After briefly reviewing the existing literature on team coaching, we propose a new
model with three distinguishing features. The model (1) focuses on the functions that
coaching serves for a team, rather than on either specific leader behaviors or leadership
styles, (2) identifies the specific times in the task performance process when
coaching interventions are most likely to have their intended effects, and (3) explicates
the conditions under which team-focused coaching is and is not likely to
facilitate performance.
1. Academy of Management Review
2005, Vol. 30, No. 2, 269–287.
A THEORY OF TEAM COACHING
J. RICHARD HACKMAN
Harvard University
RUTH WAGEMAN
Dartmouth College
After briefly reviewing the existing literature on team coaching, we propose a new
model with three distinguishing features. The model (1) focuses on the functions that
coaching serves for a team, rather than on either specific leader behaviors or lead-
ership styles, (2) identifies the specific times in the task performance process when
coaching interventions are most likely to have their intended effects, and (3) expli-
cates the conditions under which team-focused coaching is and is not likely to
facilitate performance.
Coaches help people perform tasks. Coaching TEAM COACHING
is pervasive throughout the life course, from
Team coaching is an act of leadership, but it is
childhood (e.g., a parent helping a child learn to
not the only one or necessarily the most conse-
ride a tricycle), through schooling (e.g., a teacher
quential one. Team leaders engage in many dif-
coaching a student in the proper conduct of a
ferent kinds of behaviors intended to foster team
chemistry experiment), and into adulthood (e.g.,
effectiveness, including structuring the team
a fitness coach helping with an exercise regime
or a supervisor coaching an employee in im- and establishing its purposes, arranging for the
proving his or her job performance). The main resources a team needs for its work and remov-
body of research about coaching is found in the ing organizational roadblocks that impede the
training literature, and it focuses almost entirely work, helping individual members strengthen
on individual skill acquisition (Fournies, 1978). their personal contributions to the team, and
Except for the many popular books and articles working with the team as a whole to help mem-
that extract lessons for team leaders from the bers use their collective resources well in pur-
experiences of athletic coaches, relatively little suing team purposes.
has been published that specifically addresses Leaders vary in how they allocate their time
the coaching of task-performing teams. and attention across these activities, depending
Here we propose a theory of team coaching on their own preferences; what they believe the
that is amenable to empirical testing and cor- team most needs; and the team’s own level of
rection. The theory has three distinguishing fea- authority, initiative, and maturity. Only the last
tures. One, it focuses on the functions that two sets of activities (helping individual mem-
coaching serves for a team, rather than on either bers strengthen personal contributions and
specific leader behaviors or leadership styles. working with the team to help use resources
Two, it explicitly addresses the specific times in well) are coaching behaviors, however, focusing
the task performance process when coaching respectively on individual team members and
interventions are most likely to “take” and have on the team as a whole. In this paper we deal
their intended effects. Three, it explicitly identi- exclusively with the fourth—team coaching—
fies the conditions under which team-focused which we define as direct interaction with a
coaching is most likely to facilitate perfor- team intended to help members make coordi-
mance. Overall, we show that the impact of nated and task-appropriate use of their collec-
team coaching—whether provided by a formal tive resources in accomplishing the team’s work.
team leader or by fellow group members— Although team coaching is a distinct and of-
depends directly and substantially on the de- ten consequential aspect of team leadership, re-
gree to which the proper coaching functions are cent evidence suggests that leaders focus their
fulfilled competently at appropriate times and behavior less on team coaching than on other
in appropriate circumstances. aspects of the team leadership portfolio. In a
269
2. 270 Academy of Management Review April
study of 268 task-performing teams in 88 organi- Wilson, 1991). Although varied, these models
zations, we (Wageman, Hackman, & Lehman, specify ways that team leaders can develop
2004) asked team leaders and members to rank members’ interpersonal skills, define members’
the amount of attention the team leader gave to roles and expectations, deal with conflict and
activities in each of the four categories listed interpersonal frictions, and help a team achieve
above (with a rank of “1” signifying the greatest a level of “maturity” that lessens the team’s de-
attention). For both leader and member reports, pendence on its leader (Eden, 1985; Fischer, 1993;
coaching the team as a whole came in last (the Geber, 1992; Manz & Sims, 1987; Patten, 1981;
combined mean ranks were as follows: structur- Rees, 1991; Torres & Spiegel, 2000; Woodman &
ing the team and its work, 1.75; running external Sherwood, 1980).
interference, 2.16; coaching individuals, 2.88;
and coaching the team, 3.02).
Process Consultation
The lesser attention given to team coaching
could simply mean that leaders underestimate The process consultation approach developed
the potential benefits of providing coaching as- by Schein (1969, 1988) posits that competent in-
sistance to their teams. More likely, perhaps, is terpersonal relations are essential for effective
that leaders do not coach their teams because task performance and that group members
they do not know how to do so, or they have themselves must be intimately involved in ana-
ventured a coaching intervention or two that did lyzing and improving those relationships. The
not help and thereafter focused their behavior consultant engages team members in analyzing
on seemingly more promising team leadership group processes on two levels simultaneously:
strategies. By using existing research and the- (1) the substantive level—to analyze how human
ory to identify the kinds of leader coaching be- processes are affecting work on a specific organ-
haviors that do help teams operate more effec- izational problem—and (2) the internal level—to
tively, we seek here not only to advance basic better understand the team’s own interaction
understanding about team coaching but also to processes and the ways that team processes
provide practitioners with some of what they foster or impede effective group functioning
need to know to coach their teams competently. (Schein, 1988: 11–12). Decidedly clinical in orien-
tation, this type of coaching requires the process
consultant first to directly observe the group as
EXISTING APPROACHES
it works on a substantive organizational prob-
In a review of existing research and theory, lem and then, once the group is ready, to intro-
we identified three conceptually driven ap- duce systematic or confrontive interventions in-
proaches to team coaching and one eclectic ap- tended to help the team deal with its problems
proach that is largely atheoretical. These four and exploit previously unrecognized opportuni-
approaches, described below, point the way to- ties.
ward a more comprehensive research-based
model of team coaching, and we draw on them
Behavioral Models
in developing propositions for the present the-
ory. Two distinct models of team coaching are
based on theories of individual behavior: (1) the
application of Argyris’s (1982, 1993) theory of in-
Eclectic Interventions
tervention to team-focused coaching by Schwarz
Eclectic coaching interventions are activities (1994) and (2) applications of operant condition-
that derive from no particular theoretical per- ing to the modification of team behavior, nota-
spective but have considerable face validity bly those of Komaki (1986, 1998) and her col-
nonetheless, in that a lay person would be likely leagues.
to assume that they would help a team perform In his approach, Schwarz posits that coaches
well. Eclectic models are found mainly in the should provide feedback to a team in ways that
practitioner literature as codifications of the les- help members learn new and more effective
sons learned by management consultants team behaviors, especially in how they give and
whose practice includes team facilitation (e.g., receive feedback. The coaching process in-
Fischer, 1993; Kinlaw, 1991; Wellins, Byham, & volves three phases. First is observing actual
3. 2005 Hackman and Wageman 271
group behavior both to note behaviors that are identify the issues the team needs to deal with
impeding the group’s work and to identify be- next. The focus of learning sessions for newly
haviors not presently exhibited that might facil- formed or “novice” groups (whose members are
itate group work. Second is describing to the mainly occupied with social issues of inclusion
group what has been observed and testing in- and acceptance and task issues having to do
ferences about the meanings of those behaviors. with team goals and with member skills and
And third is helping group members decide roles) differs from that for more mature or “ex-
whether they wish to change their behaviors pert” groups (whose members have become
and, if so, how they might do so. The model ready to learn strategies for self-regulation,
specifies several specific ground rules both for such as how best to detect and correct errors and
the facilitators’ behaviors and for team mem- how best to adapt to changing external de-
bers’ behaviors, such as providing specific be- mands). Because teams are unlikely to be able
havioral examples for points made, publicly to process intensive interventions when task de-
testing assumptions and inferences, and explic- mands are also high, learning sessions are re-
itly inviting questions and challenges. served for periods of relatively low cognitive
The operant conditioning approach to team demand. During intensive work periods, devel-
coaching is based on the well-established prin- opmental coaches focus mainly on gathering
ciple of individual learning that behavior is a data about behavior and performance for use
function of its consequences. Applied to teams, guiding subsequent interventions. When task
operant coaching involves three kinds of coach- demands diminish, active coaching resumes.
ing behaviors: (1) providing instructions about
how to behave, (2) monitoring the team’s perfor-
Summary
mance, and (3) providing performance-contin-
gent consequences to the team (Komaki, 1986; Most of the approaches to team coaching just
Smith, Smoll, & Curtis, 1979). Because the oper- summarized are based on well-established psy-
ant approach to team coaching does not specify chological principles and findings about human
any particular patterns of team interaction that learning and performance. Moreover, research
facilitate effectiveness across different types of conducted within each tradition has been infor-
teams and tasks, team coaches must have ex- mative and, especially for the operant and de-
tensive task knowledge so that they can issue velopmental approaches, has generated empir-
proper instructions about desirable behaviors ical findings that enrich our understanding
and reinforce the team when it does well about coaching processes and outcomes. None
(Komaki, 1998; Komaki, Deselles, & Bowman, of the existing approaches, however, is sup-
1989; Komaki & Minnich, 2002; Smoll & Smith, ported by evidence that addresses all links in
1989). the coaching intervention–team process–team
performance sequence. We seek here to provide
a conceptual model that does explicate all links
Developmental Coaching
in that sequence, that takes explicit account of
The distinguishing feature of the developmen- teams’ temporal and organizational contexts,
tal approach to coaching is the central role and that provides a sound basis for generating
given to time and timing. Two premises on guidance for team coaching practice.
which this approach is based are (1) that teams
need help with different issues at different
DOMAIN
stages of their development and (2) that there
are times in the life cycles of groups when they We begin by specifying what we mean by a
are more and less open to intervention (Kozlow- work team and what we mean by performance
ski, Gully, McHugh, Salas, & Cannon-Bowers, effectiveness, which together bound the domain
1996; Kozlowski, Gully, Salas, & Cannon-Bowers, of our model.
1996).
A key coaching intervention in the develop-
Work Teams
mental approach is the “learning session,” in
which the coach and team members review the We focus only on full-fledged teams that per-
team’s purpose, assess its progress thus far, and form tasks in social system contexts. Such teams
4. 272 Academy of Management Review April
have three features. First, they are real groups. Because we believe that these dimensions also
That is, they are intact social systems, complete are consequential for any team’s long-term or-
with boundaries, interdependence among mem- ganizational performance, we define team effec-
bers, and differentiated member roles (Alderfer, tiveness using the following three-dimensional
1977). Members of real groups can be distin- conception (adapted from Hackman, 1987).
guished reliably from nonmembers, they are in-
terdependent for some common purpose, and 1. The productive output of the team (i.e., its
they invariably develop specialized roles within product, service, or decision) meets or ex-
the group. Real groups can be either small or ceeds the standards of quantity, quality,
and timeliness of the team’s clients—the
large and either temporary or long-lived. people who receive, review, and/or use the
Second, work teams have one or more group output. It is the clients’ standards and as-
tasks to perform. They produce some outcome for sessments that count in assessing team
which members bear collective responsibility products—not those of the team itself (ex-
and for which acceptability is potentially as- cept in rare cases where the team is the
client of its own work) or those of the team’s
sessable. The kind of outcome produced is not manager (who only rarely is the person who
critical—it could be a physical product, a ser- actually receives and uses a team’s output).
vice, a decision, a performance, or a written 2. The social processes the team uses in car-
report. Nor is it necessary that the outcome ac- rying out the work enhance members’ capa-
tually be assessed; all that is required is that the bility of working together interdependently
in the future. Effective teams become adept
group produce an outcome that can be identified at detecting and correcting errors before se-
as its product and that it be theoretically possi- rious damage is done and at noticing and
ble to evaluate that product. Social groups and exploiting emerging opportunities. They are
other collectives that generate no identifiable more capable performing units when they
product fall outside our domain. finish a piece of work than when they be-
gan.
Finally, work teams operate in a social system 3. The group experience contributes positively
context. The team as a collective manages rela- to the learning and personal well-being of
tionships with other individuals or groups in individual team members. Work teams can
some larger social system. Usually this social serve as sites for personal learning and can
system is the parent organization that created spawn satisfying interpersonal relation-
ships, but they also can deskill, frustrate,
the team, but it can be people or groups outside and alienate their members. We do not
that organization as well, such as opponents for count as effective any team for which the net
an athletic team or customers for a service- impact of the group experience on members’
providing team. What is critical is that team learning and well-being is more negative
members be collectively responsible for manag- than positive.
ing consequential transactions with other indi-
Although the three criteria vary in importance
viduals and/or groups.
in different circumstances, effective teams bal-
ance them over time, never completely sacrific-
ing any one to achieve the others. In the pages
Team Performance Effectiveness that follow, we identify the coaching functions,
the temporal imperatives, and the contextual
Criterion measures in empirical research on
circumstances that affect the degree to which
team performance often consist of whatever
coaching behaviors can help a work team
quantitative indicators happen to be available
achieve and sustain a high standing on all three
or are easy to obtain (e.g., production figures for
of the criteria.
industrial workgroups or number of correct re-
sponses for teams studied in experimental lab-
oratories). Such criteria of convenience do not
FUNCTIONS
address other outcome dimensions, such as cli-
ent assessments of a team’s work, the degree to Over four decades ago, McGrath first sug-
which a team becomes stronger as a performing gested that “[The leader’s] main job is to do, or
unit over time, or the extent to which individual get done, whatever is not being adequately han-
members become more knowledgeable or dled for group needs” (1962: 5). If a leader man-
skilled as a result of their team experiences. ages, by whatever means, to ensure that all
5. 2005 Hackman and Wageman 273
functions critical to group performance are cient effort, use inappropriate strategies, and/or
taken care of, the leader has done his or her job apply inadequate talent in their work—are
well. Thus, a functional approach to leadership likely to fall short in one or more of the effective-
leaves room for an indefinite number of ways to ness criteria.
get key group functions accomplished, and Associated with each of the three performance
avoids the necessity of delineating all the spe- processes are both a characteristic “process
cific behaviors or styles a leader should exhibit loss” (Steiner, 1972) and an opportunity for pos-
in given circumstances—a trap into which it is itive synergy, which we call a “process gain.”
easy for leadership theorists to fall. That is, members may interact in ways that de-
What functions are most critical for team per- press the team’s effort, the appropriateness of its
formance effectiveness? Functions whose ac- strategy, and/or the utilization of member talent;
complishment are critical for group decision alternatively, their interaction may enhance col-
making have been identified by Hirokawa and lective effort, generate uniquely appropriate
Orlitzky (Hirokawa, 1985; Orlitzky & Hirokawa, strategies, and/or actively develop members’
2001), and those that bear on other aspects of knowledge and skills.
group behavior have been comprehensively re- Coaching functions are those interventions
viewed by Hollingshead et al. (in press). For our that inhibit process losses and foster process
specific and delimited purposes—that is, iden- gains for each of the three performance pro-
tification of the most critical functions served by cesses. Coaching that addresses effort is moti-
those who coach work teams—we focus on three vational in character; its functions are to mini-
aspects of group interaction that have been mize free riding or “social loafing” and to build
shown to be especially potent in shaping group shared commitment to the group and its work.
performance outcomes (Hackman & Morris, 1975; Coaching that addresses performance strategy
Hackman & Walton, 1986). is consultative in character; its functions are to
Specifically, we posit that team effectiveness minimize mindless adoption or execution of task
is a joint function of three performance pro- performance routines in uncertain or changing
cesses: (1) the level of effort group members col- task environments and to foster the invention of
lectively expend carrying out task work, (2) the ways of proceeding with the work that are espe-
appropriateness to the task of the performance
cially well aligned with task requirements.
strategies the group uses in its work,1 and (3) the
Coaching that addresses knowledge and skill is
amount of knowledge and skill members bring
educational in character; its functions are to
to bear on the task. Any team that expends suf-
minimize suboptimal weighting of members’
ficient effort in its work, deploys a task-
contributions (i.e., when the weight given to in-
appropriate performance strategy, and brings
dividual members’ contributions is at variance
ample talent to bear on its work is quite likely to
with their actual talents) and to foster the devel-
achieve a high standing in the three criteria of
opment of members’ knowledge and skill.
work team effectiveness specified earlier. By the
The three coaching functions specifically and
same token, teams that operate in ways that
exclusively address a team’s task performance
leave one or more of these functions unful-
filled—that is, where members expend insuffi- processes—not members’ interpersonal rela-
tionships. This focus distinguishes our model
from the great majority of writing and practice
about team coaching, especially in the eclectic
1
A team’s strategy is the set of choices members make tradition, which posits (sometimes explicitly but
about how to carry out the work. For example, a manufac- more often implicitly) that coaching interven-
turing team might decide to divide itself into three sub-
tions should primarily address the quality of
groups, each of which would produce one subassembly, with
the final product to be assembled later. Or a basketball team members’ interpersonal relationships.
might decide to use modified zone defense, with one player The pervasive emphasis on interpersonal pro-
assigned to guard the opposing team’s best shooter. Or a cesses in the team performance literature re-
team performing a task that requires a creative solution flects a logical fallacy about the role of those
might choose to free associate about possible solutions in
the first meeting, reflect for a week about the ideas that
processes in shaping performance outcomes. To
came up, and then reconvene to draft the product. All of illustrate, consider a team that is having perfor-
these are choices about task performance strategy. mance problems. Such teams often exhibit inter-
6. 274 Academy of Management Review April
personal difficulties, such as communications former to significantly outperform the latter
breakdowns, conflict among members, leader- (Kernaghan & Cooke, 1990; Woolley, 1998).2
ship struggles, and so on. Because both lay per-
Proposition 1: Coaching interventions
sons and scholars implicitly rely on an input– that focus specifically on team effort,
process– output framework in analyzing group strategy, and knowledge and skill fa-
dynamics, it is natural to infer that the observed cilitate team effectiveness more than
interpersonal troubles are causing the perfor- do interventions that focus on mem-
mance problems and, therefore, that a good way bers’ interpersonal relationships.
to improve team performance would be to fix
them. As reasonable as this inference may
seem, it is neither logical nor correct. Although TIMING
serious interpersonal conflicts sometimes do un- The efficacy of coaching interventions de-
dermine team performance (Jehn & Mannix, pends not just on their focus, as discussed
2001), it does not necessarily follow that the above, but also on the time in the group’s life
proper coaching intervention in such cases is to cycle when they are made. Regularities in group
help members improve their interpersonal rela- life cycles have been explored empirically for
tionships. many decades, beginning with Bales and Strodt-
In fact, research suggests that, in some cir- beck’s (1951) classic study of phases in group
cumstances, the causal arrow points in the op- problem solving. In a number of conceptual
posite direction—that is, performance drives in- frameworks, scholars have sought to summarize
terpersonal processes (or, at least, perceptions research findings about group development, the
of those processes), rather than vice versa. For most prominent being the “forming-storming-
example, Staw (1975) gave task-performing norming-performing” model proposed by Tuck-
man (1965). Almost all of these frameworks have
teams false feedback about their performance
treated group development as following a fixed
and then asked members to provide “objective”
set of stages, with each successive stage being
descriptions of how members had interacted.
contingent on successful completion of the prior
Teams randomly assigned to the high perfor-
one.
mance condition reported more harmonious and In recent years, research on temporal issues in
better communications, among other differ- group behavior has raised doubt about the gen-
ences, than did groups assigned to the low per- erality and validity of stage models (Ancona &
formance condition (see also Guzzo, Wagner, Chong, 1999; Gersick, 1988; Ginnett, 1993;
Maguire, Herr, & Hawley, 1986). McGrath & Kelly, 1986; Moreland & Levine, 1988;
Doubt also is cast on interpersonal ap- for a recent attempt to reconcile alternative tem-
proaches to coaching by action research that poral models, see Chang, Bordia, & Duck, 2003).
seeks to improve team performance by improv- Gersick’s findings are particularly relevant for
ing the quality of members’ interactions. Some our purposes. In a field study of the life histories
of these studies use interventions based on the of a number of project teams whose performance
process consultation approach to coaching re- periods ranged from several days to several
viewed earlier; others employ a broader set of months, Gersick (1988) found that each of the
interventions that generally are referred to as groups she tracked developed a distinctive ap-
team building or group development activities. proach toward its task as soon as it commenced
Although interventions that address members’ work, and each stayed with that approach until
relationships and interaction can be quite en- precisely halfway between its first meeting and
gaging and do affect members’ attitudes, they its project deadline. At the midpoint of their
do not reliably improve team performance (for
reviews, see Kaplan, 1979; Salas, Rozell, Mullen,
& Driskell, 1999; Tannenbaum, Beard, & Salas, 2
Woolley’s main effect finding was significantly condi-
1992; Woodman & Sherwood, 1980). Moreover, tioned by the timing of the intervention, and the Kernaghan
and Cook finding was obtained only for groups composed of
those experimental studies that have directly members with ample task-relevant abilities. We discuss the
compared teams given task-focused and inter- moderating effects of timing and of group design on the
personally focused interventions have found the impact of coaching interventions later.
7. 2005 Hackman and Wageman 275
lives, almost all teams underwent a major tran- likely to be helpful if they are provided at a time
sition. In a concentrated burst of changes, they in the life cycle when the team is not ready for
dropped old patterns of behavior, reengaged them. Indeed, ill-timed interventions may actu-
with outside supervisors, and adopted new per- ally do more harm than good by distracting or
spectives on their work.3 Following the midpoint diverting a team from other issues that do re-
transition, groups entered a period of focused quire members’ attention at that time. We next
task execution, which persisted until very near discuss the kinds of interventions that are most
the project deadline, at which time a new set of appropriate at the beginnings, midpoints, and
issues having to do with termination processes ends of work team life cycles.
arose and captured members’ attention. Gersick
(1989) subsequently replicated these findings in
Beginnings
the experimental laboratory for groups that all
had the same amount of time to complete their When team members first come together to
task (for alternative views of temporal dynamics perform a piece of work, the most pressing piece
in task-performing groups, see Seers & Woo- of business, both for members and for the team
druff, 1997, and Waller, Zellmer-Bruhn, & Giam- as a whole, is for them to get oriented to one
batista, 2002). another and to the task in preparation for the
Gersick’s findings about the natural develop- start of actual work. This involves establishing
mental processes of task-performing groups the boundary that distinguishes members from
raise the possibility, consistent with both the nonmembers, starting to differentiate roles and
process consultation and developmental ap- formulate norms about how members will work
proaches to team coaching previously summa- together, and engaging with (and, inevitably,
rized, that the readiness of work teams for coach- redefining) the group task. These activities,
ing interventions changes systematically across which involve simultaneous engagement with
their life cycles. By “readiness for coaching” we the interpersonal and task issues that dominate
mean (1) the degree to which the issues to be the start-up of any social system, create a high
addressed are among those naturally on team state of readiness for anything that shows prom-
members’ minds at the time of the intervention, ise of reducing members’ uncertainties and
coupled with (2) the degree to which the team as helping them get off to a good start. A coaching
a whole is not at that time preoccupied with intervention that helps a group have a good
more pressing or compelling matters. “launch,” therefore, can significantly enhance
We posit that coaching interventions are more members’ commitment to the team and the task,
effective when they address issues a team is and thereby enhance their motivation to perform
ready for at the time they are made and, more- the work of the team as well as they can.
over, that readiness varies systematically The power and persistence of coaching be-
across the team life cycle. In contrast, even com- haviors at the launch of a task-performing team
petently administered interventions are un- are affirmed by Ginnett’s (1993) study of the be-
havior of airline captains during the first few
3
minutes of a newly formed crew’s life. The struc-
In Gersick’s research it was not clear whether the mid- tural “shell” within which cockpit crews work is
point transition was prompted externally (i.e., by reference
to a clock or calendar) or internally (i.e., by members’ sense quite detailed: the aircraft to be flown, where it
that about half their allotted time had elapsed). Mann (2001) is to be flown, the roles of each crew member,
investigated this question experimentally by having groups basic work procedures such as checklists, and
work in a room with a clock that ran normally, in one where much more all are prespecified and well under-
the clock ran one-third faster than normal (i.e., when thirty
stood by all crewmembers. Therefore, a new
minutes had passed, the clock showed that forty minutes
had elapsed), or in one where the clock ran one-third slower crew should be able to proceed with its work
than normal (i.e., at the thirty-minute mark it showed twenty without spending time getting organized, which
minutes). Groups with the normal clock experienced a single is, in fact, what happens if a captain does not
midpoint transition, replicating earlier findings. But groups conduct a launch briefing when the crew first
with the faster and slower clocks exhibited two such transi-
meets.
tions—one at the midpoint indicated by the clock and the
other at the actual midpoint of the allotted time—showing Consistent with Gersick’s results, Ginnett
that groups pace their work in response to both internal and (1993) found that what happened in the first few
external cues about elapsed time. minutes of crewmembers’ time together carried
8. 276 Academy of Management Review April
forward throughout a significant portion of the groups did better than the “discuss your strat-
crew’s life. Crews led by captains who merely egy first” groups on the version of the task for
took the time in their preflight briefings to affirm which the obvious way of proceeding was the
the positive features of the crew shell fared bet- optimum strategy, and the reverse was true
ter than those whose captains gave no briefing when the obvious way of proceeding was sub-
at all or those whose captains undermined the optimal.
standard shell. Best of all were crews whose Perhaps the most significant finding of the
captains went beyond mere affirmation and ac- study, however, is buried in the discussion sec-
tively elaborated the shell—identifying, com- tion of the research report—namely, that it was
menting on, and engaging their crews in a dis- nearly impossible to get the experimental
cussion of the unique circumstances of the trip groups to actually have a discussion of perfor-
that was about to begin. These captains trans- mance strategy at the start of the work period.
formed a highly competent set of individual pi- Only by structuring the strategy intervention as
lots into a motivated flight crew. a “preliminary task” and explicitly requiring the
Because most work teams do not have struc- team to check off each step as it was completed
tures as detailed and elaborate as those of cock- could the experimental groups be induced to
pit crews, what happens as members come to- have more than a perfunctory discussion of their
gether and come to terms with their work should performance strategy at the beginning of the
shape their trajectories even more profoundly work period. Beginnings are not a good time for
than was the case for Ginnett’s crews. Begin- strategy discussions, but, as will be seen next,
nings provide a unique opportunity for motiva- midpoints are.
tional coaching interventions that breathe life
into a team’s structural shell—no matter how
Midpoints
rudimentary or how elaborate that shell may
be—and thereby help get a team off to a good A second window for coaching interventions
start with high motivational engagement by all opens around the midpoint of the team’s work.
members. At that point the team has logged some experi-
In contrast, and perhaps surprisingly, begin- ence with the task, providing data for members
nings are not good times to help teams formu- to use in assessing what is working well and
late work strategy. When they are just starting poorly. Moreover, the team is likely to have ex-
out, teams are not yet ready to address ques- perienced an upheaval, driven in part by mem-
tions of strategy, as Hackman, Brousseau, and bers’ anxieties about the amount of work they
Weiss (1976) inadvertently discovered in an ex- have accomplished relative to the time they
perimental study of team performance. These have remaining, that opens the possibility of
researchers asked a subset of participating significant change in how the team operates
teams to take a few minutes to reflect on their (Gersick, 1988, 1989). For these reasons, readi-
performance strategy—that is, to consider vari- ness for a strategy-focused coaching interven-
ous ways of carrying out the task— before actu- tion is high at the temporal midpoint of a team’s
ally starting work on it. The investigators hy- work.
pothesized that these teams would perform We posit that ongoing teams having no dead-
better than teams that were encouraged to line, and therefore no temporal midpoint, also
plunge immediately into the work— but only on experience increased readiness for strategy-
tasks for which the most obvious and natural focused interventions when they are about half-
way of proceeding was not the optimum task way finished with the work—for example, when
performance strategy. they have consumed half of their available re-
To test their hypothesis, the researchers struc- sources, have progressed halfway to their goal,
tured their experimental task in two different or have arrived at some other natural break
ways. In one version of the task, the most obvi- point in the work. At that time members are
ous and natural way of proceeding was, in fact, more likely than previously to welcome and be
optimum for team performance; in a second ver- helped by interventions that encourage them to
sion, that way of proceeding would introduce assess their work progress, to review how they
inefficiencies and result in suboptimal perfor- are applying members’ efforts and talents to the
mance. As expected, the “plunge right in” work, and to consider how they might want to
9. 2005 Hackman and Wageman 277
alter their task performance strategies to better how best to proceed with their work. This con-
align them with external requirements and in- clusion deals specifically with the timing of ex-
ternal resources. ternal interventions intended to foster team
The increased receptivity to coaching inter- strategy planning. When work teams spontane-
ventions that encourage reflection on team work ously engage in planning activities in their ini-
strategies by teams that have logged some ex- tial team meetings, process management norms
perience, relative to those that are just starting sometimes emerge that are subsequently help-
out, is a tenet of developmental theories of team ful in pacing and coordinating team activities
coaching (e.g., Kozlowski, Gully, McHugh, Salas, (Janicik & Bartel, 2003).
& Cannon-Bowers, 1996) and is empirically sup- In both the Woolley (1998) study and the ex-
ported by the findings of Woolley (1998), men- periment by Hackman et al. (1976) described ear-
tioned earlier. Woolley created an experimental lier, it was difficult but possible to introduce a
version of an architectural task that involved strategy-focused intervention at the beginning
constructing a model of a college residence hall of a task cycle. In other cases it is impossible to
out of LEGO® bricks. Groups were informed in do so. Total quality management programs, for
advance how the structures they created would example, involve use of such techniques as Pa-
be evaluated (criteria included sturdiness, aes- reto analyses, control charts, and cost-of-quality
thetics, and various technical indices). She de- analyses to develop improved production strat-
vised two coaching-type interventions— one in- egies (for details, see Hackman & Wageman,
tended to improve members’ interpersonal 1995). These techniques simply cannot be used
relationships and one that provided assistance until a record of experience with existing strat-
to the team in developing a task-appropriate egies has been amassed— once again affirming
performance strategy. Each team received only that consultative coaching is more appropri-
one intervention, administered either at the be- ately provided around the middle of a task cycle
ginning or at the midpoint of its work period. than at its beginning.
Woolley’s findings, shown in Figure 1, confirm
that strategy interventions are especially help-
Ends
ful when they come near the midpoint of a
team’s work cycle. When the strategy interven- The third special opportunity for coaching oc-
tion was provided at the beginning of the work curs at the end of a performance period, when
period, it did not help, further affirming that the work is finished or a significant subtask has
members need to log some experience before been accomplished (Kozlowski, Gully, Salas, &
they are ready to have a useful discussion of Cannon-Bowers, 1996). The end of a task cycle is
FIGURE 1
Woolley’s Findings About the Focus and Timing of Coaching
Adapted from Woolley (1998).
10. 278 Academy of Management Review April
the time when a team has as much data as it is can result in having leaders assigned a dispro-
likely to get for members to use in exploring portionate share of the credit for team successes
what can be learned from the collective work and minority members a disproportionate share
just completed. Moreover, a team is likely to be of the blame for team failures (Smith & Berg,
far more ready at this point than previously to 1987: Chapter 4). Competent, well-timed coach-
capture and internalize the lessons that can be ing can help members work through such im-
learned from their work experiences, for several pulses and generate collective learning that
reasons. The anxieties that invariably surface in strengthens the team’s capabilities as a task-
getting a piece of work organized, executed, and performing unit.
finished on time dissipate once that work is
completed—significant, because people do not
learn well when brimming with anxieties (Ed- Summary
mondson, 1999; Zajonc, 1965). Moreover, once the
We have seen that work teams are especially
task has been finished, there often is time for
open to coaching interventions at three times in
reflection, which typically is in short supply in
the group life cycle: (1) at the beginning, when a
the rush to completion. The postperformance pe-
group is engaging with its task; (2) at the mid-
riod, therefore, is an especially inviting time for
point, when half the allotted time has passed
educational coaching interventions (see also
and/or half the work has been done; and (3) at
Blickensderfer, Cannon-Bowers, & Salas, 1997;
the end, when a piece of work has been finished.
Butler, 1993; Ellis, Mendel, Nir, & Davidi, 2002).
Moreover, each of the three coaching functions
Such interventions not only build the team’s res-
discussed in the previous section of this article
ervoir of talent, which increases its performance
is uniquely appropriate at one of those three
capabilities for subsequent tasks, but also con-
times: motivational coaching at beginnings,
tribute directly to the personal learning of indi-
consultative coaching at midpoints, and educa-
vidual team members.
tional coaching at ends.
Absent coaching interventions, team mem-
The time dependence of the coaching func-
bers are not likely to take initiatives after the
tions is summarized in our second proposition.
work has been completed to capture and inter-
nalize the lessons that could be learned from Proposition 2: Each of the three coach-
their experiences. If the team has succeeded, ing functions has the greatest con-
members may be more interested in celebrating structive effect at specific times in the
than in reflecting, and if it has not, they may be team task cycle. Specifically, (a) moti-
driven more to rationalize why the failure was vational coaching is most helpful
not really their fault than to explore what might when provided at the beginning of a
be learned from it. Moreover, even if members performance period, (b) consultative
do take the time to reflect on possible explana- coaching is most helpful when pro-
tions for the team’s level of performance, coach- vided at the midpoint of a perfor-
ing may be required to bring those explanations mance period, and (c) educational
into alignment with reality. coaching is most helpful when pro-
In a field study of team attribution-making vided after performance activities
processes, Corn (2000) collected a diverse sam- have been completed.
ple of task-performing teams in organizations,
some of which had performed well and some of This proposition could be viewed as suggest-
which had not, and asked members of each ing that coaching is irrelevant or ineffectual
team to explain why the team had performed as during the great majority of a team’s time—that
it did. The majority of those explanations fo- is, in the extended periods that lie between its
cused on the behaviors or dispositions of indi- beginning, midpoint, and end. It is true that
vidual members or of the team leader. This at- teams are remarkably impervious to interven-
tributional bias diverts members’ attention from tions, made during times of low readiness, that
the ways that less salient structural or contex- seek to alter their established trajectories. Even
tual factors may have shaped their interaction so, it often is possible for coaching to make
and performance. Moreover, it invites the psy- small but significant contributions to the team
chodynamic phenomenon of “splitting,” which and its work during between-marker periods.
11. 2005 Hackman and Wageman 279
One such contribution is to help members co- coaching interventions are abundant for such
ordinate their activities and thereby minimize teams, because they experience multiple begin-
the risks of tacit coordination identified by Wit- nings, midpoints, and ends.
tenbaum, Vaughan, and Stasser (1998). Leaders Still other teams operate continuously, with-
who take on too much of the responsibility for out any official beginnings or ends whatsoever.
coordination do run the risk of becoming so in- Many industrial production teams, for example,
volved in the actual work that they overlook keep on turning out the same products month
opportunities to help the team develop into an after month, indefinitely into the future. Even
increasingly competent performing unit. And teams with continuous tasks, however, usually
team members may eventually abdicate to the have beginnings, midpoints, and ends. They
leader their own responsibilities for managing have them because the teams, or their manag-
team performance processes. Still, when task ers, create them.
complexity is very high and/or when members There appears to be a human compulsion to
are relatively inexperienced, helping members create temporal markers, even when no such
coordinate their activities can be an appropriate markers are actually needed. In a semiconduc-
and helpful coaching intervention. tor manufacturing plant that operated continu-
A second kind of between-marker coaching ously all year (except for an annual holiday
that can be helpful to work teams is the use of break when the entire plant closed down), for
operant techniques to reinforce constructive but instance, production activities were organized
infrequently observed team behaviors, such as around six-week performance periods, and team
verbally reinforcing a team for exhibiting good
dynamics were highly responsive to the begin-
work processes (Komaki et al., 1989). Consistent
nings, midpoints, and ends of these entirely ar-
with the tenets of the operant model of team
bitrary temporal markers (Abramis, 1990; Hack-
coaching summarized earlier, these initiatives
man, 2002). The creation of quarters to demark
can both increase the frequency of desirable
financial reporting periods and semesters (or
behaviors and decrease the frequency of unde-
quarters) to organize educational activities in
sirable behaviors. However, such initiatives
schools have the same character: they are arbi-
may be especially helpful for teams whose ba-
trary but nonetheless powerful in shaping the
sic performance processes—that is, their man-
agement of team effort, strategy, and talent— rhythm of collective activity.
are already strong. Even well-designed and Temporal rhythms are deeply rooted in hu-
well-executed reinforcement of desirable team man experience, and if temporal markers are
behaviors cannot compensate for the absence of not naturally provided (e.g., through human bi-
well-timed motivational, consultative, and edu- ology or seasonal cycles), we create them and
cational coaching interventions. then use them to pace our activities. Such mark-
Most of the research evidence supporting the ers identify the beginnings, midpoints, and ends
proposition that teams are especially open to of team life cycles, and they thereby create op-
motivational, consultative, and educational in- portunities for coaching interventions that oth-
terventions at their beginnings, midpoints, and erwise would not exist.
ends comes from teams that have a single task
to complete by a specified deadline. The task
cycles of such teams are coincident with their CONDITIONS
life cycles, and, therefore, temporal markers are
easy to identify. For many work teams in orga- The impact of motivational, consultative, and
nizations, however, tasks and deadlines are not educational coaching depends not just on the
so clear and well defined. For example, some time in the task cycle at which interventions are
teams have multiple tasks to perform, perform made but also on the degree to which two other
the same task multiple times, or have work that conditions are in place. The first of these condi-
requires members to manage multiple task cy- tions is the degree to which key performance
cles simultaneously in the service of larger per- processes are externally constrained, and the
formance goals (Marks, Mathieu, & Zaccaro, second is the degree to which the group itself is
2001). Opportunities for temporally appropriate a well-designed performing unit.
12. 280 Academy of Management Review April
Task and Organizational Constraints circumstances the relationship between the
team’s utilization of member talent and team
Not all of the coaching functions specified
performance is severely restricted.
here are salient for all team tasks, since for
In some organizational circumstances all
some tasks only one or two of the three perfor-
three of the performance processes are uncon-
mance processes drive performance outcomes.
strained, and all three, therefore, are salient in
For an arithmetic task with a self-confirming
affecting performance outcomes. Consider, for
answer, for example, the performance of a team
example, the work of product development
that has accepted and engaged with its task is
teams. The pace of the work is largely at the
almost entirely a function of the knowledge and
discretion of the team, performance procedures
skill its members apply in their work. In con- are mostly unprogrammed, and the work re-
trast, performance on a simple, self-paced pro- quires use of complex skills to deal with consid-
duction task, such as moving materials from one erable uncertainty in the environment. Motiva-
place to another, is almost entirely a function of tional, consultative, and educational coaching
the level of effort members expend. interventions, if competently provided, all can
Thus, if one of the three performance pro- be helpful in fostering the performance effec-
cesses is constrained (i.e., if variation in that tiveness of such teams.
process is controlled or limited by the task or the In other circumstances some performance pro-
organization), then any attempts by members to cesses are constrained and others are not. Sur-
manage that particular process will be ineffec- gical teams are one example (Edmondson, Boh-
tual. If, however, a performance process is not mer, & Pisano, 2001). There is little constraint
constrained, then how well members manage regarding the use of knowledge and skill by
that process can substantially affect their team members but moderate constraint on both
team’s eventual performance. strategy (some but not all procedures are pro-
The salience of effort is constrained by the grammed) and effort (some but not all task in-
degree to which work inputs are under external puts derive from the nature of the surgical pro-
control. When the arrival of the materials a team cedure and the response of the patient as the
processes is controlled externally (e.g., by cus- operation progresses). Finally, there are some
tomer demand or machine pacing), a team can circumstances in which all three performance
only respond to whatever it receives and will be processes are constrained, as for a team work-
unable to increase its output by working espe- ing on a mechanized assembly line where in-
cially hard. In such circumstances the relation- puts are machine paced, assembly procedures
ship between team effort and performance is are completely programmed, and performance
severely restricted. operations are simple and predictable. A team
The salience of strategy is constrained by the assigned such a task would be a team in name
degree to which performance operations are ex- only, since performance would depend so little
ternally determined. When work procedures are on how members interacted.
completely prespecified (e.g., by mechanical re- Teams can be helped by coaching interven-
quirements or by a manual that specifies ex- tions that focus specifically on reducing process
actly how the work is to be done), a team has losses and/or on fostering process gains only for
little latitude to develop a new or better task those aspects of team performance processes
performance strategy. In such circumstances the that are relatively unconstrained. Coaching in-
relationship between team performance strate- terventions that address team processes that
gies and performance is severely restricted. are substantially constrained will be ineffec-
Finally, the salience of knowledge and skill is tual, since they seek to improve team processes
constrained by the degree to which performance that are not salient for how well the team per-
operations are simple and predictable (versus forms. Such interventions can even compromise
complex and unpredictable). When task perfor- performance because they consume members’
mance requires the use of skills that are well time and direct their attention away from more
learned in the general population on tasks that salient aspects of their interaction.
are well understood, a team is unable to im-
prove its performance by bringing additional Proposition 3: Coaching interventions
knowledge or skill to bear on the work. In such are helpful only when they address
13. 2005 Hackman and Wageman 281
team performance processes that are (Bettenhausen & Murnighan, 1985; Gersick,
salient for a given task; those that ad- 1988), and they tend to remain in place until and
dress nonsalient processes are, at best, unless something fairly dramatic occurs to force
ineffectual. a rethinking about what behaviors are and are
not appropriate (Gersick & Hackman, 1990; Louis
& Sutton, 1991). When those upfront norms of
Group Design
conduct actively promote continuous scanning
Certain features of a team’s design, including of the performance situation and proactive plan-
properties of the social system context within ning of how members will work together, they
which it operates, can negate the impact of facilitate the development of task performance
coaching interventions, even those that are well strategies that are appropriate for the team’s
executed and that address appropriate team task and situation (Hackman et al., 1976; Wool-
performance processes. Moreover, design fea- ley, 1998). The appropriateness of a team’s per-
tures can exacerbate the effects of good and formance strategies also depends, however, on
poor coaching on team effectiveness, heighten- the degree to which the organizational informa-
ing the benefits of good coaching and making tion system makes available to the team what-
even worse the problems brought about by poor ever data and projections members may need to
coaching. select or invent ways of proceeding that are well
Each of the three performance processes that tuned to their circumstances (Abramis, 1990; Bik-
form the core of our model—the level of effort the son, Cohen, & Mankin, 1999).
team expends on its task, the appropriateness of The level of knowledge and skill a team
its performance strategies, and the amount of brings to bear on its work is influenced by the
knowledge and skill it applies to the work—is composition of the team (a structural feature)
shaped not only by coaching interventions but and by the organizational education system (a
also by how well a team is structured and by the contextual support). Well-composed teams have
level of contextual support provided (Hackman, members who bring to the group a rich array of
Wageman, Ruddy, & Ray, 2000). task-relevant knowledge and skills, and they
The effort a team expends on its work is influ- are structured so that members’ talents can be
enced by the design of its task (a structural drawn on readily in pursuing team purposes.
feature) and by the reward system of the orga- Such teams are as small as possible, given the
nization in which the team operates (a contex- work to be accomplished, they include members
tual support). A motivating team task is a whole who have appropriate skills, and they have a
and meaningful piece of work for which mem- good mix of members—people who are neither
bers share responsibility and accountability so similar to one another that they are like peas
and one that is structured so that members re- from the same pod nor so different that they risk
ceive regular and trustworthy data about how having difficulty communicating and coordinat-
they are doing. Well-designed team tasks foster ing with one another (Ancona & Caldwell, 1992;
high, task-focused effort by team members Campion, Medsker, & Higgs, 1993; Druskat, 1996;
(Hackman & Oldham, 1980). Team effort is en- Goodman & Shah, 1992; Jackson, 1992).
hanced by organizational reward systems that Even well-composed teams, however, may not
recognize and reinforce team excellence—and have within their boundaries all of the talent
that avoid the common, if usually unintended, required for excellent performance. Organiza-
problem of providing disincentives for collabo- tional education systems can supplement inter-
ration among team members by placing them in nal resources by making available to teams, at
competition with one another for individual re- the teams’ initiative, technical or educational
wards (Wageman, 1995). assistance for any aspects of the work for which
The task appropriateness of a team’s perfor- members are not already knowledgeable,
mance strategy is influenced by its core norms skilled, or experienced, including, if necessary,
of conduct (a structural feature) and by the or- assistance honing members’ skills in working
ganizational information system (a contextual together on collective tasks (Stevens & Yarish,
support). Collective expectations about accept- 1999).
able behavior are either “imported” to the group Research evidence clearly establishes the pri-
by members or established very early in its life ority of structural and contextual features over
14. 282 Academy of Management Review April
coaching behaviors as influences on team per- When, then, can coaching make a constructive
formance processes and outcomes (Cohen, Led- difference in team performance processes? We
ford, & Spreitzer, 1996). For example, Wageman propose that coaching makes relatively small
(2001) found, in a study of field service teams at adjustments to an already defined trajectory.
the Xerox Corporation, that team design fea- When a team’s performance situation is favor-
tures, including those described just above, con- able, competent coaching can be helpful to
trolled significantly more variance in both the members in minimizing process losses and cre-
level of team self-management and in perfor- ating process gains. When a team’s structure is
mance effectiveness than did team leaders’ flawed and/or its context is unsupportive, how-
coaching behaviors. For team self-management, ever, even competent process-focused coaching
design features controlled 42 percent of the vari- may do more harm than good.
ance, compared to less than 10 percent for mea-
Proposition 4: Competent coaching in-
sures that assessed the quality of leaders’
terventions (i.e., those that foster col-
coaching activities; for team performance, de-
lective effort, task-appropriate perfor-
sign controlled 37 percent of the variance, com-
mance strategies, and good use of
pared to less than 1 percent for coaching. These
member knowledge and skill) are
findings are consistent with other evidence,
more beneficial for groups that are
cited earlier, showing that even highly compe-
well structured and supported than for
tent process-focused coaching by team leaders
those that are not; poor coaching in-
or consultants rarely generates substantial or
terventions (i.e., those that subvert
enduring improvements in team processes or
team performance processes) are more
performance, and with the more general finding
detrimental for teams that are poorly
that coaching cannot prevail against strong
structured and supported than for
structural and contextual forces (Hackman,
those that are well designed.
1987). It is nearly impossible to coach a team to
greatness in a performance situation that under- Further findings from the Wageman (2001)
mines rather than supports teamwork. study described above provide evidence in sup-
FIGURE 2
How Team Design and Leader Coaching Jointly Affect Team Self-Management
15. 2005 Hackman and Wageman 283
port of this proposition. As seen in the left panel CONCLUSION
of Figure 2, competent coaching (e.g., conduct-
In the present model we posit that team
ing a problem-solving process) helped well-
coaching can foster team effectiveness only
designed teams exploit their favorable circum-
when four conditions are present. Two of these
stances but made little difference for poorly
conditions have to do with organizational cir-
designed teams. Poor coaching (e.g., identifying
cumstances and two with coaches’ actions.
a team’s problems and telling members how
they should solve them), in contrast, was much
more deleterious for poorly designed teams than 1. The group performance processes that are
for those that had an enabling team structure key to performance effectiveness (i.e., effort,
strategy, and knowledge and skill) are rela-
and a supportive organizational context (right tively unconstrained by task or organization-
panel of Figure 2). al requirements.
The interaction between a team’s design and 2. The team is well designed and the organi-
the efficacy of coaching interventions may help zational context within which it operates
explain the finding from “brainstorming” re- supports rather than impedes team work.
3. Coaching behaviors focus on salient task
search that the pooled ideas of individuals performance processes rather than on mem-
working alone generally exceed in both quan- bers’ interpersonal relationships or on pro-
tity and quality the product of interacting brain- cesses that are not under the team’s control.
storming groups (Taylor, Berry, & Block, 1957). 4. Coaching interventions are made at times
Studies of brainstorming typically use either ad when the team is ready for them and able to
deal with them—that is, at the beginning for
hoc groups created especially for research pur- effort-related (motivational) interventions,
poses (e.g., Cohen, Whitmyre, & Funk, 1960) or near the midpoint for strategy-related (con-
existing organizational groups whose members sultative) interventions, and at the end of a
are asked to take time from their regular work to task cycle for (educational) interventions
participate in the research (e.g., Osborn, 1963). that address knowledge and skill.
On the one hand, if the participating teams’
design features are suboptimal—not unlikely When these four conditions are present, skill-
for ad hoc or serendipitously obtained groups— fully provided coaching can yield substantial
it would not be surprising to find that brain- and enduring improvements in team effective-
storming fails to facilitate creative team perfor- ness. Yet these conditions are not commonly
mance. For teams that have enabling structures found in traditionally designed and managed
and supportive contexts, on the other hand, this work organizations. Organizational work de-
particular coaching intervention—as well as signs often constrain one or more of the three
others that require teams to be able to use non- performance processes that drive team perfor-
traditional and unfamiliar group process tools— mance; organizational systems do not provide
might well generate substantial performance the supports that work teams need; and coaches,
benefits. when trained at all, are taught the leadership
In sum, even competent coaching is unlikely styles preferred by trainers rather than helped
to be of much help to groups that have poor to learn how to provide well-timed and appro-
structures and/or unsupportive organizational priately focused interventions using their own
contexts. Favorable performance situations, preferred styles (Hackman, 2002; Hackman &
however, can yield a double benefit: teams are Walton, 1986).
likely to have less need for coaching (because One could conclude, therefore, that few schol-
they encounter fewer problems that lie beyond arly resources should be expended on research
their own capabilities), and the coaching that on team coaching because it is of so little con-
they do receive is likely to be more helpful to sequence. Moreover, one could view the reports
them (because they are not preoccupied with from the field (cited in the introduction to this
more basic, structurally rooted difficulties). Over paper) that team leaders spend less time on
time, such teams may become skilled at coach- team coaching than on any other category of
ing themselves and may even enter into a self- leader behavior as a sign of team leaders’ wis-
fueling spiral of ever-increasing team capabil- dom. Rather than spend time on an activity that
ity and performance effectiveness (Lindsley, so rarely makes a difference, leaders might be
Brass, & Thomas, 1995). better advised to focus on aspects of their lead-
16. 284 Academy of Management Review April
ership portfolio for which there is a greater re- organizational circumstances within which
turn from effort expended. their teams operate. It is highly doubtful that
We believe such conclusions would be too any single laboratory or field study could com-
pessimistic—for scholars and practitioners prehensively assess all the propositions of the
alike. Scholars with an interest in senior execu- present model. But studies that test individual
tive leadership have for many years debated propositions by creating specific coaching inter-
just how much of a difference CEOs make in the ventions and documenting their effects in con-
performance of their firms. These disputes, texts that are thoughtfully created or selected
which probably can never be resolved empiri- can, over time, provide the knowledge required
cally, have now given way to a more tractable to correct and refine the model.
question—namely, under what conditions does The challenges for coaching practitioners are
senior leadership matter (Wasserman, Anand, & just as great as for scholars, and for the same
Nohria, 2001)? We suggest that a similar refram- reasons. Rather than simply taking as given the
ing of research questions about team coaching circumstances in which their teams operate,
may be warranted. That is, instead of asking, practitioners who lead work teams should give
“How much difference does team coaching first priority to determining whether the basic
make?” scholars might more productively ex- structural and contextual conditions that foster
pend resources in further research on the struc- team effectiveness are in place (Hackman, 2002).
tural and contextual conditions under which If they are not, team leaders would be well ad-
competent team coaching does (and does not) vised to exercise influence with their own peers
significantly affect team performance. and supervisors to create those conditions, and
The conduct of such research poses several thereby to make competent team coaching pos-
significant challenges. For one thing, it is far sible. We hope the model of team coaching set
from straightforward to measure leader behav- forth in this article is of some use to practition-
iors and group processes reliably as they unfold ers in orienting and prioritizing such initiatives,
in real time in fluid organizational circum- as well as to scholars in conducting informative
stances. And since the effects of team coaching research about work teams and the behaviors of
are determined jointly by factors that exist at those who lead them.
multiple levels of analysis (i.e., the organization-
al, team, and individual levels), it is necessary REFERENCES
to locate or create research settings where there
Abramis, D. J. 1990. Semiconductor manufacturing team. In
is ample variation at all three levels (Hackman, J. R. Hackman (Ed.), Groups that work (and those that
2003). Studies of coaching effectiveness cannot don’t): 449 – 470. San Francisco: Jossey-Bass.
simply take as given whatever structural and Alderfer, C. P. 1977. Group and intergroup relations. In J. R.
contextual features are commonly found either Hackman & J. L. Suttle (Eds.), Improving life at work:
in the experimental laboratory or in organiza- 227–296. Santa Monica, CA: Goodyear.
tional life. Ancona, D. G., & Caldwell, D. F. 1992. Demography and
Indeed, it may be necessary in research on design: Predictors of new product team performance.
team coaching, as sometimes must be done in Organization Science, 3: 321–341.
other scientific fields, such as medical research Ancona, D. G., & Chong, C. L. 1999. Cycles and synchrony:
and subatomic physics, to first create one’s phe- The temporal role of context in team behavior. In
R. Wageman (Ed.), Groups in context: 33– 48. Stamford,
nomenon of interest before conducting research
CT: JAI Press.
on its dynamics (Argyris, 1969). Research in lab-
Argyris, C. 1969. The incompleteness of social psychological
oratory settings, for example, could involve the
theory: Examples from small group, cognitive consis-
construction and administration of model- tency, and attribution research. American Psychologist,
specified coaching interventions at times either 24: 893–908.
consistent or inconsistent with the model’s prop- Argyris, C. 1982. Reasoning, learning, and action. San Fran-
ositions and then assessment of the conse- cisco: Jossey-Bass.
quences for group dynamics and performance. Argyris, C. 1993. Education for leading-learning. Organiza-
And action research in field settings could as- tional Dynamics, 21(3): 5–17.
sess the impact of educational programs in- Bales, R. F., & Strodtbeck, F. L. 1951. Phases in group problem
tended to help team coaches design interven- solving. Journal of Abnormal and Social Psychology, 46:
tions specifically tailored to the task and 485– 495.
17. 2005 Hackman and Wageman 285
Bettenhausen, K., & Murnighan, J. K. 1985. The emergence of Geber, B. 1992. From manager into coach. Training, Febru-
norms in competitive decision-making groups. Adminis- ary: 25–31.
trative Science Quarterly, 30: 350 –372.
Gersick, C. J. G. 1988. Time and transition in work teams:
Bikson, T. K., Cohen, S. G., & Mankin, D. 1999. Information Toward a new model of group development. Academy of
technology and high-performance teams. In E. Sund- Management Journal, 31: 9 – 41.
strom (Ed.), Supporting work team effectiveness: 215–
Gersick, C. J. G. 1989. Marking time: Predictable transitions
245. San Francisco: Jossey-Bass.
in task groups. Academy of Management Journal, 31:
Blickensderfer, E., Cannon-Bowers, J. A., & Salas, E. 1997. 9 – 41.
Training teams to self-correct: An empirical investiga-
Gersick, C. J. G., & Hackman, J. R. 1990. Habitual routines in
tion. Paper presented at the annual meeting of the So-
task-performing teams. Organizational Behavior and
ciety for Industrial and Organizational Psychology, St.
Human Decision Processes, 47: 65–97.
Louis.
Ginnett, R. C. 1993. Crews as groups: Their formation and
Butler, R. E. 1993. LOFT: Full-mission simulation as crew
their leadership. In E. L. Wiener, B. G. Kanki, & R. L.
resource management training. In E. L. Wiener, B. G.
Helmreich (Eds.), Cockpit resource management: 71–98.
Kanki, & R. L. Helmreich (Eds.), Cockpit resource man-
Orlando, FL: Academic Press.
agement: 231–259. Orlando, FL: Academic Press.
Goodman, P. S., & Shah, S. 1992. Familiarity and work group
Campion, M. A., Medsker, G. J., & Higgs, A. C. 1993. Relations
outcomes. In S. Worchel, W. Wood., & J. Simpson (Eds.),
between work group characteristics and effectiveness:
Group process and productivity: 276 –298. London: Sage.
Implications for designing effective work groups. Per-
sonnel Psychology, 46: 823– 850. Guzzo, R. A., Wagner, D. B., Maguire, E., Herr, B., & Hawley, C.
1986. Implicit theories and the evaluation of group pro-
Chang, A., Bordia, P., & Duck, J. 2003. Punctuated equilibrium
and linear progression: Toward a new understanding of cess and performance. Organizational Behavior and Hu-
group development. Academy of Management Journal, man Decision Processes, 37: 279 –295.
46: 106 –117. Hackman, J. R. 1987. The design of work teams. In J. W. Lorsch
Cohen, D., Whitmyre, J., & Funk, W. 1960. Effect of group (Ed.), Handbook of organizational behavior: 315–342.
cohesiveness and training upon creative thinking. Jour- Englewood Cliffs, NJ: Prentice-Hall.
nal of Applied Psychology, 44: 319 –322. Hackman, J. R. 2002. Leading teams: Creating conditions for
Cohen, S. G., Ledford, G. E., Jr., & Spreitzer, G. M. 1996. A great performances. Boston: Harvard Business School
predictive model of self-managing work team effective- Press.
ness. Human Relations, 49: 643– 676. Hackman, J. R. 2003. Learning more from crossing levels:
Corn, R. 2000. Why poor teams get poorer. The influence of Evidence from airplanes, orchestras, and hospitals.
team effectiveness and design quality on the quality of Journal of Organizational Behavior, 24: 1–18.
group diagnostic processes. Unpublished doctoral dis- Hackman, J. R., Brousseau, K. R., & Weiss, J. A. 1976. The
sertation, Harvard University, Cambridge, MA. interaction of task design and group performance strat-
Druskat, V. U. 1996. Team-level competencies in superior egies in determining group effectiveness. Organiza-
performing self-managing work teams. Paper presented tional Behavior and Human Performance, 16: 350 –365.
at the annual meeting of the Academy of Management, Hackman, J. R., & Morris, C. G. 1975. Group tasks, group
Cincinnati, OH. interaction process, and group performance effective-
Eden, D. 1985. Team development: A true field experiment at ness: A review and proposed integration. In L. Berkowitz
three levels of rigor. Journal of Applied Psychology, 70: (Ed.), Advances in experimental social psychology, vol.
94 –100. 8: 45–99. New York: Academic Press.
Edmondson, A. E. 1999. Psychological safety and learning Hackman, J. R., & Oldham, G. R. 1980. Work redesign. Read-
behavior in work teams. Administrative Science Quar- ing, MA: Addison-Wesley.
terly, 44: 350 –383. Hackman, J. R., & Wageman, R. 1995. Total quality manage-
Edmondson, A. E., Bohmer, R. M., & Pisano, G. P. 2001. Dis- ment: Empirical, conceptual, and practical issues. Ad-
rupted routines: Team learning and new technology im- ministrative Science Quarterly, 40: 309 –342.
plementation in hospitals. Administrative Science Quar- Hackman, J. R., Wageman, R., Ruddy, T. M., & Ray, C. R. 2000.
terly, 46: 685–716. Team effectiveness in theory and practice. In C. Cooper
Ellis, S., Mendel, R., Nir, M., & Davidi, I. 2002. After-event & E. A. Locke (Eds.), Industrial and organizational psy-
reviews: Drawing lessons from successful vs. failed ex- chology: Theory and practice: 109 –129. Oxford: Black-
perience. Working paper, Recanati Graduate School of well.
Business Administration, Tel Aviv University, Tel Aviv. Hackman, J. R., & Walton, R. E. 1986. Leading groups in
Fischer, N. 1993. Leading self-directed work teams: A guide to organizations. In P. S. Goodman (Ed.), Designing effec-
developing new team leadership skills. New York: tive work groups: 72–119. San Francisco: Jossey-Bass.
McGraw-Hill. Hirokawa, R. Y. 1985. Discussion procedures and decision-
Fournies, F. F. 1978. Coaching for improved work perfor- making performance: A test of a functional perspective.
mance. Bridgewater, NJ: Van Nostrand Reinhold. Human Communication Research, 12: 203–224.
18. 286 Academy of Management Review April
Hollingshead, A. B., Wittenbaum, G. M., Paulus, P. B., Hiro- From habits of mind to active thinking. Human Rela-
kawa, R. Y., Peterson, R. S., Jehn, K. A., & Yoon, K. In tions, 44: 55–76.
press. A look at groups from the functional perspective. Mann, J. B. 2001. Time for a change: The role of internal and
In A. B. Hollingshead & M. S. Poole (Eds.), Theories of external pacing mechanisms in prompting the midpoint
small groups: An interdisciplinary perspective. Thou- transition. Unpublished honors thesis, Department of
sand Oaks, CA: Sage. Psychology, Harvard University, Cambridge, MA.
Jackson, S. 1992. Team composition in organizations. In Manz, C. C., & Sims, H. P. 1987. Leading workers to lead
S. Worchel, W. Wood, & J. Simpson (Eds.), Group process themselves: The external leadership of self-managing
and productivity: 138 –173. London: Sage. work teams. Administrative Science Quarterly, 32: 106 –
Janicik, G. A., & Bartel, C. A. 2003. Talking about time: Effects 128.
of temporal planning and time awareness norms on Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. 2001. A temporally
group coordination and performance. Group Dynamics, based framework and taxonomy of team processes.
7(2): 122–134. Academy of Management Review, 26: 356 –376.
Jehn, K. A., & Mannix, E. A. 2001. The dynamic nature of McGrath, J. E. 1962. Leadership behavior: Some requirements
conflict: A longitudinal study of intragroup conflict and for leadership training. Washington, DC: U.S. Civil Ser-
group performance. Academy of Management Journal, vice Commission.
44: 238 –251.
McGrath, J. E., & Kelly, J. R. 1986. Time and human interac-
Kaplan, R. E. 1979. The conspicuous absence of evidence that tion: Toward a social psychology of time. New York:
process consultation enhances task performance. Jour- Guilford Press.
nal of Applied Behavioral Science, 15: 346 –360.
Moreland, R. L., & Levine, J. M. 1988. Group dynamics over
Kernaghan, J. A., & Cooke, R. A. 1990. Teamwork in planning time: Development and socialization in small groups. In
innovative projects: Improving group performance by J. E. McGrath (Ed.), The social psychology of time: New
rational and interpersonal interventions in group pro- perspectives: 151–181. Newbury Park, CA: Sage.
cess. IEEE Transactions on Engineering Management, Orlitzky, M., & Hirokawa, R. Y. 2001. To err is human, to
37: 109 –116. correct is divine: A meta-analysis of research testing the
Kinlaw, D. C. 1991. Developing superior work teams. San functional theory of group decision-making effective-
Diego: Lexington Books. ness. Small Group Research, 32: 313–341.
Komaki, J. L. 1986. Toward effective supervision: An operant Osborn, A. 1963. Applied imagination (3rd ed.). New York:
analysis and comparison of managers at work. Journal Scribner.
of Applied Psychology, 71: 270 –279. Patten, T. H. 1981. Organizational development through
Komaki, J. L. 1998. Leadership from an operant perspective. team-building. New York: Wiley.
New York: Routledge. Rees, F. 1991. How to lead work teams. San Francisco: Jossey-
Komaki, J. L., Deselles, M. L., & Bowman, E. D. 1989. Definitely Bass/Pfeiffer.
not a breeze: Extending an operant model of effective Salas, E., Rozell, D., Mullen, B., & Driskell, J. E. 1999. The effect
supervision to teams. Journal of Applied Psychology, 74: of team building on performance: An integration. Small
522–529. Group Research, 30: 309 –329.
Komaki, J. L., & Minnich, M. R. 2002. Crosscurrents at sea: The Schein, E. H. 1969. Process consultation: Its role in organiza-
ebb and flow of leaders in response to the shifting de- tion development. Reading, MA: Addison-Wesley.
mands of racing sailboats. Group and Organization
Schein, E. H. 1988. Process consultation, vol. 1. Reading, MA:
Management, 27: 113–141.
Addison-Wesley.
Kozlowski, S. W. J., Gully, S. M., McHugh, P. P., Salas, E., &
Schwarz, R. 1994. Team facilitation. Englewood Cliffs, NJ:
Cannon-Bowers, J. A. 1996. A dynamic theory of leader-
Prentice-Hall.
ship and team effectiveness: Developmental and task
contingent leader roles. In G. R. Ferris (Ed.), Research in Seers, A., & Woodruff, S. 1997. Temporal pacing in task forc-
personnel and human resource management, vol. 14: es: Group development or deadline pressure. Journal of
253–305. Greenwich, CT: JAI Press. Management, 23: 169 –187.
Kozlowski, S. W. J., Gully, S. M., Salas, E., & Cannon-Bowers, Smith, K. K., & Berg, D. N. 1987. Paradoxes of group life. San
J. A. 1996. Team leadership and development: Theory, Francisco: Jossey-Bass.
principles, and guidelines for training leaders and Smith, R. E., Smoll, F. L., & Curtis, B. 1979. Coach effective-
teams. In M. Beyerlein, D. Johnson, & S. Beyerlein (Eds.), ness training: A cognitive-behavioral approach to en-
Advances in interdisciplinary studies of work teams: hancing relationship skills in youth sport coaches. Jour-
Team leadership, vol. 3: 251–289. Greenwich, CT: JAI nal of Sport Psychology, 1: 59 –75.
Press.
Smoll, F. L., & Smith, R. E. 1989. Leader behaviors in sport: A
Lindsley, D. H., Brass, D. J., & Thomas, J. B. 1995. Efficacy- theoretical model and research paradigm. Journal of
performance spirals: A multilevel perspective. Academy Applied Social Psychology, 19: 1522–1551.
of Management Review, 20: 645– 678.
Staw, B. M. 1975. Attribution of the “causes” of performance:
Louis, M. R., & Sutton, R. I. 1991. Switching cognitive gears: A general alternative interpretation of cross-sectional