America Global client
Information Local customer
Office, USA relationship
IT strategy Customer Unit Manager
operations Overall coordination Overall coordination
Local Program Manager
SD Supervisor Local customer
Belgium Relationship Client Executive
Asian Local Local customer
0.4 FTE SD Manager
Poland Service delivery operations
7.4 FTE IT professional
2 3.5 FTE IT professional
India Service delivery operations
Figure 7.3 Governance structure of the CPG Inc. offshore outsourcing of infrastructure
to the previous domestic outsourcing. The client anticipated that the new offshore out-
sourcing engagement would require more staff time since CPG Inc. in Belgium had
limited experience in offshoring systems management relationships. Hence, these gov-
ernance roles were understood to be somewhat temporary: as the client gained off-
shoring experience the governance staf´¬üng needs would be reduced. There were also
other considerations for these roles. First, the assigned client Program Manager had a
long history with Atos Origin and was therefore needed for his experience in oversight.
Second, the Contract Manager was to gather the Balanced Scorecard data that came
from local operations in Belgium.
148 Managerial competency
Identify those offshore projects that require complex knowledge transfer and manage these
engagements patiently, devoting suf´¬ücient time and money to make them successful.
Successful knowledge transfer revolves around two essentials: staff rotations and, separately,
careful knowledge documentation. There are no short-cuts around these two essentials.
Organizational acceptance of offshoring often requires creating new centralized
organizational units, such as ÔÇťglobal sourcing units.ÔÇŁ
Organizational acceptance of offshoring requires internal selling and education efforts.
Organizational acceptance of offshoring requires organizational goals that are implemented
through measurement and rewards for managers.
Organizational acceptance of offshoring is helped by funding demonstration projects out of
Organizational acceptance of offshoring is most successful when it is spearheaded by a
respected offshore champion that acts as a catalyst for change.
As with change management, offshore outsourcing governance requires a new
organizational unit, such as an Information Of´¬üce, that centralizes many of the coordination
and relationship functions.
Effective governance requires many open channels of communication between client and
provider at multiple hierarchy levels.
Effective governance is fostered by good personal relations and trust between the
individuals in the liaison roles.
Effective governance can bene´¬üt from speci´¬üc documents: the SLA and, if appropriate, a
Effective governance requires personnel assigned to new roles. The governance overhead
that needs to be budgeted is about 6ÔÇ“7% of the contract value.
8 Overcoming distance and time
All things being equal, any manager would prefer to manage a co-located team
rather than a distributed team.
Offshoring requires distributed collaboration in which people work across distance and
time. This is a key dif´¬üculty in offshoring. In this chapter, we ´¬ürst take a close look at
the root of the problem: Why is it so dif´¬ücult to collaborate across distance and time?
We then present the many small solutions to this problem. The dif´¬üculties in distributed
collaboration cannot be eliminated, but they can be mitigated somewhat through a
mosaic of solutions described in this chapter: applying principles of formalisms and
informalisms, managing time differences, using a mix of collaborative technologies,
selecting the right staff, and designing the optimal organizational structure.
We like to be close
In spite of the hype about our new ÔÇťvirtual worldÔÇŁ in which ÔÇťdistance is dead,ÔÇŁ we humans
like proximity. We perform better when we are close together. We thrive when we have
face-to-face interaction.1 We crave proximity.2
In order to understand why distributed work is more dif´¬ücult for us humans, Kiesler
and Cummings3 gathered the results of decades of group psychology research. We
begin by summarizing their thought-provoking ´¬ündings.
The mere proximity to another human introduces a ÔÇťsocial facilitation effect.ÔÇŁ That
is, our physiologic performance changes: alertness increases, our heart rate goes up,
and our blood pressure increases. Television producers introduced laugh tracks to com-
edy shows because we all tend to laugh when others laugh; we smile when others
smile. Researchers found that when we experience an event with someone else the
event tends to be more memorable. Even more interesting, food tastes better when we
are with others. We feel more involved when we are with others; we are ÔÇťenergized.ÔÇŁ
Simply being familiar with someone tends to increase liking and to heighten identity
to the group or team. Simply being closer to someone else makes us more likely to con-
form, or obey orders. In one classic 1970s experiment, subjects were more likely to
apply dangerous electrical shocks to their fellow students when the orders were
received from someone close by.
