Content uploaded by Joseph P. Broschak
Author content
All content in this area was uploaded by Joseph P. Broschak on Nov 10, 2020
Content may be subject to copyright.
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Use of collaborative technologies and knowledge sharing in co-located
and distributed teams: Towards the 24-h knowledge factory
Amar Gupta
a,*
, Elisa Mattarelli
b
, Satwik Seshasai
c
, Joseph Broschak
a
a
Eller College of Management, University of Arizona, Arizona, United States
b
Department of Engineering Science and Methods, University of Modena and Reggio Emilia, Italy
c
College of Engineering, MIT, United States
article info
Article history:
Received 27 February 2008
Received in revised form 21 June 2009
Accepted 10 July 2009
Available online 12 August 2009
Keywords:
Globally distributed teams
24-h Knowledge factory
Knowledge sharing
abstract
The relocation of knowledge work to emerging countries is leading to an increasing use of
globally distributed teams (GDT) engaged in complex tasks. In the present study, we inves-
tigate a particular type of GDT working ‘around the clock’: the 24-h knowledge factory
(Gupta, 2008). Adopting the productivity perspective on knowledge sharing (Haas and
Hansen, 2005, 2007), we hypothesize how a 24-h knowledge factory and a co-located team
will differ in technology use, knowledge sharing processes, and performance. We con-
ducted a quasi-experiment in IBM, collecting both quantitative and qualitative data, over
a period of 12 months, on a GDT and a co-located team. Both teams were composed of
the same number of professionals, provided with the same technologies, engaged in similar
tasks, and given similar deadlines. We found significant differences in their use of technol-
ogies and in knowledge sharing processes, but not in efficiency and quality of outcomes.
We show how the co-located team and the GDT enacted a knowledge codification strategy
and a personalization strategy, respectively; in each case grafting elements of the other
strategy in order to attain both knowledge re-use and creativity. We conclude by discuss-
ing theoretical contributions to knowledge sharing and GDT literatures, and by highlight-
ing managerial implications to those organizations interested in developing a fully
functional 24-h knowledge factory.
Ó2009 Elsevier B.V. All rights reserved.
1. Introduction
The relocation of knowledge work to emerging countries has been largely analyzed as a cost-savings driven phenomenon
(Manning et al., 2008). Reports from the Association of Computer Machinery (ACM), the Institute of Electrical and Electronics
Engineers (IEEE) and the National Society of Professional Engineers (NSPE) have described offshoring in the context of jobs
being gained or lost due to cost savings between nations like in a competitive zero-sum situation, where work can only be
done in one country or the other (IEEE, 2004; White, 2004; Asprey et al., 2006).
Research in organizational theory, strategy, and psychology reinforces the idea that knowledge work, such as product
development, can be done most productively in a single location. For instance, Thompson’s (1967, pp. 54–61) early work
on structural contingency theory posited that activities like product development create reciprocal interdependence be-
tween individuals and subunits. Reciprocal interdependence is most effectively managed by locating individuals in close
0963-8687/$ - see front matter Ó2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jsis.2009.07.001
*Corresponding author. Tel.: +1 520 626 9842; fax: +1 520 621 8105.
E-mail addresses: gupta@arizona.edu (A. Gupta), elisa.mattarelli@unimore.it (E. Mattarelli), satwik@mit.edu (S. Seshasai), broschak@email.arizona.edu
(J. Broschak).
Journal of Strategic Information Systems 18 (2009) 147–161
Contents lists available at ScienceDirect
Journal of Strategic Information Systems
journal homepage: www.elsevier.com/locate/jsis
Author's personal copy
proximity to facilitate high levels of communication between them. Similarly, transaction cost theorists have suggested that
knowledge work, such as information technology development, is best performed internally due to concerns over the loss of
control over work (Loh and Venkatraman, 1995), high transaction costs (Ang and Straub, 1998), and threat of knowledge loss
(Duncan, 1998) when knowledge work is outsourced. Finally, studies on inter-personal communication have shown that
geographic distance reduces the opportunity for face-to-face interaction (Conrath, 1973), which is necessary for transferring
tacit knowledge between individuals and organizations (Espinosa et al., 2007; Kogut and Zander, 1992; Porter, 1998;
Tallman et al., 2004). Traditionally, physical distance was considered detrimental to inter-personal and inter-organizational
collaboration, which is why many firms in the 1980s and 1990s preferred co-locating large cross-functional teams at a single
site (Eppinger and Chitkara, 2006).
However, recent advances in information technology have enabled virtual distributed teams to perform knowledge work
effectively without meeting face-to-face (Cummings, 2004; Humphrey, 1995; Maznevski and Chudoba, 2000). By virtual
teams, we mean groups of workers who are geographically and temporally dispersed and are assembled via technology
to accomplish an organizational task (Jarvenpaa et al., 1998; Lipnack and Stamps, 1997). When virtual teams are based in
different countries, they are referred to as GDTs, i.e. globally distributed teams. The rich and vibrant body of research on
virtual teams and GDTs (Gibson and Cohen, 2003), and the increasing reliance of organizations on virtual teams in diverse
activities such as research and development laboratories (Brockhoff, 1998), IS development (Chakrabarty, 2006), and
software development (Carmel, 1999) suggest the potential for a new model of distributed knowledge production that
can leverage geographic distance for strategic advantage.
Over time, the view of offshoring as primarily a cost saving exercise has gradually transitioned to a perspective that views
offshoring as a mechanism for utilizing a globally distributed workforce in a new manner made possible by advances in
information systems (Venkatraman, 2004; Cullen et al., 2005; Walsham, 2005; Manning et al., 2008; Gupta, 2008). And while
this development may seem obvious to managers of organizations practicing the model described in this paper, the academic
literature contains many gaps in our knowledge about the functioning and performance of virtual teams in a distributed
knowledge work environment. For instance, while there has been a considerable amount of research on inter-personal issues
such as conflict, trust, and identity in virtual teams (Jarvenpaa et al., 1998; Jarvenpaa and Leidner, 1999.Montoya-Weiss
et al., 2001; Cramton and Hinds, 2005; Hinds and Mortensen, 2005), there has been less research on the use of tools and
methods in distributed teams when increased handoffs between team members exist; the same void exists for analyzing
the conditions under which the use of such tools can improve the effectiveness of the distributed model, and in understand-
ing how differently structured GDTs actually work (O’Leary and Cummings, 2007). In addition, one common criticism of
research on globally distributed teams is the lack of extended field experiments – conducted in commercial environments
– that have compared the behaviors of co-located and distributed teams and how these behaviors are related to the perfor-
mance of the distributed model. These issues are important for understanding the effective management of geographically
distributed teams: how can distributed teams work effectively with frequent transfer (handoffs) of work-in-progress with
each other?; how can subsets of team members work during daytime in their respective countries and still achieve
round-the-clock operation for the entire team?; and how effective can geographically distributed teams be in comparison
to collocated teams?
In this paper, we advance our knowledge of globally distributed teams by conducting a field study that compares the col-
laboration activities between members of a globally distributed team with the collaboration activities between co-located
team members performing a similar task. Here, we consider offshoring in a mutually beneficial perspective where the
interests of workers in high-income economies are aligned with workers in other countries and customers worldwide.
The research question that guides this research is: in a commercial setting how do distributed and co-located teams
performing the same task differ in their patterns of communication and knowledge sharing, and in their performance? In
investigating this issue, we use the productivity perspective on knowledge sharing in organizations proposed by Haas and
Hansen (2005, 2007).
Specifically, our focus will be on one type of globally distributed team, i.e., the 24-h knowledge factory model (Gupta and
Seshasai, 2007), which advocates continuous work on knowledge-based tasks by individuals located in time zones that allow
for 24-h engagement. Each individual in such work environments work the normal workday hours that pertain to his or her
local time zone, and then pass the task to fellow workers located in a different time zone.
Our setting is a case study of a two-site, global work environment (in contrast to a fully localized work arrangement); we
believe that the insights gained from examining this case study can serve as the basis for analyzing the characteristics of true
24-h knowledge factories that are rapidly evolving in different industries.
This paper is organized as follows: first, we introduce the concept of the 24-h knowledge factory; then we discuss the
productivity perspective on knowledge sharing and develop hypotheses. After describing the methodology followed to
conduct our longitudinal field experiment of two teams (one co-located and one distributed), we present our results and
conclude with a reflection on the theoretical and practical contributions of our study.
2. A specific type of globally distributed team: the 24-h knowledge factory
We begin by providing a definition of a knowledge factory. A knowledge factory is defined as a collection of knowledge-
driven workers tasked with producing a knowledge-based asset, with the workers frequently creating incremental assets
148 A. Gupta et al. /Journal of Strategic Information Systems 18 (2009) 147–161
Author's personal copy
that are handed off (i.e. passed back and forth) among fellow workers. A globally distributed call center is, in some ways, a
knowledge factory because when calls are handled, the knowledge pertaining to the particular call is stored centrally and is
available to the next individual who has to handle the same topic or the same caller. The software test and fix cycle envi-
ronment is another knowledge factory in which software is the knowledge-based asset, and the knowledge of whether
the software accomplishes its function is passed back and forth between software developers and testers. The 24-h aspect,
mentioned above, can be considered to be a manifestation of a knowledge factory where work is performed on a continuous
basis around the clock. This structure allows for tasks to be executed with faster turnaround time, which is one of the major
potential benefits of distributing work across time zones (Gupta, 2009b; Treinin and Miller-Frost, 2006; see also Eppinger
and Chitkara, 2006; Majchrzak et al., 2000).
Understanding the implications of spatial and temporal separations between workers on the overall performance of the
24-h knowledge factory paradigm requires looking at the historical precedents of this model as well as analyzing a number
of interrelated technical, strategic, organizational, and economic issues. The notion of shifts can be traced back to the indus-
trial revolution. Since installed manufacturing equipment was scarce and costly, different sets of employees were scheduled
to work in successive shifts so that the manufacturing facilities could be used on a round-the-clock basis. The use of the 8-h
shift system evolved over time. Initially, each worker was directed to work 12–16 h a day so that all machines could be used
for extended periods of time. Then, the notion of having two shifts evolved. Based on new legislation on both sides of the
Atlantic, the work hours were gradually reduced. The introduction of the shift system yielded benefits in terms of higher
productivity of each machine, reduced production times, and lower prices to customers. However, it also created social
and health issues by requiring people to work in an urban setting, usually away from other members of their families,
and also at odd hours and changing work schedules determined by the idiosyncrasies managers in charge of assigning work-
ers to different shifts.
Global workforces provide firms with access to high-talent designers; however, in the absence of the 24-h knowledge fac-
tory model, these workers would need to relocate to a different country, or work at odd hours of the night, often referred to
as the ‘‘graveyard shift”, in order to collaborate in real time with their globally distributed co-workers (Gupta, 2009a). His-
torically, observers from around the world deemed the time difference between globally distributed workers to be a major
impediment when implementing distributed information systems. Recently, the perception has switched around; for many
projects, the time difference is viewed as a strategic plus (Gupta, 2008; Gupta et al., 2007). However, both views of the effec-
tiveness of globally distributed teams are based on largely untested assumptions regarding the nature of work by co-located
and distributed teams and the feasibility of handing off tasks across shifts.
The knowledge factory examined in this paper is set in the computer software industry. Here R&D teams are character-
ized by a development cycle that relies heavily on sequential performance of specific functions, such as development, testing,
and verification. In a traditional software development environment, where all parties are located in the same geographic
location, a code developer typically waits until a fully functional portion of the product is available before passing it onto
an engineer to test it. However with the potential for receiving testing feedback overnight, the developer in a 24-h knowl-
edge factory model now has the unprecedented opportunity to build portions of the product on a daily basis (Treinin and
Miller-Frost, 2006). Examples of collaborative technologies that enable the 24-h knowledge factory in the software industry
are general technologies (e.g., emails) and software specific technologies, such as software problem reports system and
source code control systems.
Previous research has acknowledged that coordination and knowledge sharing across time and space during handoffs are
critical in the 24-h knowledge factory model (Gupta and Seshasai, 2007). This is perhaps why this approach is not yet in
widespread use (Espinosa and Carmel, 2003; Treinin and Miller-Frost, 2006). Most of the existing applications of the model
are in fact based on two shifts. Nevertheless, the potential importance of a full application of the model calls for more studies
about the challenges that the 24-h factory model poses and how to overcome them.
3. Communication, knowledge sharing, and performance: development of hypotheses
Effective knowledge sharing is considered essential for high performance in both co-located and distributed settings
(Cummings, 2004; Tagliaventi and Mattarelli, 2006; Kotlarsky et al., 2008). Haas and Hansen (2007) outlined two distinct
ways of sharing knowledge: through written documents that are made available in paper or in electronic format, and
through direct contact between individuals. Accordingly, two different knowledge management strategies can be applied
in organizational contexts: codification and personalization (Hansen et al., 1999). Teams that adopt a codification strategy
‘automate’ knowledge management; they make use of information and communication technologies to codify and store
knowledge into databases, with the objective of re-using codified knowledge in a ‘people-to-documents’ fashion. Teams that
adopt a personalization strategy rely on individual members to share knowledge and develop networks where tacit knowl-
edge can be shared on a person-to-person basis.
In two studies of co-located consultancy teams, Haas and Hansen (2005, 2007) introduced the productivity perspective on
knowledge sharing, based on the idea that different types of knowledge sharing affect task performance dimensions differ-
ently. For instance, they found that the level of quality of electronic documents’ used directly affects the time saved on tasks,
but only indirectly affects the quality of team output. At the same time, personal advice from external members favors the
quality of work but is not linked to the timely responses of the team.
A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161 149
Author's personal copy
In this paper, we adopt the productivity perspective framework and extend it to the case of globally distributed teams and
the 24-h knowledge factory. Specifically, for software development teams, the use of collaborative technologies (e.g., source
code system) is essential for their functioning in both collocated and distributed settings. At the same time, the different per-
ceptions and experiences of team members may induce ways of using the collaborative technologies differently from what
was originally expected. Users may adopt a collaborative technology, ignoring part of its properties or inventing new ones,
going the extra mile or contradicting the requirements of its original design (Orlikowski, 2000). Technologies, in fact, do not
exist in the abstract, but are manifest only when one introduces them into a social network, where they are necessarily sub-
ject to re-definition and re-structuring (Friedberg, 1993; Crozier and Friedberg, 1994).
Thus, even though in distributed settings the use of collaborative technologies is expected to be more intense (Lipnack
and Stamps, 1997; Maznevski and Chudoba, 2000), it must be recognized that some informal person-to-person practices
for sharing tacit knowledge will still emerge (Kotlarsky and Oshri, 2005; Oshri et al., 2007; Mattarelli and Gupta, 2009).
In other words, given the same set of communication technologies, we expect that co-located and distributed teams will de-
velop different strategies for sharing codified and tacit knowledge, as well as a different mix of ‘codification’ and ‘personal-
ization’ practices. Given this premise, we develop a comparative framework of distributed and co-located teams based on the
constructs depicted in Fig. 1.
3.1. Technology use and patterns of communication
There has long been a sense that face-to-face interaction can facilitate creative interaction and produce more and better
ideas (Osborn, 1957). However, there is an equally long history of experimental findings that show that the aggregate output
of so-called ‘‘nominal” or ‘‘concocted” groups of individuals working alone outstrips the aggregate output of ‘‘real” groups of
the same number of individuals working together in person on creative tasks such as idea generation (Lorge et al., 1958;
Mullen et al., 1991; Taylor et al., 1958). Real interactive groups consistently incur a ‘‘process loss” during group interaction
that nominal groups avoid (Steiner, 1972). The inability of all real group members to contribute their ideas simultaneously
can create a bottleneck that blocks potentially valuable contributions from some members and thereby reduces the effec-
tiveness of real groups (Diehl and Stroebe, 1987, 1991).
A number of researchers have noted that the use of communication technology can enhance the performance of both real
and nominal teams (e.g., members working more independently than collaboratively). For instance, the use of information
technology tools by real interactive groups can simultaneously enable creative production and removal of social inhibition,
thereby eliminating the production blocking problem (Paulus et al., 1996; McLeod et al., 1997). As a result, real groups can
sometimes be even more productive than nominal groups (Dennis and Valacich, 1993; Valacich et al., 1994). Globally dis-
tributed teams share key characteristics with nominal groups and also with electronic interacting groups (team member
interaction mediated by technology). The social psychology literature on small group dynamics implies that global virtual
teams may enjoy certain advantages relative to collocated teams (e.g., Kirkman et al., 2004), and with the aid of electronic
communication, the advantage of distributed teams over co-located interactive teams grows even further as group size in-
creases (Gallupe et al., 1992).
One issue that has not been suitably addressed by this literature is how co-located and distributed teams differ in their
use of technologies. In globally distributed teams, especially when time differences separate participants, the occasions for
synchronous communications to discuss task-relevant issues are reduced. Moreover, such teams ‘will be more likely to
transfer knowledge in explicit rather than tacit forms because the technology supports the declarative nature of explicit
knowledge’ (Griffith et al., 2003, p. 271). This means that, given a set of collaborative technologies provided to the team,
if a team is characterized by higher ‘virtualness’ (in terms of geographic and temporal separation), it will also rely more
on codified and ‘written’ forms of communication to discuss issues that are relevant to the completion of tasks. Thus, if
we compare a GDT with a co-localized team, we can hypothesize that:
Fig. 1. A framework for technology use, knowledge sharing, and team performance.
150 A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161
Author's personal copy
HP 1. The distributed team will rely more heavily, than the co-located team, on written communication for team discussion.
Among the different types of collaborative technologies, emails are probably the most widespread and diffused. In the
workplace, emails are used both for formal and informal communications. In the first case, emails represent a way to assess
the state of work, share formal documents, define meetings; in other words, they belong to the codification strategy de-
scribed above. In the second case, emails are used for quick and informal messages, in addition to other means of commu-
nication (e.g., face-to-face, instant messaging), coherently with a personalization strategy (Hansen et al., 1999). As such, we
expect that many messages of the second type will be sent when team members are co-located in the same time zone and
have other means of synchronous communication. On the other hand, when team members are globally distributed and sep-
arated by time and space, the former use is preferred and fewer emails are exchanged. Thus, we hypothesize that:
HP 2. The distributed team will rely less, than the co-located team, on broadcast style email messages.
Consistent with what was discussed above, we expect that distributed teams will use emails mainly as a codification
strategy. This means that emails will contain detailed discussions that team members would not be otherwise able to con-
duct and formalize, given the time difference across sites. Thus, it follows that:
HP 3. The distributed team will conduct longer discussions, than the co-located team, primarily in written (email) form.
Among the different types of content of emails, logistical messages are those related to a specific task or action to be com-
pleted in very short time (e.g., less than a week’s time). A logistical message is focused on logistics of a specific action and the
language is very focused, as opposed to a message that is trying to gather a broader set of opinions. Accordingly, logistical
messages can be interpreted as informal reminders that team members share in order to synchronize their pace, when they
work in the same time zone. But, when team members are separated, a schedule of activities and work is defined in advance
(Carmel, 1999) and thus, we expect that:
HP 4. The distributed team will send fewer logistical messages, than the co-located team, to members of the group.
3.2. Knowledge sharing
While electronic communication tools, such as email, allow distributed teams to work interactively (to some extent) and
productively on creative tasks, they do not resolve the challenge of tacit knowledge, which is considered to be essential to
innovative activities but is difficult to transfer without face-to-face interaction (Kogut and Zander, 1992; Nonaka et al., 2000;
Sternberg et al., 2000). The accessibility of ambient tacit knowledge has been posited as a major reason firms locate in close
geographic proximity to other organizations within the same industry (Audretsch and Stephan, 1996; Porter, 1998; Rosen-
feld, 1997; Tallman et al., 2004). If correct, globally distributed teams may be missing a key ingredient that would help them
function effectively, suggesting that co-located product development teams may be preferable after all.
The logic underlying the following hypotheses is based on the notion that a globally distributed team requires more hand-
offs of knowledge, and thus requires more formal systems to facilitate these handoffs (Mattarelli and Gupta, 2009). Accord-
ingly, the distributed team adapts the technical design and processes to reduce the number of interactions required. This has
an impact on the nature of discussions, the nature of tasks, and the nature of assigning technical modules, as described in the
hypotheses.
Among the technical design and processes used by software development teams, of particular relevance is the source code
modification process: a computer-based system for logging changes made to the computer programs being developed by the
team. When programming, team members must consider the ‘‘feature freeze” date, i.e. the deadline to complete program-
ming work, other required tasks, and all features within the given software release. Consistent with a codification perspec-
tive, we expect that distributed teams will make major use of the source code modification process when approaching the
deadline of the feature freeze date in order to translate tacit knowledge into easily sharable codified knowledge (Griffith
et al., 2003). For example, the individuals on the distributed team may each commit the source code changes they are
responsible for into the system before discussing with other individuals. On the other hand, members of the co-located team
would discuss a particular code modification before committing it to the project. In other words, co-located team members
will rely more on informal contacts and discussions in order to share tacit knowledge about the product. Thus, we expect:
HP 5. The distributed team, as compared with the co-located team, will make greater use of the source code modification
process to resolve issues, in place of informal collaboration, before the ‘feature freeze’ date.
When working on software development, teams are assigned different modules. A team may decide to have a single per-
son to take care of each module or to have multiple individuals working together on each module. In the latter case, the so-
cio-technical system is more interconnected, because two or more individuals associated with the same module must share
and build tacit and codified knowledge through repeated interactions. Consistent with a codification strategy, and with the
perspective that members of virtual teams are less likely to acquire tacit knowledge from their distant teammates (Griffith
et al., 2003), we expect that:
HP 6. The socio-technical system of the distributed team will be less interconnected as compared to the co-located team.
A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161 151
Author's personal copy
Knowledge sharing among team members also occurs through the mechanism of meetings. Periodic meetings are consid-
ered to be fundamental for the proper functioning of co-located and distributed teams (Hackman, 1990; Kiesler and Cum-
mings, 2002; Ganesan et al., 2005). Meetings can be face-to-face, through videoconference, or by phone. They can deal
with long-term strategic decisions, such as the technical architecture of the product or with short term tactical issues, such
as the discussion of the specific content of work tasks (e.g., if a piece of coding is better done in one way versus another).
Logically, co-located teams have the opportunity to discuss day-to-day tactical decisions in informal ways, for instance
through a chat in the hallway or in front of the coffee machine. This is not possible for globally distributed teams separated
by physical distance and time. Thus, in the latter case, formal meetings become necessary for both strategic and tactical deci-
sions and the latter become the most frequent rationale. We can hypothesis that:
HP 7. The distributed team will rely more, than the co-located team, on meetings for handling short term issues.
Also, meetings can be organized to assign tasks to team members according to their expertise (Hackman, 1990). We ex-
pect that distributed teams will make more use of meetings to formalize task assignments, consistent with HP 7. On the
other hand, task assignment is done mainly informally and through face-to-face coordination in co-located settings.
HP 8. The distributed team is more likely than the co-located team to formally assign tasks to team members in meeting format.
Finally, when it comes to the overall knowledge management strategy, the above discussion suggests that distributed
teams will establish a codification strategy. In other words, given the same set of collaborative technologies, distributed
teams will rely more on relevant technologies to codify knowledge and make this knowledge available to all team members.
HP 9. The distributed team will rely more on formal systems for knowledge capture, as compared to the co-located team.
3.3. Performance
Previous studies provide contradictory advantages and disadvantages for distributed and co-located teams, making it dif-
ficult to formulate general predictions about how each type of team would perform on a similar product development task.
But, in the case of a team that has the time to develop trust and acquaintance of working together (e.g., a software devel-
opment team whose members work together on a project for 1 year), it has been found that some of the limits of distribution
are overcome and that a distributed team can produce an output similar to that of the traditional co-located team (Dennis
and Garfield, 2003; Espinosa et al., 2007). In other words for long-term teams, the following hypothesis is likely:
HP 10. The output of the distributed team will be similar, in terms of quality, as that of the co-located team.
However, several studies have shown that the overall efficiency, generally defined as a measure of output from produc-
tion processes per unit of input, of a distributed team is lower than that of a co-located one (e.g., Montoya-Weiss et al., 2001;
Powell et al., 2004; Hightower et al., 2007). In the case of a team working on a 24-h knowledge factory basis, efficiency is
reduced by the overhead involved in transferring tasks back and forth on an incremental basis. Thus, we hypothesize that:
HP 11. The efficiency of the distributed team will be lower than that of the co-located team.
4. Method
While several studies have investigated some of the differences between co-located and distributed work in laboratory
settings, limited empirical evidence has been collected in real world settings, especially when teams are globally distributed
(McGrath, 1991; Montoya-Weiss et al., 2001; Massey et al., 2003; Martins et al., 2004). Moreover, extant work tends to treat
all virtual teams alike (Bell and Kozlowski, 2002), while in practice, virtual teams may differ significantly from one another in
terms of their structure, duration, and tasks (Saunders and Ahuja, 2006; O’Leary and Cummings, 2007). On a related note,
significant empirical research has been performed using cross-sectional surveys of hundreds of teams (e.g., Cummings,
2004). A notable exception to the mainstream literature on virtual teams is provided by the exemplar work of Majchrzak
and colleagues (2000), who studied a product development team that was distributed across three organizations in different
locations and collected longitudinal ethnographic and quantitative data to develop an in-depth understanding of the oper-
ation of the team from social, organizational, and technical perspectives.
Inspired by this type of in-depth data collection strategy, a longitudinal study of co-located and distributed software
development teams was conducted at IBM Corporation. We compared two teams within a single firm and we manipulated
the key variable of organizational structure (geographic distribution). One team was entirely based in Boston, MA, while a
second team was distributed between Boston and Bangalore (India). In contrast to descriptive case studies, the present study
is a controlled field experiment that compares two teams with nearly identical characteristics except for the critical variable
of interest: co-location versus geographic distribution of team members. The design is a ‘‘quasi-experiment” (Cook and
Campbell, 1979) in the sense that team members were not randomly assigned to each type of team, but the twin features
of similar composition of team and exercise of controls for other possible explanatory factors allowed us to infer that the
difference in the structure of the two teams was the basis for observed differences in team performance.
152 A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161
Author's personal copy
In the following sub-sections, we describe the characteristics of the two teams, the data sources used, and the data anal-
ysis process followed, and the measures used to test the hypotheses.
4.1. Team characteristics
Both teams belonged to IBM and worked within the department responsible for building new collaborative software. The
two teams worked on two parts of the same software package, with one team producing a document management product
and the other producing a team collaboration product. The two teams were subject to identical time schedules (12 months)
and deadlines, and were under the same environment in terms of project management, resources, and work rules. Teams
were provided with the same collaborative technologies: an email system, an instant messaging system, and the same pro-
cesses for managing tasks and source code.
Each team was assigned seven members (three in the US and four in India for the distributed team) with similar positions,
qualifications and experience. Of the seven team members, one was the lead, six were developers, each with 5–20 years
experience and seniority. The average professional and organizational tenures for the co-located and distributed teams were
both approximately 10 years.
All the co-located team members worked on the software during the same work hours, whereas the globally distributed
team members shifted work back and forth across time zones in an asynchronous manner. The two sequential work shifts of
the distributed team provide less coverage than the three consecutive 8-h work shifts in the ideal 24-h knowledge factory
model, but dispersion of the team across 10 time zones forced team members to work more independently during their
respective shifts, providing a conservative test of the key feature of the model.
4.2. Data sources and analysis
Quantitative and qualitative data were collected systematically from the two teams over a period of 1 full calendar year.
Since the main project deliverable was on a 1-year timeframe, this period covered every major point in the project lifecycle
from the kick-off to the delivery of the end product. Within this year, the teams devoted a significant amount of time to short-
term tasks such as attending to customer deployment issues and fixing bugs for maintenance releases; as such, the 1-year
timeframe provided an opportunity to gain insights on knowledge sharing for multiple scopes and varieties of tasks. The data
collection process was designed to provide a complete picture of the knowledge sharing that occurred over the 1-year period
in terms of technical, organizational, social, and group process dimensions. The experimental design and quantitative mea-
sures enabled direct comparisons between the co-located and distributed groups on the key dimensions of interest. The data
sources employed were: interviews, observations of weekly meetings, and archival data. They are described below.
4.2.1. Interviews
Two hour-long structured interviews (Gubrium and Holstein, 2001) were conducted with each of the developers on each
team. While the focus of these interviews was primarily to gain qualitative insight on work content, specific quantitative
questions were asked in order to elicit the developers’ own views of their knowledge sharing requirements. In particular,
interviewees were asked to elucidate about and provide the number of: informal interactions (i.e., interactions that did
not begin with an intention of discussing business) with fellow team members and with main developers; formal interac-
tions with main developers; informal communication in person, via instant messaging, and via phone; tactical decisions
made informally (decisions that were minor in scope, with minimal knowledge sharing requirements and minimal impact
on other developers’ work); strategic decisions made informally (decisions that were major in scope, with significant knowl-
edge sharing requirements and long-term impact on other developers’ work); strategic decisions that were speeded up infor-
mally; and tactical decisions that were speeded up informally. Interviews were transcribed into files and inductively coded
to obtain the quantitative measures described below.
4.2.2. Observations of weekly meetings
The weekly meetings of each team were observed (in Boston) by one of the authors to gain insights into the processes of
formal task allocation and knowledge sharing, on a group-wide basis, for each team. The teams organized three meetings per
week (one meeting for co-located team, one meeting for US team members of the distributed US–India team, and one US–
India team joint session). The subgroup in India did not hold formal meetings, because the project manager was a US-based
employee and only ran meetings which involved US-based employees in the US time zone. The Indian subgroup members sat
next to each other, and would often discuss items, but did not have a formal meeting.
The minutes were recorded by the project manager for the teams, who maintained item-by-item details of the discussion
and shared them with the researchers. The co-located team held one face-to-face team-wide meeting per week, while the
distributed team held 1 weekly face-to-face meeting for only the US-based team members and 1 weekly coordination meet-
ing via telephone between the development leads from the US and India.
Minutes were inductively coded to obtain quantitative measures such as: the number of Tactical Tasks Assigned, the
number of Strategic Tasks Assigned, the number of Tactical Status Requests, the number of Strategic Status Requests. Tasks
are defined as future actions required of a team member while status requests are queries on past actions. Tactical refers to a
very specific scope with a definitive action, and strategic refers to a more broadly scoped question without a specific action.
A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161 153
Author's personal copy
4.2.3. Archival data
Three types of archival data were used. First, each development team kept track of fixes requested or made to the code
base via Software Problem Reports (SPRs). These SPRs contained information on the problem being reported, as well as the
history of knowledge provided by various developers in resolving the issue and information regarding the actual fix to the
issue. SPRs were stored in a central database for each team. Modifications to SPR states were performed according to a for-
malized process. The formalized process of storing, tracking, and transforming SPRs, from their creation to problem resolu-
tion, constituted a formalized knowledge capture system. For purposes of this study, a software tool was written to collect
the data from the SPR archive. This tool analyzed the software problems that were resolved over the 12-month period of
study, and collected the specific types of data (e.g., average delay between developer inputs) for each developer, on a weekly
basis. The measures used in the analysis are described in the next sub-section.
Second, each of the two teams used a source control system to log the modifications made to each element of the source
code for the team’s product. The source control system stored the date, time, developer making the change, and a comment
regarding the particular change. The comments often cited particular SPRs if there was an SPR that initiated the particular
change to be made. The goal of collecting data from the source control system was to ensure a clear depiction of the technical
system, which would complement the social and organizational systems described by the other forms of data that were col-
lected. The data from the source control system provided a representation of the technical dependencies between developers
on the teams, and the rate of technical collaboration within the teams. Different data were collected, with respect to each
developer, on a weekly basis (e.g., Delay between check-ins). The measures used in the analysis are described in next
sub-section.
Finally, a software tool was written to analyze email messages sent to all members of each team. A ‘‘thread” refers to the
entire set of messages written in response to an initial electronic broadcast or request for information. These data provided
insights into the use of broadcast messages to share knowledge on the teams. Different data were collected, with respect to
each developer, on a weekly basis (e.g., number of threads contributed to). Moreover, email messages were coded, according
to their content, into logistical and non-logistical messages.
4.3. Measures
The measures of the variables used to test the hypotheses of this study (derived from the data sources described above)
are shown in column 2 of Table 1. Specifically, data derived from observations of meetings, and the hypotheses to which they
applied, were as follows:
Table 1
Comparison of outcomes for key process variables.
Hypothesis Process variable Distributed
team
Collocated
team
tdf t-Test
(p< 0.05)
Mean SD Mean SD
HP 1: The distributed team will rely more heavily, than the co-
located team, on written communication for team discussion
Contributors per email
thread
1.73 1.55 1.50 0.74 0.94 12 Inconclusive
HP 2: The distributed team will rely less, than the co-located
team, on broadcast style email messages
Average weekly email
threads
10.42 5.05 19.85 10.75 5.56 12 Confirmed
HP 3: The distributed team will conduct longer discussions, than
the co-located team, primarily in written (email) form
Average emails per thread 2.32 2.25 1.75 0.95 1.63 12 Inconclusive
HP 4: The distributed team will send fewer logistical messages,
than the co-located team, to members of the group
Average logistical weekly
emails
17.06 10.13 29.91 19.55 4.09 12 Confirmed
HP 5: The distributed team, as compared with the co-located
team, will make greater use of the source code modification
process to resolve issues, in place of informal collaboration,
before the ‘feature freeze’ date
Source code check-ins
prior to deadline
53.82 74.56 11.56 11.0 3.93 12 Confirmed
HP 6: The socio-technical system of the distributed team will be
less interconnected as compared to the co-located team
Average number of
developers per code
element
1.10 0.2 1.63 1.04 3.50 12 Confirmed
HP 7: The distributed team will rely more, than the co-located
team, on meetings for handling short term issues
Fraction of tactical (versus
strategic) meeting items
0.81 0.17 0.39 0.22 10.57 12 Confirmed
HP 8: The distributed team is more likely than the co-located
team to formally assign tasks to team members in meeting
format
Percent of task assignment
(versus status) meeting
agenda items
0.35 0.13 0.24 0.17 3.60 12 Confirmed
HP 9: The distributed team will rely more on formal systems for
knowledge capture, as compared to the co-located team
Average # of individuals
modifying SPR state
3.25 0.97 1.74 0.34 10.28 12 Confirmed
HP 10: The output of the distributed team will be similar, in
terms of quality, as that of the co-located team
Average SPR actions per
week
134.21 168.3 104.37 152.39 0.92 12 Confirmed
HP 11: The efficiency of the distributed team will be lower than
that of the co-located team
Average SPR time to
resolution
113.80 83.17 120.72 130.45 0.31 12 Inconclusive
154 A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161
Author's personal copy
the fraction of tactical (versus strategic) meeting items as a proxy of the discussion of short term issues during meetings
(HP 7);
the percent of task assignment (versus status) meeting agenda items as a proxy of the use of meetings for task assignment
(HP 8).
From SPR data, the following were derived:
the number of source code check-ins prior to deadline in week 41 as a proxy of use of source code modification processes
(HP 5);
the average SPR time to resolution, that is the average time it takes for an SPR to move from being approved by the man-
agement team to being fixed, and finally to actually being logged as fixed, as a proxy of team efficiency (HP 11).
From source control system data, we derived:
the number of individuals working on each module as a proxy of the interconnection of the socio-technical systems (HP
6);
the average number of individuals modifying SPR state – the number of individuals modifying SPR state corresponds to
the number of people who were directly involved in resolving a particular problem relying on the formal system of knowl-
edge capture; thus, it is a proxy of reliance on formal systems of knowledge capture (HP 9);
the average number of SPR actions per week – the larger the number the SPR actions, the greater is the number of software
problems that were reported. Hence it is a proxy for output quality (HP 10); actually, the number of SPR is inversely
related to software quality.
Finally, from the analysis of email messages, we derived:
the number of contributions per email thread, as a proxy of amount of written communication (HP 1);
the average weekly email thread initiated, as a proxy of the amount of broadcast style email messages (HP 2);
the average length of initiated threads as a proxy of length of discussions in written form (HP 3);
the average number of logistical weekly emails as a proxy of intensity of logistical messages (HP 4).
5. Results
5.1. Quantitative analysis
Comparisons of outcomes for the key process variables for the distributed and collocated teams are presented in Table 1,
based on the set of 11 hypotheses formulated earlier in the paper. The table contains means and standard deviations of each
observed variable. Additionally, a t-test was used to compare means across groups and validate the formulated hypotheses.
No statistical difference was found for HP 1; the two teams did not differ in terms of the number of contributors per email
thread. This may be explained by the small size of the team (seven members) and by the similar division of labor across the
two teams.
However, consistent with HP 2 and 4, our data revealed that the number of email threads initiated and the number of
logistical emails sent per week are significantly larger for the co-located team than for distributed team. In other words,
the co-located team communicated more frequently via email messages than did members of the globally distributed team;
this was despite the fact that many of the co-located team members worked in the same hallway of the same building.
Though it appears that the number of email threads initiated is larger for co-located teams, it should be noted that the num-
ber of emails exchanged within each initiated thread does not change significantly across the two teams (thus disproving HP
3).
The two teams differed most dramatically in the number of source code modifications prior to the ‘‘feature freeze” dead-
line in week 41, with the distributed team making 53.8 modifications compared to only 11.6 modifications by the co-located
team (t= 3,93, p< 0,05). In other words, the co-located team was able to approach a key product development deadline with
much fewer last-minute changes, and its work on the software code involved more person-to-person consultation than the
work by the globally distributed team. This supports HP 5.
On average, there were more developers per code element for the co-located team, as compared to the distributed team,
thereby supporting HP 6. This is consistent with the consideration that the socio-technical system of the distributed team is
less interconnected than the co-located one.
Consistent with HP 7–9 which were all validated by the case study, the two teams used team meetings for very different
purposes; the meetings of the distributed team featured a significantly higher percentage of tactical (cf. strategic) agenda
items (HP 7)and also a much higher percentage of assignment items (cf. status items, HP 8). Overall, the distributed team
relied more on formal systems for knowledge capture, as evidenced by the intensity of use of SPR (HP 9).
A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161 155
Author's personal copy
We found support for HP 10, but not for HP 11. In other words, both the output quality and efficiency (measured by
weekly SPR actions and average time to resolve SPR’s, respectively) of the two teams were similar. This means that, despite
the very different usage of information systems and meeting behaviors, each team exhibited similar performance in terms of
the quality and speed of their work. In the next section, we triangulate the quantitative results with our qualitative evidence.
5.2. Qualitative analysis
Qualitative data from interviews and meetings helped us in understanding more deeply the context under study, in tri-
angulating the evidence from the quantitative data, and in adding insights on the underlying processes. In the following
paragraphs, we present the perceptions of our informants on (i) technology use and knowledge sharing; (ii) the link between
social relationship and technical behavior; (iii) the outputs of distributed and co-located teams; and (iv) the major advan-
tages of geographic distribution.
5.2.1. Technology use and knowledge sharing processes are enacted differently
Our quantitative evidence shows that, while the two teams were provided with the same technologies, their use and the
consequent knowledge sharing processes enacted very differently. For instance, the distributed team was more parsimoni-
ous in the use of emails. During interviews, members of the globally distributed team confirmed that they used emails for
specific large scale purposes, and not to address short term issues related to the advancement of work. In other words, GDT
members used email messages as part of the codification strategy (Hansen et al., 1999) and to extend the knowledge capture
processes guaranteed by the SPR. On the other hand, co-located team members used email messages in a more informal fash-
ion, even with few lines and short logistical questions. In other words, email messages were perceived to be a continuation of
face-to-face interactions.
It is also evident that the distributed team relied more on the source code system to manage the modification of software
and on formal knowledge capture systems to codify all the knowledge produced by the team. Specifically, distributed team
members affirmed that they used this technology as a means of transferring information and knowledge between team
members and maintaining a record of status, while the co-located team members affirmed that they could rely on synchro-
nous communication for purposes of information sharing and status reporting. Interviews with GDT members also high-
lighted that the use of formal systems increased individual confidence in team results and improved the management of
physical and temporal distance across team members.
Also, the nature of the meetings differed in the two cases. An analysis of the minutes of the meetings revealed that while
the agenda categories were generally the same between meetings, the number of tactical items and number of task assign-
ments were much higher in the case of the distributed team. In other words, meetings were used by distributed members to
keep updated on individual work details and to redefine the workload assigned to each developer (see also Orlikowski and
Yates, 2002). These two issues were not perceived as being of primary importance by co-located members, who could infor-
mally discuss such issues in the hallway or with a word over the cubicle, outside of the context of the formal weekly
meetings.
Overall, these differences suggest that technology and processes that support knowledge sharing can be used to explicitly
serve different purposes (cf. Maznevski and Chudoba, 2000; Haas and Hansen, 2007). Barley (1986) provides a framework for
assessing the role of technology in a knowledge-based work environment and suggests that the context in which the work is
performed can significantly impact the way the technology is used. Teams will gradually adapt available technologies to suit
their specific spatial and temporal structures.
5.2.2. Social relationships and technical behavior are linked
HP 6 was confirmed, highlighting that the co-located team had more examples of code elements that were modified by
multiple team members; interviews confirmed that this was because of the greater degree of social interaction on this team,
rather than any piece of software requiring more intertwined technical interaction than the other. The interview sessions
also revealed many cases where casual interactions led to technical decisions. For instance, one of the developers stated:
‘‘While such social relationships are much easier to form when the team is co-located, the experience of one US developer
on the distributed team who traveled to India suggests that social relationships can be built across distant geographic and
cultural boundaries and these relationships can be leveraged to satisfy technical goals”.
Based on the above, the degree of social interaction between developers on a team was shown to have an impact on the
technical behavior of the team, which then led to tighter social relationships. Developers on both teams cited the comfort
level between team members as being important in facilitating creative discussions, so that developers did not have to worry
about feeling embarrassed by a poor idea.
5.2.3. Geographic structure does not define output
The geographic structure of the teams in this study led to different forms of value being achieved from their knowledge
sharing processes; however, it does not follow that the output of each team is necessarily defined by its structure. The struc-
ture of the distributed team led its members to have a higher degree of documented decisions. Interviews with members of
the distributed team confirmed that a very valuable, though perhaps unintended, outcome of this documentation process
was that the history of the decision making process of the team was better retained. On the other hand, the co-located team
156 A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161
Author's personal copy
cited more frequent informal communications as a process that led to higher incidence of finding new and creative solutions.
Even though these informal meetings generally occurred face-to-face, the distributed team could still achieve a similar out-
come. During interviews on this specific topic, distributed team members mentioned the importance of a one-time face-to-
face meeting that would introduce team members and incorporate a social component to the relationships, and the use of
explicitly informal phone calls where no agenda or topic was preplanned so that team members could discuss any open-
ended topic.
It is worth noting that co-located team members did not let unacknowledged the importance of documented processes,
especially those related to decision making. During interviews on this specific topic, co-located team members mentioned on
the one hand the importance of scheduling some midpoints in which each team member should put effort in codification
and, on the other hand, the usefulness of automated tools to facilitate such processes.
5.2.4. Advantages of geographic distribution
Based on the analysis of the data from the interviews, Table 2 summarizes the major advantages of globally distributed
and co-located teams. Our finding that both co-located and geographically distributed teams were capable of successful col-
laboration suggests that common themes in the literatures on offshoring (offshoring is a win–lose zero-sum proposition),
innovation (geographic distribution is a barrier to overcome) and social and organizational psychology (face-to-face groups
are more productive) may all be inaccurate. Numerous benefits from leveraging a dispersed geographic structure were cited
in interviews with the distributed team. Examples include: an increase in documentation and history retention; enhanced
ability to share short-term tasks which required immediate attention so that work could be performed around the clock;
and a more structured and explicit definition of work tasks and distribution of work items.
6. Discussion
6.1. A productivity perspective on knowledge sharing in globally distributed teams
This study was aimed at enhancing our understanding of the differences between co-located and distributed teams. We
have proposed a set of hypothesis on the use of collaborative technologies, knowledge sharing processes, and performance.
We conducted a quasi-experiment in IBM and collected both quantitative and qualitative data in order to compare the per-
formance of a distributed team working around the clock as a knowledge factory with the performance of a traditional co-
located team. The two teams we studied were composed of the same number of individuals, were provided with the same
technologies, were engaged in similar tasks, and were given similar deadlines.
We found support for 8 of the 11 hypotheses we formulated. Specifically, as regard to technology use, while both team
members relied heavily on written communication for group discussion and engaged in written discussions of similar length
(disconfirming HP 1 and 3), we found that the distributed team sent a smaller number of broadcast style email messages and
fewer logistical messages than the collocated team (supporting HP 2 and 4). In other words, while email messages were
deemed to be essential in both cases, they were interpreted as a mechanism to share broad information by distributed team
members, versus as a continuation of their informal interactions by co-located team members.
Moreover, the distributed team made greater use of documentation processes for knowledge sharing. Specifically, distrib-
uted team members used the formal source codification process to share knowledge and resolve issues prior to the feature
freeze date (HP 5), and relied on formal systems for knowledge capture (HP 9). On the other hand, co-located teams relied
heavily on informal face-to-face interactions to share knowledge, and tended not to document decision making processes.
This may also explain why, on average, more than one person worked on each code element; in other words, the socio-tech-
nical system of the co-located team was more interconnected (HP 6).
The scope of meetings varied greatly across the two teams. The distributed teams used meetings for short term issues and
to assign tasks (HP 7 and 8). On the other hand, the co-located team conducted strategic meetings, more similar to brain-
storming sessions, to discuss the status of the overall coding process and the future directions to take. Co-located team mem-
bers had the opportunity to discuss short term issues and task assignments informally face-to-face and did not need to
document such decisions.
The strategies followed by the two teams to share information and knowledge resemble the two knowledge management
strategies described by Hansen et al. (1999): a codification strategy for the globally distributed team, based on documented
Table 2
The major advantages of each type of team mentioned by informants during interviews.
Distributed team Co-located team
Use of collaborative technologies Exploiting technology for collaboration Using technologies as an addition to informal, face-to-face, interactions
Processes and interactions Structured use of formal processes Incidental interaction that leads to efficiency
Meetings Meetings focused on role and tasks definition Meetings focused on strategic discussion
Knowledge sharing Formal logging of knowledge Issues resolved informally, in a timely manner
A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161 157
Author's personal copy
decisions and a personalization strategy for the co-located team, based on informal communication (see Fig. 2). Differently
from the knowledge sharing strategies of Hansen et al. (1999), which are planned by top management and need to reflect the
overall strategy of the firm in order to be successful, the strategies for knowledge sharing and technology use that we have
just described were emergent and triggered by geographical and temporal distribution. Further, the difference in knowledge
sharing strategies does not depend on the availability of a certain technology, but on the actual use and interpretation of that
technology (Orlikowski, 2000).
Haas and Hansen (2005, 2007), in their productivity perspective on knowledge sharing, pointed out that different ways of
sharing knowledge bring about different outcomes. Our qualitative data support this perspective and operationalize it for a
different context (distributed versus co-located teams). Specifically, while documented decisions are associated with the
possibility of retaining history and re-using knowledge in a timely fashion, informal communication is associated with cre-
ativity (see Fig. 2).
Our data also show that there is not a statistically significant difference in the overall efficiency and quality across the two
teams (HP 10 and 11). As far as quality is concerned, previous literature on distributed teams has already shown that dis-
tributed teams members that are together over time attain the same level of quality outcomes as that of their co-located
counterparts (e.g., Dennis and Garfield, 2003). On the contrary, as far as efficiency is concerned, preliminary evidence on dis-
tributed versus co-located teams seems to suggest that co-located teams outperform distributed teams, especially for dis-
tributed teams in extreme situations, such as those characterized by a high time and space separation (e.g., Powell et al.,
2004). Our evidence, instead, shows that both teams attained the same efficiency level.
A possible explanation for the similarity of performance can be found in Fig. 2. Our qualitative data show that distributed
team members did not strictly adhere to a pure codification strategy, but grafted elements of personalization through the
introduction of a face-to-face meeting and informal phone calls. Such elements improved their ability to develop new
and creative solutions. At the same time, co-located team members grafted seeds of codification into their personalization
strategy with scheduled documentation times and the use of automated documentation tools. Such grafted strategies seem
to level off team performance.
6.2. Managerial implications
This study supports the contention that offshore decentralization of knowledge intensive work, such as software or infor-
mation systems development, can succeed with proper design and management of the dispersed team, and use of appropri-
ate collaborative technologies. The collaborative systems can facilitate effective group interaction while preserving some
advantages enjoyed by ‘‘nominal” groups of individual team members working independently.
This study also shows that the geographic structure of a team (co-located or globally distributed) does not predetermine
team outcomes. Neither structure is inherently superior; both are workable models with proper adaptations. The results also
indicate that geographic distribution can be leveraged by taking advantage of the possibility of continuous engagement on
tasks across time zones.
At the outset, we referred to the 24-h knowledge factory model as the evolving model for leveraging geographic and tem-
poral differences. Over time, this notion has been applied to applications of greater sophistication and with less inherent
structure, by placing greater reliance on technologies to provide the necessary collaboration for handling semi-structured
Fig. 2. A productivity perspective on knowledge sharing in globally distributed teams: a field model.
158 A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161
Author's personal copy
work. Two kinds of environments are especially relevant to the information systems community. One is the design, devel-
opment, and implementation of information systems in a manner that leverages the distributed workforce paradigm; this is
already happening. The other is the development of new information system approaches that will enable this paradigm to be
applied to a broad range of white-collar activities ranging from medical services to logistics planning, and from financial
analysis to product design; this is where the greater challenge and opportunity lie. While we considered the scenario of dis-
tributed and sequential software development, the same principles could be applied to perform distributed product design
and development work (as General Motors is doing), to create new marketing plans, to analyze data from accounting and
auditing perspectives, to mine information from customers, and to conduct long-term research in medicine, biotechnology,
and other fields. The challenge in each of these areas lies in being able to take traditional tasks and decompose them into a
series of components, just as what happens in the case of large IT endeavors. This ‘‘commodity-based” approach allows dif-
ferent people to perform the mini-tasks. Further, when tasks are modular in nature, natural breaks can serve as good hand-
off points.
From a managerial perspective, the concept of 24-h knowledge factory raises several new issues. Should the work be per-
formed exclusively on a peer-to-peer basis, or should the manager get involved? Should the pay for the workers in the dif-
ferent countries e the same as they are performing very similar functions, or should it be different to reflect the dissimilar
cost-of-living statistics in the respective geographic settings? Should the manager be accessible around the clock in case of
emergencies, or should the management function itself be transformed to a set of three managers, for each work in shifts of
8-h? If the latter concept is accepted, how far up in the organizational hierarchy should this concept go? As an extreme case,
should it apply to the corporate CEO too? Many of these emerging managerial challenges are currently being handled on a
case-by-case basis, based on the type of the organization, the type of the professional work involved, and the specific choice
of the three locations.
6.3. Limitations and future research directions
This research is characterized by several limitations. First, it was conducted in a single organization and with a limited
number of respondents. Even though our evidence may not be generalized to other settings, the access that we were able
to gain in this context enabled us to collect quantitative and qualitative evidence and to create and analyze a detailed picture
of technology use and knowledge sharing processes in co-located and distributed teams. In addition, the subjects we studied
are software professionals, who are similar, for many aspects of their work, to workers in many types of IT knowledge-based
industries.
Second, we acknowledge that several of the problems and overheads for the distributed team occur because there of the
need for frequent handoffs between people. If the individuals were not in distributed locations, but still had the same num-
ber of handoffs, we suspect that many of the same characteristics would have been observed. However, the distributed team
is the primary instance where knowledge-based work will involve repeated handoffs and thus was chosen to be one of the
key foundations of our quasi-experiment.
Third, we only investigated processes related to technology use and internal knowledge sharing, and did not investigate
other social processes, such as external knowledge sharing, subgroup dynamics, conflict, and trust. We do not know how
these emerging processes may influence the experimental results. For instance, as far as subgroup dynamics are concerned,
we noticed that the subgroup members in India sat next to each other, and would often discuss items but did not have a
formal meeting. This in itself is an interesting fact, which likely did have some effect on the team’s functioning. However
that was not studied as part of this project.
Future research directions can be framed in light of these limitations. Future studies could compare distributed and co-
located teams in other settings and explore if the model we propose in Fig. 2 still holds; further, the model could be ex-
panded to incorporate other social processes. For example, the study of Cummings (2004) focused on external knowledge
sharing as opposed to intra-team knowledge sharing and linked structural diversity to a higher degree of external knowledge
sharing. Our study focused primarily on internal knowledge sharing, but reached a similar conclusion that having structural
diversity does lead to a change in knowledge sharing practices and knowledge re-use. An extension to the present study
could involve the distinction between internal and external knowledge sharing.
7. Conclusions
This paper described the potential characteristics of the 24-h knowledge factory that utilizes multiple collaborating cen-
ters located at carefully selected time zones that are operational during daytimes in their respective countries. The efficacy of
such a work environment was evaluated by creating a set of 11 hypotheses that were tested in a controlled field experiment
involving one co-located team and one distributed team, characterized by similar composition, tasks, and collaborative tech-
nologies. The results show that the two teams differed in their use of technologies and in knowledge sharing processes, but
not in efficiency and quality of outcomes. The co-located team and distributed team enacted a codification strategy and a
personalization strategy, respectively; in each case, they grafted elements of the other strategy in order to attain both knowl-
edge re-use and creativity.
A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161 159
Author's personal copy
This work contributed to the literature on knowledge sharing in distributed teams, expanding the framework of Haas and
Hansen (2005) that was previously developed at the organizational level. Moreover, it offered a unique comparison of a co-
localized team and a globally distributed team in a real setting. To the best of our knowledge, no previous studies have pre-
sented quasi-experiments with these aims and characteristics. We also attempted to contribute to managerial practice, by
offering suggestions to managers and organizations that are interested in developing and deploying the 24-h knowledge fac-
tory model and taking greater advantage of a globally distributed knowledge workforce. Our results suggest that the intro-
duction of spatial and temporal separations between workers implies a corresponding introduction of new challenges; these
can be overcome – and even leveraged – for strategic advantage.
Acknowledgement
The authors acknowledge, with sincere thanks, the help provided by several colleagues and students, especially Igor Crk
and Curtis Prendergast.
References
Ang, S., Straub, D.W., 1998. Production and transaction economies and IS outsourcing: a study of the US banking industry. MIS Quarterly 22 (4), 535–552.
Asprey, W., Mayadas, F., Vardi, M., 2006. Globalization and the Offshoring of Services: A Report of the ACM Job Migration Task Force. Association of
Computing Machinery Report.
Audretsch, D.B., Stephan, P.E., 1996. Company-scientist location links: the case of biotechnology. American Economic Review 86 (June), 641–652.
Barley, S.R., 1986. Technology as an occasion for structuring: evidence from observations of CT scanners and the social order of radiology departments.
Administrative Science Quarterly 31 (1), 78–108.
Bell, B.S., Kozlowski, S.W., 2002. A typology of virtual teams. implications for effective leadership. Group and Organization Management 27 (1), 14–49.
Brockhoff, K., 1998. Internationalization of Research and Development. Springer, Berlin.
Carmel, E., 1999. Global Software Teams: Collaborating Across Borders and Time Zones. Prentice Hall PTR, Upper Saddle River, NJ.
Chakrabarty, S., 2006. Real life case studies of offshore outsourced IS projects: analysis of issues and socio-economic paradigms. In: Kehal, H.S., Singh, V.P.
(Eds.), Outsourcing and Offshoring in the 21st Century: A Socio-Economic Perspective. IGI Publishing, Hershey, PA, pp. 48–301.
Conrath, D.W., 1973. Communication environment and its relationship to organizational structure. Management Science 20, 586–603.
Cook, T.D., Campbell, D.T., 1979. Quasi-Experimentation: Design and Analysis Issues for Field Settings. Houghton-Mifflin, Boston.
Cramton, C.D., Hinds, P.J., 2005. Subgroup dynamics in internationally distributed teams: ethnocentrism or cross-national learning? Research in
Organizational Behavior 26, 233–265.
Crozier, M., Friedberg, E., 1994. L’Acteur et le Système: Les Contraintes de l’Action Collective. Editions de Seuil, Paris.
Cullen, S., Seddon, P.B., Willcocks, L.P., 2005. IT outsourcing configuration: research into defining and designing outsourcing arrangements. Journal of
Strategic Information Systems 14 (4), 357–387.
Cummings, J.N., 2004. Work groups, structural diversity, and knowledge sharing in a global organization. Management Science 50 (3), 352–364.
Dennis, A.R., Garfield, M.J., 2003. The adoption and use of GSS in project teams: toward more participative processes and outcomes. MIS Quarterly 27 (2),
289–323.
Dennis, A.R., Valacich, J.S., 1993. Computer brainstorms: more heads are better than one. Journal of Applied Psychology 78 (4), 531–537.
Diehl, M., Stroebe, W., 1987. Productivity loss in brainstorming groups: toward the solution of a riddle. Journal of Personality and Social Psychology 53 (3),
497–509.
Diehl, M., Stroebe, W., 1991. Productivity loss in idea-generating groups: tracking down the blocking effect. Journal of Personality and Social Psychology 61
(3), 393–403.
Duncan, N.B., 1998. Beyond opportunism: a resource-based view of outsourcing risk. In: Proceedings of the 31st Annual Hawaii International Conference on
System Sciences, pp. 675–684.
Eppinger, S.D., Chitkara, A.R., 2006. The new practice of global product development. MIT Sloan Management Review 47 (4), 22–30.
Espinosa, J.A., Carmel, E., 2003. The impact of time separation on coordination in global software teams: a conceptual foundation. Software Process
Improvement and Practice 8, 249–266.
Espinosa, A., Slaughter, S., Kraut, R., Herbsleb, J., 2007. Familiarity and performance in geographically distributed software development. Organization
Science 18 (4), 613–630.
Friedberg, E., 1993. Le pouvoir et la règle. Dynamiques de l’action organisée. Editions de Seuil, Paris.
Gallupe, R.B., Dennis, A.R., Cooper, W.H., Valacich, J.S., Bastianutti, L.M., Nunamaker Jr., J.F., 1992. Electronic brainstorming and group size. Academy of
Management Journal 35 (2), 350–369.
Ganesan, S., Malter, A.J., Rindfleisch, A., 2005. Does distance still matter? Geographic proximity and new product development. Journal of Marketing 69 (4),
44–60.
Gibson, C.B., Cohen, S.G., 2003. Virtual Teams that Work: Creating Conditions for Virtual Team Effectiveness. John Wiley and Sons, New York.
Griffith, T.L., Sawyer, J.E., Neale, M.A., 2003. Virtualness and knowledge in teams: managing the love triangle of organizations, individuals, and information
technology. MIS Quarterly 27 (2), 265–287.
Gubrium, J.F., Holstein, J.A., 2001. Handbook of Interview Research: Context and Method. Sage Publications, Thousand Oaks, CA.
Gupta, A. (Ed.), 2008. Outsourcing and Offshoring of Professional Services: Business Optimization in a Global Economy. Information Science Reference,
Hershey, PA.
Gupta, A., 2009a. The 24-hour knowledge factory: can IT replace the graveyard shift? IEEE Computer Society 42 (1), 66–73.
Gupta, A., 2009b. Deriving mutual benefits from offshore outsourcing: the 24-hour knowledge factory scenario. Communications of the ACM 52 (6), 122–
126.
Gupta, A., Seshasai, S., 2007. 24-Hour knowledge factory: using internet technology to leverage spatial and temporal separations. ACM Transactionson
Internet Technology 7 (3).
Gupta, A., Seshasai, S., Mukherji, S., Ganguly, A., 2007. Offshoring: the transition from economic drivers toward strategic global partnership and 24-hour
knowledge factory. Journal of Electronic Commerce in Organizations 5 (2), 1–23.
Haas, M.R., Hansen, M.T., 2005. When using knowledge can hurt performance: the value of organizational capabilities in a management consulting
company. Strategic Management Journal 26 (1), 1–24.
Haas, M.R., Hansen, M.T., 2007. Different knowledge, different benefits: toward a productivity perspective on knowledge sharing in organizations. Strategic
Management Journal 28 (11), 1133–1153.
Hackman, J.R., 1990. Groups that Work (and those that don’t): Creating Conditions for Effective Teamwork. Jossey-Bass, San Francisco.
Hansen, M.T., Nohria, N., Tierney, T., 1999. What’s your strategy for managing knowledge? Harvard Business Review 77 (2), 106–116.
Hightower, R., Sayeed, L., Warkentin, M., 2007. A longitudinal study of information exchange in computer-mediated and face-to-face groups. In: Kulkarni, U.,
Power, D.J., Sharda, R. (Eds.), Decision Support for Global Enterprises. Springer, New York.
160 A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161
Author's personal copy
Hinds, P.J., Mortensen, M., 2005. Understanding conflict in geographically distributed teams: the moderating effects of shared identity, shared context, and
spontaneous communication. Organization Science 16 (3), 290–309.
Humphrey, W.S., 1995. Introducing the personal software process. Annals of Software Engineering 1, 311–325.
IEEE, 2004. IEEE Position Statement on Offshore Outsourcing. IEEE-USA Career and Workforce Policy Committee Report.
Jarvenpaa, S.L., Leidner, D.E., 1999. Communication and trust in global virtual teams. Organization Science 10, 791–815.
Jarvenpaa, S.L., Knoll, K., Leidner, D.E., 1998. Is anybody out there? Antecedents of trust in global virtual teams. Journal of Management Information Systems
14 (4), 29–64.
Kiesler, S., Cummings, J., 2002. What do we know about proximity and distance in work groups? A legacy of research. In: Hinds, P., Kiesler, S. (Eds.),
Distributed Work. MIT Press, Cambridge, MA, pp. 57–80.
Kirkman, B.L., Rosen, B., Tesluk, P.E., Gibson, C.B., 2004. The impact of team empowerment on virtual team performance: the moderating role of face-to-face
interaction. Academy of Management Journal 47, 175–192.
Kogut, B., Zander, U., 1992. Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science 3, 383–397.
Kotlarsky, J., Oshri, I., 2005. Social ties, knowledge sharing and successful collaboration in globally distributed system development projects. European
Journal of Information Systems 14 (1), 37–48.
Kotlarsky, J., Oshri, I., van Fenema, P., 2008. Knowledge Processes in Globally Distributed Contexts. Palgrave Macmillan, Hampshire, UK.
Lipnack, J., Stamps, J., 1997. Virtual Teams – Reaching Across Space, Time, and Organizations with Technology. John Wiley and Sons, New York.
Loh, L., Venkatraman, N., 1995. An empirical study of information technology outsourcing: benefits, risks, and performance implications. In: Proceedings of
the 16th International Conference on Information Systems, Amsterdam, The Netherlands, pp. 277–288.
Lorge, I., Fox, D., Davitz, J., Brenner, M., 1958. A survey of studies contrasting the quality of group performance and individual performance, 1920–1957.
Psychological Bulletin 55 (6), 337–372.
Majchrzak, A., Rice, R.E., Malhotra, A., King, N., Ba, S., 2000. Technology adaptation: the case of a computer-supported inter-organizational virtual team. MIS
Quarterly 24 (4), 569–600.
Manning, S., Massini, S., Lewin, A.Y., 2008. The globalization of innovation: a dynamic perspective on offshoring. Academy of Management Perspectives 22
(3), 35–54.
Martins, L.L., Gilson, L.L., Maynard, M.T., 2004. Virtual teams: what do we know and where do we go from here. Journal of Management 30 (6), 805–835.
Massey, A.P., Montoya-Weiss, M.M., Hung, Y.-T., 2003. Because time matters: temporal coordination in global virtual project teams. Journal of Management
Information Systems 19 (4), 129–155.
Mattarelli, E., Gupta, A., 2009. Offshore-onsite subgroup dynamics in globally distributed teams. Journal of Information Technology and People, forthcoming.
Maznevski, M.L., Chudoba, K.M., 2000. Bridging space over time: global virtual team dynamics and effectiveness. Organization Science 11 (5), 473–492.
Mcgrath, J.E., 1991. Time, interaction, and performance (TIP): a theory of groups. Small Group research 22, 147–173.
McLeod, P., Baron, R., Marti, M., Yoon, K., 1997. The eyes have it: minority influence in face-to-face and computer-mediated group discussion. Journalof
Applied Psychology 82, 706–718.
Montoya-Weiss, M.M., Massey, A.P., Song, M., 2001. Getting it together: temporal coordination and conflict management in global virtual teams. Academy of
Management Journal 44 (6), 1251–1262.
Mullen, B., Johnson, C., Salas, E., 1991. Productivity loss in brainstorming groups: a meta-analytic integration. Basic and Applied Social Psychology 12 (1), 3–
23.
Nonaka, I., Toyama, R., Konno, N., 2000. SECI, Ba, and leadership: a unified model of dynamic knowledge creation. Long Range Planning 33, 5–34.
O’Leary, M.B., Cummings, J.N., 2007. The spatial, temporal, and configurational characteristics of geographic dispersion in teams. MIS Quarterly 31 (3), 1–20.
Orlikowski, W., 2000. Using technology and constituting structures: a practice lens for studying technology in organizations. Organization Science 11 (4),
404–428.
Orlikowski, W., Yates, J., 2002. It’s about time: temporal structuring in organizations. Organization Science 13 (6), 684–700.
Osborn, A.F., 1957. Applied imagination, Rev. ed. Scribner, New York.
Oshri, I., Kotlarsky, J., Willcocks, L.P., 2007. Global software development: exploring socialization and face-to-face meetings in distributed strategic projects.
Journal of Strategic Information Systems 16, 25–49.
Paulus, P.B., Larey, T.S., Putnam, V.L., Leggett, K.L., Roland, E.J., 1996. Social influence processes in computer brainstorming. Basic and Applied Social
Psychology 18 (1), 3–14.
Porter, M.E., 1998. Clusters and the new economy of competition. Harvard Business Review 76, 77–90.
Powell, A., Piccoli, G., Ives, B., 2004. Virtual teams: a review of current literature and directions for future research. ACM SIGMIS Database 35 (1), 6–36.
Rosenfeld, S.A., 1997. Bringing business clusters into the mainstream of economic development. European Planning Studies 5 (1), 3–23.
Saunders, C.S., Ahuja, M.K., 2006. Are all distributed teams the same? Differentiating between temporary and ongoing distributed teams. Small Group
Research 37, 662–700.
Steiner, I.D., 1972. Group Process and Productivity. Academic Press, New York.
Sternberg, R.J., Forsythe, G.B., Hedlund, J., Horvath, J.A., Wagner, R.K., Williams, W.M., Snook, S.A., Grigorenko, E.L., 2000. Practical Intelligence in Everyday
Life. Cambridge University Press, New York.
Tagliaventi, M.R., Mattarelli, E., 2006. The role of networks of practice, value sharing, and operational proximity in knowledge flows between professional
groups. Human Relations 59 (3), 291–319.
Tallman, S., Jenkins, M., Henry, N., Pinch, S., 2004. Knowledge, clusters, and competitive advantage. Academy of Management Review 29 (2), 258–271.
Taylor, D.W., Berry, P.C., Block, C.H., 1958. Does group participation when using brainstorming facilitate or inhibit creative thinking? Administrative Science
Quarterly 3, 23–47.
Thompson, JD., 1967. Organizations in Action. McGraw-Hill Publishing, New York.
Treinin, J.J., Miller-Frost, S.L., 2006. Following the sun: case studies in global software development. IBM Systems Journal 45 (4), 773–783.
Valacich, J.S., Dennis, A.R., Connolly, T., 1994. Idea generation in computer-based groups: a new ending to an old story. Organizational Behavior and Human
Decision Processes 57, 448–467.
Venkatraman, N., 2004. Offshoring without guilt. Sloan Management Review 45 (3), 14–16.
Walsham, G., 2005. Development, global futures and is research: a polemic. Journal of Strategic Information Systems 14 (1), 5–15.
White, L., 2004. NSPE Issue Brief: Offshore Outsourcing of Engineering Work. National Society of Professional Engineers, Publication No. 4065.
A. Gupta et al. / Journal of Strategic Information Systems 18 (2009) 147–161 161
