulating knowledge. We focus specifically on middle managers’ knowledge search practices at the micro-level, investigating 38 knowledge search processes that we observed in our selected case organizations. Our case study design allowed us to develop a holistic understanding of a complex phenomenon (Yin, 2009) by unpacking the multifaceted micro-level practices of middle managers in complex organizational contexts. Our research setting consisted of Irish subsidiaries of geographically distributed ICT organizations. We focused on a single industry so as to reduce extraneous variation, and our setting also offered one main advantage – it is a dynamic industry (Brown & Eisenhardt, 1997) in which unusual problems occur regularly and consequently is an attractive setting for the study of the phenomenon under investigation: middle managers’ knowledge search practices in response to non-routine challenges.

Case selection

Having defined the study’s population, we then selected four sample organizations and analysed their knowledge search practices in depth. To safeguard their confidentiality we call them Epsilon, Gamma, Omega and Sigma. The case selection was guided by the theoretical sampling principle: all were wholly owned subsidiaries of multinational corporations (MNCs) and the cornerstones of all their strategies were technology and innovation leadership. We chose organizations that varied substantially in their organizational characteristics to introduce constructive variance to the sample: at the corporate (MNC) level, the companies differed in terms of their predominant business domains (software, hardware, services, solution provision), while at the subsidiary level, the sampled organizations varied in size (indicating the different levels of their local knowledge stocks) and in the varying number and nature of their mandates, reflecting the disparate concentrations and scope of that knowledge (Gupta & Govindarajan, 1991; Hansen & Løvås, 2004; Van Wijk, Jansen & Lyles, 2008). Having secured access to the case organizations and considered their middle management structures, we adopted a wide definition of middle managers to include various mid-level professionals, all of whom were classified by their seniors as middle management based on the scope of their responsibilities and their access to top management (Wooldridge, Schmid & Floyd, 2008). We broadened our sample by including middle managers from areas as diverse as R&D, operations, sales, services and support units. While those we interviewed mostly had line management roles (with people and performance management as their main responsibilities), it was significant that their tasks also involved the drive for improvements (a requirement also shared by our few project-based respondents). They had varying company tenures (from one to over fifteen years) and thus a range of time spans for building their interpersonal networks within the organization: most had progressed into their managerial roles via internal career paths, with only a few being hired externally. Our sample subsidiaries are briefly characterized below and in Table 1:

Epsilon is highly committed to technology leadership, as reflected in its substantial engagement in, and subsidiary mandate for, R&D activities. The unit we collected data from had global responsibility for developing a certain product, which meant it had considerable autonomy.

Gamma nourishes a particularly strong innovation culture with continuous technological improvement at the heart of its vision and cost as only a secondary consideration. The Gamma subsidiary from which we collected data hosts mainly sales activities, and we were granted extensive access to one specific sales unit. Gamma’s global sales activities are organized by region, and it replicates its business processes globally, largely via automated software applications.

Omega is a diversified technology leader with a substantial market and technology reach. The subsidiary from which we collected data hosts a range of different activities – we were given access to its operational and R&D units, which have sister units in other, globally distributed subsidiaries of the MNC.

Sigma pursues a strategy centred on innovation-driven growth. The subsidiary we used for data collection manages a range of different activities, and we included its largest units – services and support, and sales – in this study. Sigma’s middle managers often operate in tightly defined and globally integrated activities.

OPEN IN VIEWER

Table 1.

Description of Sample Organizations.

Characteristics	Gamma	Omega	Sigma	Epsilon
Corporation (MNC)
Domain	ICT services	ICT hardware, software, solutions and related services	Software solutions and related services	ICT solutions and related services
Focal subsidiary
Mandates	Sales	Operations, Sales, Services, R&D	Sales, Services and Support, R&D	R&D, Services
Size	Two separate units <1,500 employees	Four separate units >1,500 employees	Over 15 separate units <1,500 employees	Two separate units >1,500 employees
Unit(s) used for data collection	Sales	Operations and R&D	Sales, Services and Support	R&D
Scope of mandate(s) a	Regional responsibilities	Regional to global responsibilities	Regional to global responsibilities	Global responsibilities
Autonomy a (decision-making authority of subsidiary compared with HQ and other units)	High autonomy (many decisions made by subsidiary)	Moderate autonomy (decisions made by subsidiary, other units and HQ)	Low autonomy (few decisions made by subsidiary)	High autonomy (many decisions made by subsidiary)
Middle management a
Role of middle managers	People management, drive sales and operational improvements	People management, drive operational improvements	People management, drive process improvements	People management, drive process and product development
Company tenure (at point of study)	Low (mostly 2–5 years)	High (mostly 10–15 years)	Moderate (mostly 5–10 years)	High (mostly 15–20 years)
Career path	Internal promotion and managers hired externally	Dominance of internal promotion	Mostly internal promotion	Dominance of internal promotion

a

Refers only to the unit(s) used/middle managers interviewed for data collection.

Knowledge accumulation and circulation

Having negotiated access, the data which we gathered and analysed also offered insights into how each organization generally accumulated and circulated its knowledge, which also provided contextual information on how they supported their middle managers in identifying and locating knowledge and how such knowledge was normally retrieved and searching facilitated. This responds to the need to attend to organizational-level factors as context conditions for knowledge-related activities of individuals (Currie & White, 2012; Foss, Husted & Michailova, 2010).

To accumulate knowledge, codification strategies (Cabrera & Cabrera, 2002; Hansen, Nohria & Tierney, 1999; Hsiao, Tsai & Lee, 2006) played prominent roles at Epsilon and at Sigma, which both maintained large-scale and well-developed central knowledge repositories to provide a common point for searching and for access, in particular so it could be independent of the knowledge holder’s geographic location and was available across time. These databases contained codified knowledge, for example on previous projects, improvement initiatives and business processes. The subsidiary middle managers themselves also frequently contributed to the continuous restocking of these central repositories by codifying their own and their team’s learning experiences to help create business process templates that were then available to peers. This codification and central storage required knowledge inputs to be clearly indexed and categorized so as to structure the corporations’ organizational knowledge in predetermined ways that were easily accessible. Gamma, in contrast, was a good example of personalization in managing knowledge (Hansen et al., 1999). The organization did not store its knowledge centrally but rather it continued to be distributed throughout the organization. Having no central point for knowledge accumulation meant Gamma’s middle managers faced greater uncertainty about where exactly particular knowledge resided in the organization, and so perceived their firm’s knowledge accumulation as being rather unsystematic and ‘fuzzy’. Organizational knowledge was less codified – usually only when business processes were embedded in software tools – so more knowledge remained tacit and embedded in organizational routines or individuals’ expertise and skills. The fourth case organization, Omega, pursued a mixed approach, presenting a certain level of codification and central storage, but a lot of its knowledge remained tacit and distributed and was exchanged interpersonally.

The case organizations also employed mechanisms to promote the circulation of knowledge: two general mechanisms were evident from the middle managers’ perspectives – networking (including expatriation and opportunities for face-to-face interactions) and the establishment of a supportive climate for knowledge sharing (including informal knowledge sharing and an openness/willingness to help). Expatriation has been noted as being an important means by which geographically distributed organizations can stimulate intra-organizational knowledge exchange (Hocking, Brown & Harzing, 2004, 2007); however, although all our case organizations offered their middle managers opportunities to gain international experience if they chose, only those at Epsilon and Omega appeared to have taken advantage of them. Providing opportunities for face-to-face interactions between employees, peers, seniors and other units locally as well as with peers from geographically distant subsidiaries is also important for facilitating knowledge circulation (Gupta & Govindarajan, 2000; Nohria & Eccles, 1992). Intense IT-enabled communication via email, telephone and video conferences was the norm in these organizations (in order to bridge distances), so frequent face-to-face interactions with colleagues working in other locations may have seemed a luxury. During our study period, cost curtailment measures at Sigma (and to a lesser degree at Omega) restricted global travel opportunities – and thus the possibilities for such interactions – but Gamma and Epsilon both organized international conferences at least annually and sometimes more often. These gatherings gave their middle managers the opportunity to network and socialize with peers from other sites, and additional business visits to other sites for specific purposes facilitated further face-to-face communication. Our case companies all showed similar levels of supporting local face-to-face interactions, except at Sigma where management peers and seniors were often located at considerable distances from one another. All four case organizations promoted a climate for supportive knowledge sharing, including informal knowledge sharing and a general openness and willingness to help.

Table 2 summarizes the comparison of the organizational contexts for accumulating and circulating knowledge that constitute the general background for managers’ knowledge searching. It also shows the implications of this organizational context for how knowledge is normally accumulated and circulated: the organizations exhibited similarities and differences which are constructive for theory building. This included mechanisms for knowledge accumulation which ranged from the highly distributed and unsystematic (Gamma and Omega) to the more systematic and centralized (Sigma and Epsilon); while modes of knowledge circulation included facilitation through central storage and interpersonal exchanges (Sigma and Epsilon), as well as through primarily interpersonal exchanges (Gamma and Omega).

OPEN IN VIEWER

Table 2.

Organizational Context for Accumulating and Circulating Knowledge.

	Gamma	Omega	Sigma	Epsilon
Knowledge accumulation
Codification	Low (most knowledge remains tacit)	Moderate (a lot of knowledge remains tacit)	High (emphasis on codification, but knowledge also remains tacit)	High (emphasis on codification, but knowledge also remains tacit)
Central storage	Low (lack of central repositories)	Moderate (mostly for technical knowledge, but weaknesses in implementation as not a managerial priority)	High (emphasis on central knowledge storage)	High (emphasis on systematic, central knowledge storage)
Knowledge circulation
Networking
Expatriation a	Low	High	Low	High
Opportunities for face-to-face interactions b	High locally Moderate globally	High locally Low globally	Moderate locally Low globally	High locally Moderate globally
Supportive knowledge sharing climate
Informal knowledge sharing c	High	Moderate–high	Moderate	Moderate
Openness/willingness to help d	High	High	High	High
Implications for knowledge accumulation	Knowledge accumulation is highly distributed, unsystematic	Knowledge accumulation is mainly distributed	Knowledge accumulation is somewhat systematic, centralized; tacit knowledge remains distributed.	Knowledge accumulation is somewhat systematic, centralized; tacit knowledge remains distributed.
Implications for knowledge circulation	Knowledge circulates freely among employees, interpersonal exchanges dominate.	Knowledge circulates freely among employees; interpersonal exchanges dominate.	Knowledge circulates among employees, facilitated through central storage and interpersonal exchanges.	Knowledge circulates among employees, facilitated through central storage and interpersonal exchanges.

a

This refers only to the middle managers interviewed. All organizations generally supported expatriation.

b

This relates to how favourable the circumstances are to arrange face-to-face meetings and have informal encounters at a local/global level: “high” means very favourable with plentiful opportunities; “moderate” means somewhat favourable with some opportunities; “low” means unfavourable with few opportunities.

c

This refers to the informality of knowledge sharing: “high” means that it is the main mechanism for sharing knowledge; “moderate–high” means that it is an important mechanism for sharing knowledge; “moderate” means that it is a mechanism for sharing knowledge alongside codification.

d

This relates to the perceived willingness of colleagues to openly share their knowledge, and “high” denotes an overall openness and willingness to help.

Data collection

We negotiated access to our case subsidiaries with their senior management, and promised confidentiality in order to facilitate the openness of our interviewees and enhance the possibility of extensive access (Huber & Power, 1985; Miller, Cardinal & Glick, 1997). Data collection comprised four phases: (1) study of secondary sources; (2) interviews with middle managers; (3) interviews with senior-level informants; and (4) review of archival materials. Our analysis focused first on secondary data about each organization at both MNC and subsidiary levels, from annual reports, press releases, the organizations’ websites and other commentaries, all of which helped us develop an understanding of their strategies and subsidiary mandates and set the broader context for middle managements’ tasks and actions. Second (as our main data collection technique) we conducted 34 semi-structured interviews with subsidiary middle managers, lasting approximately an hour (some up to 75 minutes) which yielded data about 42 discrete processes – our unit of analysis for the micro-level data. Four knowledge search processes were excluded from the dataset due to missing information, leaving 38 processes for analysis. The middle manager interviews aimed at gathering material on specific aspects of their actual knowledge search practices. Respondents were asked to recall one or two specific non-routine problems, which had to represent breakdowns or significant interruptions in the unit’s routinized processes and practices for which it had no predetermined organizational response (Nelson & Winter, 1982). The attribution of ‘non-routineness’ was organization specific. Middle managers referred to a variety of situations, including examples such as a significant increase in customer demand with which current sales practices could not cope; a drastic decline in customer retention across several market segments; and severe escalations of technical difficulties, to mention a few. Each manager was then asked to explain, for each particular incident, how exactly they had searched for knowledge as part of their solution-finding activities, and to elaborate on the outcomes of their respective knowledge searches and what solutions were developed. We used open-ended questions and probes to encourage detailed responses as well as to promote the more accurate recall of past actions (Miller et al., 1997), and sought to mitigate further the danger of retrospective bias by asking the middle managers to choose concrete situations that had occurred during the last twelve months, a time frame short enough to ensure a more precise recall of events (Huber & Power, 1985; Miller et al., 1997). We also asked them to focus on their actual doings rather than their opinions, intentions or beliefs, again to increase the level of accuracy in the accounts (Golden, 1992; Miller et al., 1997) and to describe how their actual knowledge search practices unfolded. At the end of each interview we also enquired more specifically about the organization’s provisions for accumulating, storing, accessing and circulating knowledge in order to collect additional organizational context data. All interviews were recorded and transcribed verbatim, and followed up with questions via phone or emails to clarify any ambiguous answers.

For our third data collection activity, we interviewed seven senior-level subsidiary managers (including business directors, general managers and other key informants) to explore what knowledge management tools, knowledge sharing procedures and support structures were in place, how the subsidiary exchanged knowledge with other parts of the organization, and the outcomes of knowledge search processes. Where feasible, we interviewed multiple informants from each organization to guard against possible individual response bias (Golden, 1992; Miller et al., 1997). Finally, we reviewed the available archival material, including selected internal reports, communications, strategy documents and intranet information, which yielded rich contextual data on the subsidiary units and important detail on the outcomes of knowledge search processes. Table 3 summarizes the interview data and non-routine problems included in the dataset.

OPEN IN VIEWER

Table 3.

Summary of Interviews and Non-routine Problems.

	Middle manager interviews	Senior manager interviews	Non-routine problems
Gamma	8	2	14
Omega	8	2	9
Sigma	10	2	8
Epsilon	8	1	7
Total	34	7	38

Data analysis

Our analysis proceeded in three main steps: analysis of knowledge search practices, of outcomes and their implications for the organization and of typical knowledge search paths. The data was coded using an activity perspective focused on managers’ actual actions as they unfolded in their everyday organizational lives (Johnson, Melin & Whittington, 2003; Whittington, 2006) and, more precisely, on the actions and interactions between individuals performing their roles (Jarzabkowski, 2005; Rasche & Chia, 2009). We used the activity perspective in this study as our empirical lens – rather than as a theoretical underpinning (Feldman & Orlikowski, 2011) – to understand middle managers’ actual micro-level actions and interactions, as they sought to fill the knowledge gaps revealed by the specific non-routine problems they faced.

In order to analyse the managers’ actual micro-level knowledge search practices, we applied the broad definition, which includes locating and mobilizing knowledge as well as using it for solution development. We examined all data relating to their practices in detail, along with what middle managers then did with the knowledge they accessed in terms of crafting solutions. Starting by using inductive qualitative techniques, we developed ‘in vivo’ codes (similar to open coding) to reflect the respondents’ language which generated a very detailed representation of the data (Strauss & Corbin, 2008). Examples of these codes included ‘ask for advice from other managers’ (Gamma, process 9), ‘bouncing ideas off people’ (Epsilon, process 2), ‘trying to gain knowledge from a specialist team’ (Gamma, process 13), ‘establish the contacts’ (Sigma, process 6) and ‘get experts on the case’ (Sigma, process 6). As our understanding of the data developed, we aggregated similar, recurrent codes thematically under broader, first-order concepts. These were then summarized in turn under common, second-order themes, each describing a particular search practice, and clearly labelled and defined to avoid ambiguity. We finally clustered these themes under higher-level categories, cycling between the themes emerging from the data and the notion of core theoretical organizational knowledge concepts (Miles & Huberman, 1994; Strauss & Corbin, 2008) in order to integrate and conceptualize the data. These theoretical concepts included knowledge ‘distribution’ (Bartlett & Ghoshal, 1998; Tsoukas, 1996) to aggregate activities that tackled knowledge distribution challenges; ‘socialization’ as described in Nonaka’s (1994) early explication of the knowledge creation theory to theorize activities that comprised tacit-to-tacit knowledge conversions; ‘limited and biased search’ (e.g. Cyert & March, 1963) to cluster isolating activities; and ‘recombination’ (Fleming, 2001; Galunic & Rodan, 1998; Kogut & Zander, 1993) to conceptualize activities to master solution development. Table 4 illustrates how this category-building process progressed.

OPEN IN VIEWER

Table 4.

Progression of Category Building for Micro-level Knowledge Search Practices.

First-order (informant) concepts	Second-order themes		Aggregate categories
Do my own thing My own idea	→ Siloed problem solving	↘	Isolating
Have experience Have been here so long	→ Using own experience	↗
Have/reach out to contact Worked with person before	→ Using existing personal links	↘	Overcoming knowledge distribution challenges
Make contact with experts Reach out to unknown specialist	→ Searching for new link	↘
Use different sources Engage a number of different teams	→ Searching numerous diverse sources	↗
Exchange regularly Set up regular meetings/calls	→ Seeking regular updates	↗
Understand what is going on Figure out what we need to know	→ Interactive scoping of knowledge needs	↘	Socializing
Discuss/ask/work with manager/peers/team Get another opinion/feedback/brainstorm with manager/peers/team	→ Discussing within focal team (manager, peers, team)	→
Set up team Pull in experts	→ Addressing knowledge search collectively	↗
Discuss/brainstorm ideas Collaborate/work/collaborate with others Have person coming here/go there to understand	→ Collaborating with specialists	↗
Ask peers for solution Look at how others did it Search database/repository/ internet/intranet	→ Searching for a replicable solution	↘	Mastering solution development
Localize/adapt/don’t replicate exactly/tweak existing solution	→ Adapting an existing solution	→
Improve/learn from old/existing solution Innovate/design/create New/innovative solution/way of doing things	→ Integrating diverse knowledge to innovate	↗

We also analysed the outcomes from each knowledge search process, namely, the kind of solutions that were created and implemented, in terms of their implications for organizational knowledge development. To make our analysis as comprehensive as possible, we triangulated data from middle managers’ and senior informants’ interviews against our archival data wherever possible, and examined the implications on two levels: the nature of the impact of the solutions found on the organization’s routines and on its higher-order capabilities. To undertake this analysis, we compared the nature of the original routine (data captured in descriptions of the non-routineness of the problem faced) and the kind of solution developed, implemented and habitualized (data captured on the outcomes and their implications) to conclude on the extent of routine change that occurred. Finally, to echo the process nature of non-routine problem solving and to examine the dynamic and complex relationships between knowledge search practices and outcomes, we analysed each search process with respect to the stepwise occurrence and linkages between the practice themes and categories we had identified. For this purpose, we followed the process-tracing technique (George & Bennett, 2005) which brings to the forefront the causal logic within each search process.

After completing our within-case analysis, cross-case comparisons (Eisenhardt, 1989; Eisenhardt & Graebner, 2007) between the 38 knowledge search processes enabled us to identify two common search paths, namely, the modification of an existing routine and the creation of a new routine. To uncover the influence of the organizational context on these search paths, we first analysed the influence exerted by organizational context on the micro-level knowledge search practices to better understand and explain the nature of interaction and causal mechanisms. We applied an analysis procedure similar to Chreim, Williams and Hinings (2007), including examining the knowledge search processes for statements where the middle managers directly referred to aspects of the organizational context in describing their knowledge searching practices, for example: ‘It would be documented in different documents or in PowerPoint slides. It can be pulled together… There are documents here, here, here, and here of each of the individual subcomponents within the overall process and how they work’ (Epsilon, middle manager 3). Such explicit references to the organizational context in shaping search paths signifies the operation of direct cross-level effects and strengthens our confidence in the validity of the specified explanatory conditions. Second, to compare knowledge search paths across the four organizations, we used two categories for comparison: knowledge accumulation (comprised of the second-level indicators codification and central storage), and knowledge circulation (comprised of the second-level indicators networking and supportive knowledge-sharing climate) (see Table 2). As all the second-level indicators interact and interpenetrate to create a particular organizational context for knowledge accumulation and circulation, we aggregated our comparison to the distinct ways of accumulating and circulating knowledge. Process tracing allowed us to track the operation of the identified cross-level effects and to give primacy to within-case analysis over quasi-controlled comparison (George & Bennett, 2005).

We used NVivo9 to manage the data analysis process in a systematic and consistent way, and so increase its reliability (Yin, 2009). Although the data was coded manually, the software proved helpful for storing, fragmenting, reassembling and recoding the data to generate findings progressively. We employed multiple measures to strengthen the trustworthiness of the qualitative data and of our analysis (Lincoln & Guba, 1985): multiple data analysis iterations, moving constantly back and forth between data and theory; triangulation; protecting the respondents’ confidentiality and confirming the validity of our initial analyses with them, including seeking feedback on subsequent analysis iterations. Replicating the study in multiple, heterogeneous organizations, investigating a number of knowledge search processes within each organization, and focusing on commonalities in the responses while acknowledging that each knowledge search process is idiosyncratic, allowed us to generate more robust and transferable insights (Eisenhardt, 1989; Eisenhardt & Graebner, 2007).

Context of search: Challenges of knowledge distribution

Despite the assumption that middle managers enjoy certain advantages in relation to knowledge access, they also clearly encountered challenges in their knowledge searches caused by organizational knowledge distribution. Knowledge is distributed in organizations among individuals and units, and (in geographically dispersed organizations such as MNCs) is further spread across locations in different countries. Much organizational knowledge is also specialized and relates to a particular function, product or business unit. This distributed nature of organizational knowledge, and the fact that managers were often unsure about what knowledge they actually needed, caused two particular challenges: difficulties locating knowledge and feeling isolated from knowledge.

Middle managers’ knowledge search practices

The challenges of knowledge distribution set the broader context for managers’ actual search practices. This section describes the main knowledge search practices (aggregate categories) and their constituent activities (second-order themes) used by middle managers to find solutions to the non-routine problems they encountered. Table 5 offers additional illustrative quotes for each theme.

OPEN IN VIEWER

Table 5.

Middle Managers’ Knowledge Search Practices.

Knowledge search practices (aggregate categories and their constituent themes)	Illustrative quote(s)
Isolating
Using own experience	‘My manager here, Pat, he is like us all. He is as experienced as I am. He’s been at Epsilon as long as I have. Over the years we have built up a lot of experience.’ (Epsilon, process 1)
Siloed problem solving	‘Everyone is trying to sort out the issue as best as they can for their area.’ (Epsilon, process 2)
Overcoming knowledge distribution challenges
Using existing personal links	‘I have my contacts. I can easily put them on cc on an email. And I’d just shout and raise a hand and say: “Hey guys, does anyone know how it works?” And two hours later I’d had it in my inbox: the right contact person.’ (Gamma, process 14)
Searching for new link	‘They [the vendor] had a lot of technical expertise that we wouldn’t have. And we got access to that – quite a high-tech research lab. [They had a] person, a well-qualified PhD, an expert, who really got involved and did a lot of very in-depth work. … And then we fully understood what was causing the problem initially.’ (Omega, process 7)
Searching numerous diverse sources	‘I am in regular contact with engineering groups, globally. I’m in contact with our legal team, our compliance team, our different operational teams, operational managers … at this point, I think it has really touched all the different departments within Gamma.’ (Gamma, process 12)
Seeking regular updates	‘I would have set up monthly meetings or calls or VCs [video conferences] with the operations manager in Central Europe… [It is] an update session. And from time to time we would discover something that may be beneficial for the team on either side.’ (Gamma, process 6)
Socializing
Interactive scoping of knowledge needs	‘Trying to get clarity, get understanding of what happened … They, pretty quickly, sent three of their people on the site for a full week to understand our issue and to see it first-hand.’ (Sigma, process 1)
Addressing knowledge search collectively	‘We set up our own group.’ (Epsilon, process 2)
Discussing within focal team (manager, peers, team)	‘What we did from the very beginning, we actually brought all of those people on board, brought them in. They were part of the development of those strategies. … And we all sat down together.’ (Omega, process 5) ‘We [team]’d all come together and check different things and test different things.’ (Sigma, process 5)
Collaborating with specialists	‘Talking with engineering about the types of solutions they could offer us.’ (Gamma, process 12)
Mastering solution development
Searching for a replicable solution	‘I just went and asked other people because they will have had similar issues.’ (Gamma, process 9)
Adapting an existing solution	‘We took ideas from them [expert unit]. But then again we have a portal, a corporate portal where all, or a lot, of this information is stored. We looked at how they do it: templates. And then we filled it with our own knowledge.’ (Sigma, process 3) ‘We try to learn from the results of the activities already done, and then we try to adapt them.’ (Gamma, process 7)
Integrating diverse knowledge to innovate	‘Develop the tool and the process.’ (Omega, process 9)

Knowledge search paths and their implications

The previous section described the different knowledge search practices which, through stages of combination and interaction, comprised middle managers’ knowledge search process. This section extends these findings to take a broader look at the outcomes achieved and their implications for organizational knowledge development as well as some of the organizational mechanisms which facilitate that development.

Our examination of these outcomes revealed that the solutions managers created not only resolved their initial challenges but could also, either by modifying existing routines or creating new ones, affect organizational-level capability development in important ways. These are the two faces of the knowledge search: crafting new solutions and developing organizational capabilities. From the organization’s perspective, the practices managers pursued had important and differentiated outcomes in terms of reshaping capabilities; the solutions developed became effectively the by-products of their search activities. Searching mobilizes and uses organizational knowledge and incorporates and materializes these efforts into solutions which in turn become habitual routine once implemented and embedded within the organization and so alter organizational behaviours (Levitt & March, 1988). Taking knowledge search outcomes – the modification and creation of routines – as a starting point for analysing typical, cross-case commonalities and differences in knowledge search practices, the data suggests two common knowledge search paths, the modification of existing routines for capability exploitation and the creation of new ones for capability exploration (see Figure 1).

OPEN IN VIEWER

Figure 1.

Knowledge Search Paths: Search Practices, Implications and Facilitating Organizational Conditions.

Knowledge use advantage

The main contribution of this paper to organization studies is to differentiate between knowledge access and knowledge use and to demonstrate that they are both very important for bottom-up capability development (Burgelman, 1983, 1994; Floyd & Wooldridge, 1999; Helfat & Peteraf, 2003). Promoting knowledge access refers to an organization’s facilitation mechanisms for knowledge accumulation and knowledge circulation. Ensuring knowledge access has been the common emphasis in the literature on organizational knowledge, learning and memory (Felin & Foss, 2005, 2009; Fiedler & Welpe, 2010; Friedman, 2001; Olivera, 2000), and our findings are firmly aligned with the core idea that organizations must first structure their knowledge access mechanisms to supply individuals (and especially middle managers) with channels and a context that enables them to access different pockets of organizational knowledge. We argue, furthermore, that focusing on search activities, namely, the individuals’ proactive and self-initiated actions in looking for and identifying knowledge for reuse, extension and recombination, leads us on to the notion of knowledge use – how individuals actively engage with and employ their organizations’ accumulated knowledge in their own idiosyncratic ways. Their searches are driven by the need for additional knowledge, and they may utilize located knowledge to relate what they find to their specific problem in novel ways, thereby making active use of a mix of individual and organizational knowledge. Despite organizational provisions to facilitate the circulation of organizational knowledge and promote its more general supply, individuals still need to engage actively with their organization’s knowledge system through time-consuming and costly search activities: middle managers need to activate their knowledge access advantage. Based on our findings we argue that capability development requires organizations to develop a knowledge use advantage, the ability to mobilize accessible knowledge for circulation, as well as knowledge socializing and mastering solution development, especially at the organization’s mid-levels.

Middle managers and organizational knowledge processes

Our perspective on knowledge use also contributes to the literature on the role of middle managers in organizational knowledge processes, in that we surface evidence about their knowledge search activities that challenges the prevailing and perhaps over-simplistic notion of their knowledge access advantage (Hedlund, 1994; Mom et al., 2007; Nonaka, 1994), based on the idea that their central positions on both the organization’s vertical (top-down/bottom-up) and lateral axes offer them more ‘immediate’ access to knowledge and thus facilitate a wider range of possible interactions. In fact, we find that middle managers are confronted with significant challenges associated with knowledge distribution, and often find it difficult to locate knowledge and, when operating in large, geographically dispersed organizations, they may feel isolated from knowledge in practice, despite strong organizational efforts to promote knowledge circulation. By applying a different empirical perspective, and investigating knowledge search not in abstract or general terms but as a ‘real-life’ activity performed by individuals, our findings suggest that the practice of searching involves finding ways to overcome and minimize the isolation caused by an organization’s knowledge distribution problems. Indeed, we find that middle managers may pursue isolating practices at two levels: that of the individual relying (too) heavily on the untransformed reuse of their experience to deal with new organizational challenges, and that caused by the functional specialization and geographic distribution of organizational knowledge. Specialized knowledge produces boundaries (Carlile, 2002), the overcoming of which may require managers to make substantial proactive search efforts. We also found that the structural advantage middle managers are often assumed to enjoy does not necessarily suffice to give them access to the knowledge they may require, and they may need to depart from exclusively mobilizing existing links to search for new ones. Searching for knowledge among previously unknown colleagues is different from searching among known sources, and middle managers often must approach such sources based on more or less informed ‘hunches’ about where required knowledge could be located, but without the security of exploiting established contacts. Searching for new links is particularly important in helping avoid ‘network rigidities’ (Floyd & Wooldridge, 1999), where interpersonal interactions take on stable patterns over time, while searching for specific new links also adds immediately valuable connections to the social network, and may prevent an organization’s members from ‘engaging in unproductive, yet costly social networking’ that may reduce the value it can derive from knowledge exchanges, given the investment required in building social relations within organizations (Bae & Koo, 2008, p. 249).

Managerial implications

Most established organizations encounter difficulties in breaking with inertial forces and developing their existing capabilities in response to environmental changes. As many capability development processes are driven from the bottom up, engaging with and supporting middle managers’ knowledge search practices in order to develop solutions to an organization’s non-routine problems can represent an important contribution to its long-term renewal. Our findings therefore have direct managerial implications.

First, middle managers need to develop an awareness of the value of non-routine problems in testing the viability of entrenched routines and capabilities. Here, their knowledge search practices are essential as they can initiate the development of new routines. But this potential can be reduced if they merely adapt existing solutions that may have limited suitability when more substantial organizational adaptation is what is really required.

Second, the low level of managers’ use of knowledge repositories for knowledge searching that we observed, together with the benefits of interactive searches for knowledge creation, underlines previous recommendations about the development of organizations’ social structures and linkages as interpersonal search channels (Cross & Sproull, 2004; Mors, 2010; Pappas & Wooldridge, 2007) and channels for exerting agency (Daudigeos, 2013). At the same time, our findings imply that providing knowledge access points and promoting knowledge circulation via, for example, a supportive climate alone are not enough, both because existing ties may not be utilized and because locating dissimilar, novel knowledge may involve searching for new links. So it seems important, as well as providing for enhanced social interaction and establishing a knowledge-sharing climate, that organizations find ways to develop their middle managers’ knowledge search practices, given that they have to deal with the uncertainties of identifying knowledge as an integral part of their search processes. Many of the middle managers interviewed for this study progressed through internal career paths and would previously have worked within highly procedural knowledge management contexts, so it will be important to develop their awareness that the knowledge demands for dealing with novel non-routine problems differ from those involved in handling routine frontline problems.

Limitations and future research

As is typical in exploratory qualitative research, questions arise about the extent to which findings are transferable to other contexts. We employed certain measures to increase transferability: the replication of our study across four case organizations, that exhibit both similar and dissimilar organizational conditions, and the investigation of 38 knowledge search processes, for example, yielded more robust findings. We have also studied a phenomenon, the middle managers’ knowledge searches in response to non-routine problems, which we believe occurs regularly and in similar forms in other organizations, especially those operating in dynamic industries. At the same time, the knowledge search practices and the two paths we identify reflected typical knowledge search patterns in our case subsidiaries. Future research with a wider sample could develop our results and help establish the generalizability of our findings to other organizational and industry contexts.

Another area that may warrant closer examination is identifying the antecedents and drivers of search actions and search paths, including middle managers’ characteristics, the subjective meaning of different search actions (such as expectations of reciprocity) and considerations of wider organizational attributes as well as of the non-routine problems. It may also be worth differentiating between problems according to their novelty, for example, examining whether they are new to the industry or merely to the organization (Spender, 1989), differences in the kinds of outcomes sought (Haas & Hansen, 2005, 2007) as well as the need(s) of the organization for explorative or exploitative capability development. Similarly, we only studied knowledge searching in response to non-routine situations, which can represent critical learning opportunities and which are usually more explorative in nature than more day-to-day, operational challenges. Further research could investigate these more common problems in detail, perhaps linked with how managers use modern ICT in practice (Orlikowski, 2000; Vaast, 2007). The managers interviewed for this study exhibited a preference for using internal knowledge, but other studies have found a preference for seeking external knowledge as possessing greater perceived novelty (King & Lekse, 2006; Menon & Pfeffer, 2003), so more insights are also needed on the micro-level characteristics of knowledge searches that are internal and external to the organization. Last but not least, future research is needed to analyse the efficiency of different search practices and paths.