150 Managerial competency
As social beings, we behave differently in different settings (in a bar, a church, or a
school). We all work in such a shared social setting, such as an of´¬üce. We get territorial
about social setting and about our personal territory. We all have territorial ÔÇťbubblesÔÇŁ
around us. We donÔÇ™t like it when others puncture these bubbles. We have territories
around our desk space or the group of team cubicles. The territory tends to strengthen
the ties we develop with the group of people we work with. Furthermore, this shared
social setting leads us to be more satis´¬üed with that team at large. Thus, cohesive teams
tend to sit together. The problem with our territorial attachment is that it interferes with
our identi´¬ücation with the larger distributed team: the distant programmer in Manila is
not a member of our territory.
Proximity also leads to spontaneous communication. These are chance-encounters
that lead to conversation: an encounter in the hallway, in the of´¬üce kitchen area, or
before a meeting. Americans call these spontaneous conversations ÔÇťwater coolerÔÇŁ con-
versations, still using an image that is fast disappearing from most American buildings,
namely a dispenser of cooled, piped water. These spontaneous conversations are enor-
mously powerful in organizational life because it is through these chance-encounters
that we get to know what others are working on and how well they are progressing. We
do quick problem solving, thus facilitating coordination. These chance conversations are
also important to non-task objectives: they help solidify relationships. We are more likely
to like people we chat with and we are more likely to be in´¬‚uenced by these people.
Of´¬üce distance has the highest impact on spontaneous communication. Many of us
have discovered from experience that we hardly ever see Ariel, who has an of´¬üce just
one building away from us. Tom Allen was the ´¬ürst to con´¬ürm this when he measured
workersÔÇ™ chance-encounters and found that spontaneous encounters for people whose
of´¬üces were more than 30 m apart reduced to a chance of only 10% per day (Figure 8.1).
Silicon Valley illustrates how proximity makes a difference. Most high-tech ´¬ürms
bene´¬üt from being in high-tech clusters, the most successful of which is Silicon Valley.
Probability of communication in a
given time period
30 meters individuals
Figure 8.1 The Allen curve showing the probability of spontaneous communication between
151 Overcoming distance and time
All else being equal, each ´¬ürm in the cluster is better off from its location in the clus-
ter rather than in a distant, remote location. Innovative ´¬ürms bene´¬üt from the many
face-to-face interactions that clusters enable.
Understanding the problems of distance
We use a physics metaphor to frame the problem of distance.5 A centrifugal force is a
physical force that propels an object away from the center. Distributed software col-
laboration is like a centrifugal force that propels the team members apart from each other
(see Figure 8.2). Each of the ´¬üve centrifugal forces is introduced and explained here.
We human beings communicate best when we are close. Why? Because we are condi-
tioned to convey and read each other via more than the naked text that we utter. The way
the text is delivered, via tone of voice, the pauses in our speech, an accent that we place
on a phrase, all convey so much. Furthermore, our body tells a story. We open our eyes,
furrow our brow, smile, frown, gesture with our hands, point with our ´¬üngers. All of these
are part of our communication that is lost when we try to send a message, or convey a
vision, over a narrow communication channel, such as e-mail. Some say that 80% of the
message we convey is in the non-naked text. In fact, some go further and make an evolu-
tionary argument: during most of our evolution as a species, our ancestors communicated
primarily face-to-face, so our brains are hard-wired for this form of communication.6
r r i er s
Figure 8.2 The ´¬üve centrifugal forces that make distributed software work dif´¬ücult.
152 Managerial competency
We also know that the more complex and important the task, the more we need face-
to-face communication, or at least a telephone call.7 A serious design session is dif´¬ü-
cult to do over distance. A large contract negotiation is rarely done over distance.
Requirements gathering will not be comprehensive when done over distance.
Successful communication has an exacting de´¬ünition that is worth pondering: com-
munication is complete when the information has been transmitted, received, acknowl-
edged, understood, and acted upon. Accordingly, when our messages are not understood
properly, which is much more likely when we are far apart, we say that we have a mis-
communication. In distributed collaboration this miscommunication leads to delays
(because of the need to clarify); to rework (because we didnÔÇ™t really understand what
was meant); and, most painfully, miscommunication leads to con´¬‚ict because, all of
us, in all countries, get fussy over personal slights that we interpret in the message text.
Software development cycles require frequent ÔÇťsmall adjustments.ÔÇŁ8 Coordination is
the act of making those adjustments, or more formally, it is the act of integrating each
task and organizational unit, so that it contributes to the overall objective. We coordinate
work via countless small adjustments: a question, a request for clari´¬ücation, a small
improvement, an ad hoc solution resulting from a one-minute chat while standing in
line at the cafe. When you are working in the same location, you are aware of who to
contact for help because sheÔÇ™s one ´¬‚oor below you and she sits next to Josepha with
whom you play tennis. In fact, awareness and its close cousin ÔÇťshared knowledgeÔÇŁ are
vital elements of coordination.9
In distributed collaboration all of these small adjustments are dif´¬ücult, since much
of our coordination results from spontaneous conversations or from the small cues
about what is happening in the project. And, time separation makes all this worse.
When coordination slows or breaks down, several dynamics occur. Problem solving
gets delayed again and again until it becomes very expensive to ´¬üx. Things move along
the wrong path for so long that it becomes dif´¬ücult to renegotiate all the sides back
Control is the process of adhering to goals, policies, and standards. Years of experience
teach us that successful control takes place when managers can roam around to
see, observe, and dialogue with their staff. Hence, MBWA (Management By Walking
When a software project manager is supervising developers many kilometers away,
roaming around and getting a ÔÇťfeelÔÇŁ for whatÔÇ™s happening becomes an unusual event.
Sometimes it never happens at all. And, when managers cannot roam, they have to rely
on collecting information and imposing their will by means of technology, through
telephone and e-mail. This is less effective then face-to-face. As an added insult, managers
153 Overcoming distance and time
tend to pay more attention to those who are close to them and less to those far away,
and thus the saying: ÔÇťout of sight, out of mind.ÔÇŁ The result of poor control in distrib-
uted collaboration is wasteful duplication of effort, discovering problems late, and the
subsequent need for rework.
Groups that are close together jell and bond.10 People get to like each other, trust each
other, help each other, work harder for each other. It is not surprising that in distributed
collaboration it is very dif´¬ücult to foster cohesion unless there is history. That is, unless
the people have worked together before in the same location.
Distributed teams are more diverse, by culture and organizational background. Such
teams are less likely to rally around the same vision. Often, members are working on
multiple projects and on multiple teams. For example, at Intel, one of the most dis-
persed technology companies, a 2003 survey found that 63% of Intel employees were
members of at least two teams and most of these teams were, themselves, virtual teams.
(The Intel case appears later in this chapter.)
Cohesive project teams trust each other. But building this trust takes time, even for
co-located teams, let alone for distributed teams. Trust presents a paradox for distributed
collaboration: on the one hand you need trust in order to work effectively together over
distance; on the other hand it is quite dif´¬ücult to develop trust over distance.
Trust is an important, but elusive, concept that merits a brief digression. First, it should
be evident that the opposite of trust, known as mistrust, is to be avoided. Mistrust, once
it begins, can ooze and fester and destroy everything that it touches. So, it is critical to
actively reduce the chances that mistrust appears.
Our favorite de´¬ünition of trust is ÔÇťthe comfort to make yourself vulnerable.ÔÇŁ Social
scientists have formulated many more de´¬ünitions, types, and levels of trust. For exam-
ple, one level of trust is knowing what the other will do. Some call this ÔÇťcognitive
trustÔÇŁ, a belief about the otherÔÇ™s reliability. But an even higher level of trust is allowing
the other side to act in your place. Complicating matters further, some individuals and
cultures trust quickly (Americans are said to be in this category), other cultures more
Geert Hofstede, the famed culture scholar, calls culture ÔÇťthe programming of the mind
that separates one group from another.ÔÇŁ11 Culture de´¬ünes each personÔÇ™s principles,
values, beliefs, and behaviors, including communication behavior. As a result, in any
cross-cultural communication, the receiver is more likely to misinterpret messages or
cues. Hence, the familiar complaint of miscommunication across cultures. Worse,
small cultural gaffes lead to culture clashes, mistrust, and eventually con´¬‚ict. In fact,
culture is so important in offshoring that we devote an entire chapter to the topic
154 Managerial competency
Each of these ´¬üve centrifugal forces lead to problems, and, in some cases, failed projects.
The most dif´¬ücult outcomes were noted by Jim Herbsleb, one of the pioneers in the study
of global software teams.12 He recently summarized several years of studying these kinds
of teams. He concluded that there were two particularly serious outcomes. First, these dis-
persed teams cannot deal with unexpected events. Second, these teams suffer from issue
resolution paralysis, which means that it is dif´¬ücult for them to arrive at closure on dif´¬ücult
issues. Both outcomes stem from combinations of the centrifugal forces presented above.
Formalize and informalize
The ´¬üve centrifugal forces (the ´¬üve core problems of distance and time) can be miti-
gated by applying a wide range of organizational and technological solutions. In this
section, and in the rest of this chapter, we present these solutions.
First, the dictum that should guide you is: