Problems of Matrix Organizations
Problems of Matrix Organizations
Problems of Matrix
Organizations
by Stanley M. Davis and Paul R. Lawrence
FROM THE MAY 1978 ISSUE
What Is a Matrix?
founded on solid business reasons that will persist. The matrixs most basic advantage over
the familiar functional or product structure is that it facilitates a rapid management
response to changing market and technical requirements. Further, it helps middle
managers make trade-off decisions from a general management perspective.
Because the matrix is a relatively new form, however, the companies that have adopted it
have of necessity been learning on a trial and error basis. The mistakes as well as the
successes of these pioneers can be very informative to companies that follow their lead.
Here, we present some of the more common problems that occur when a company uses a
matrix form. For the sake of easy reference, we diagnose each pathology first, then discuss
its prevention and treatment. By using this format, however, we do not mean to suggest
that simple first-aid treatment of pathologies will cure them.
Through firsthand experience we know of only one organization that, using a latent
matrix form, quite literally came apart at the seams during a rather mild economic
recession. Following a fast-growth strategy, this company used its high stock multiple to
acquire, and then completely assimilate, smaller companies in the recreation equipment
field. Within a period of about six months the company changed from an exciting success
to a dramatic disaster. Its entire manufacturing, distribution, and financial systems went
out of control leaving unfilled orders, closed factories, distressed inventories, and huge
debts in their wake.
Of course, there are many possible reasons why this might have happened, but one
perfectly reasonable explanation is that the organization design failed under stress. What
was that design?
The product manager would find himself working across functional lines to try to
coordinate production schedules, inventories, cash flow, and distribution patterns without
any explicit and formal agreements about the nature of his relationships with the
functional managers. Because he was locked into his approved marketing plan, when sales
slipped behind schedule, his response was to exhort people to try harder rather than to cut
back on production runs.
But once one or two things began to crumble, there was not enough reserve in the system
to keep everything else from going wrong. As the product manager lost control, a power
vacuum developed, into which the functional managers fell, each grabbing for total
control. The result was that a mild recession triggered conditions approaching anarchy.
PreventionWe believe the lesson of this experience is loud and clear. Organizations
should not rely too much on an informal or latent matrix to coordinate critical tasks.
Relationships between functional and product managers should be explicit so that people
are in approximate agreement about who is to do what under various circumstances.
Properly used, a matrix does not leave such matters in an indefinite status; it is a definite
structure and not a free form organization.
A useful anarchy index is how many people in an organization do not recognize one boss
to whom they feel responsible for a major part of their work. In a study of five medical
schools, which are notoriously anarchical, the one with the most explicit matrix structure
was also the one with the least number of bossless people.1
TreatmentShould the worst happen and a company plunge into anarchy, true crisis
management would be the best response. The crisis response is really no mystery. The CEO
must pull all key people and critical information into the center. He or she must personally
make all important decisions on a round-the-clock schedule until the crisis is over. Then
and only then can he undertake the work of reshaping the organization so that it can
withstand any future shock such as a minor recession.
Power struggles
Managers jockey for power in many organizations, but a matrix design almost encourages
them to do so.
DiagnosisThe essence of a matrix is dual command. For such a form to survive there
needs to be a balance of power, where its locus seems to shift constantly, each party always
jockeying to gain an advantage. It is not enough simply to create the balance, but there
must also be continual mechanisms for checking the imbalances that creep in.
In business organizations that operate with a balance of power form, there is a constant
tendency toward imbalance. As long as each group or dimension in an organization tries to
maximize its own advantage vis--vis others, there will be a continual balancing struggle
for dominant power. A power struggle in a matrix is qualitatively different from that in a
traditionally structured hierarchy because in the latter it is clearly illegitimate. In the
matrix, however, power struggles are inevitable; the boundaries of authority and
responsibility overlap prompting people to maximize their own advantage.
PreventionMost top managers will find it exceedingly difficult to forestall all power
struggles. Equal strength on the part of the two parties, however, will prevent struggles
from reaching destructive heights. Friendly competition should be encouraged, but all-out
combat severely punished. Managers in a matrix should push for their advantages but
never with the intention of eliminating those with whom they share power, and always
with a perspective that encompasses both positions.
TreatmentThe best way to ensure that power struggles do not undermine the matrix is to
make managers on the power axes aware that to win power absolutely is to lose it
ultimately. These managers need to see that the total victory of one dimension only ends
the balance, finishes the duality of command, and destroys the matrix. They must see this
sharing of power as an underlying principle, before and during all of the ensuing and
inevitable power struggles.
Matrix managers have to recognize that they need worthy adversaries, counterparts who
can match them, to turn the conflict to constructive ends. For this successful outcome
three things are necessary.
First, matrix managers always have to maintain an institutional point of view, seeing their
struggles from a larger, shared perspective. Second, they have to jointly agree to remove
other matrix managers who, through weakness or whatever inability, are losing
irretrievable ground. And, third, that they replace these weak managers with the strongest
available peopleeven if to do so means placing very strong managers in weakened parts
of the organization and reversing their power initiatives.
Another key element in stopping power struggles before they get out of hand and destroy
the balance is the top level superior to whom the duelling managers report. Because of this
element, the matrix is a paradoxa shared-power system that depends on a strong
individual, one who does not share the authority that is delegated to him (say by the
board), to arbitrate between his power-sharing subordinates.
The top manager has many vehicles for doing this: the amount of time he spends with one
side of the matrix or the other, pay differentials, velocity of promotion, direct orders issued
to one dimension and not to the other, and so forth. What he must do above all, however,
is protect the weak dimension in the organization, not necessarily the weak manager in
charge of that dimension.
Severe groupitis
The mistaken belief that matrix management is the same as group decision making.
DiagnosisThe confusion of matrix behavior with group decision making probably arises
from the fact that a matrix often evolves out of new project or business teams, which do
suggest a group decision process. Under many circumstances, of course, it is perfectly
sensible for managers to make decisions in groups. But managers should expect difficulties
to arise if they believe group decision making to be the essence of matrix behavior.
We have seen one matrix organization that had a severe case of groupitis. This
multiproduct electronics company had a product manager and a product team, comprised
of specialists drawn from the ranks of every functional department, assigned to every
product. So far so good. But somehow the idea that the matrix structure requires that all
Many of the decisions that had to be made about each product involved detailed matters
with which only one or two people were regularly conversant. Yet all team members were
constrained to listen to these issues being discussed until a decision was made, and were
even expected to participate in the discussion and influence the choice. Some individuals
seemed to enjoy the steady diet of meetings and the chance to practice being a generalist.
However, a larger number of people felt that their time was being wasted and would have
preferred leaving the decisions to the most informed people. The engineers, in particular,
complained that the time they were spending in meetings was robbing them of
opportunities to strengthen their special competence and identities. As well as noting
these personal reactions, senior managers reported a general disappointment with the
speed and flexibility of organizational responses.
TreatmentIn the case of the multiproducts electronics company, the problem came to
light while we were researching the matrix approach. Once senior people had clearly
diagnosed the problem, it was 90% cured. Top management emphatically stated that there
was nothing sacred about group decisions and that it was not sensible to have all product
team members involved all the time. Once the line between individual and group matters
was drawn according to who had information really relevant to a decision, meetings
became fewer and smaller and work proceeded on a more economical and responsive
basis. The concept of teamwork was put in perspective: as often as necessary and as little
as possible.
However, if these companies follow the normal business cycle, there will be a period of two
to five years before they experience another economic crunch which is more than enough
time for the matrix concept to spread throughout a company. By that time the matrix
occupies a central place in company conversations and is a familiar part of these
organizations. Although there may still be some problems, the matrix seems there to stay.
When the down part of the economic cycle begins, senior management in these companies
may become appreciably bothered by the conflict between subordinates as well as by the
apparent slowness with which they respond to the situation. We need decisive action is
their rallying cry.
In an authoritarian structure top management can act quickly because it need not consider
the spectrum of opinion. Thinking there is no time for organizational toys and tinkering,
the top level managers take command in an almost, but not quite, forgotten way, and ram
their directives down the line. The matrix is done in.
PreventionTop management can prevent this kind of collapse of the matrix by employing
general managerial excellence, independent of the matrix, long before the crunch arrives.
Good planning, for example, can often forecast downturns in the economic cycle.
Corporate structures such as the matrix should not have to change because of standard
changes in the business cycle. When management planning has been poor, however, the
matrix is a readily available scapegoat.
Companies that experience severe economic crunches often make drastic changes in many
directions at once: trimming product lines, closing offices, making massive personnel and
budget cuts, and tightening managerial controls. The matrix is often done in during this
period for several reasons: it represents too great a risk; it never really worked properly
and giving it the coup de grace can disguise the failure of implementation; and the quality
of decision making had not improved performance sufficiently to counterbalance the hard
times. Measures management can take to prevent this pathology do not lie within the
matrix itself, as much as with improvements in basic managerial skills and planning.
A real estate and construction company provides an example of how a company can
anticipate and flexibly respond to an economic crunch that demonstrates the strength
rather than the weakness of the matrix. The company has developed a structure as well as
procedures that are especially well suited to the economic uncertainties of the business.
These include a set of fully owned subsidiaries each the equivalent of a functional
department in a manufacturing company and each the home base for varied specialists
needed to execute all phases of a major building project. The heads of the subsidiaries act
as chief salesmen for their various services, and often head up the bidding teams that put
together sophisticated proposals.
As a proposal project proceeds, the selected project manager is drawn into the team in
anticipation of securing the contract. This ensures an orderly transition to the project
management phase. The project office is given first-line responsibility for control of costs,
schedules, and quality of the project, but the top management team from the parent
company reviews the project regularly as a backup.
The company has used the matrix to advantage in weathering major shifts in both the
availability of business by market segment, for example, from schools to hospitals, and the
level of construction activity. It maintains a cadre of professional specialists and project
managers, who can be kept busy during the lows of the cycle, which it rapidly expands
during the highs by subcontracting for temporary services.
Excessive overhead
The fear of high costs associated with a matrix.
DiagnosisOn the face of it, a matrix organization would seem to double management costs
because of its dual chain of command. This issue deserves thoughtful consideration.
The limited amount of research on matrix overhead costs indicates that in the initial
phases overhead costs do in fact rise, but that, as a matrix matures, these extra costs
disappear and are offset by productivity gains.2 Our experience supports this finding. In a
large electronics company we observed in some detail how initial overhead increases not
only necessarily occur in a matrix but also how they can inflate unnecessarily. In this case,
the company decided to employ the matrix design from the outset in setting up its new
operating division at a new plant site.
This unique organizational experiment had a number of positive attributes, but one of its
problems was with overhead costs. In staffing the new division, top management filled
every functional office and every product managers slot with one full-time person. This
resulted in a relatively small division having top level managers as well as full-time
functional group and full-time product managers. Within months, however, this top heavy
division was pared down to more reasonable staffing levels; by assigning individuals to two
or more slots, management got costs under control.
PreventionThe divisions problem was caused by top managements assumption that each
managerial slot requires a full-time incumbent. Overstaffing is much less liable to occur
when an organization evolves gradually from a conventional design into a matrix, and
managers perform as both functional and product managers. While this technique can be
justified as a transition strategy, it also has its hazards. A safer route is to assign managers
roles on the same side of the matrix (i.e., two functional jobs or two product management
jobs).
As a final argument against the fear of overhead costs, consider that no well-run
organization would adopt a matrix structure without the longer run expectation that, at a
given level of output, the costs of operations would be lower than with other
organizational forms. In what way can such economies be achieved?
The potential economies come from two general sources: fewer bad decisions and less
featherbedding. First and most important, the matrix can improve quality of business
decisions because it helps bring the needed information and emphasis to bear on critical
decisions in a timely fashion. The second source, less featherbedding, is not so obvious,
but potentially of greater significance. How can it work?
As one senior manager in a consulting firm put it, There is no place to hide in a matrix
organization. This fact makes clear-cut demands on middle level people and consequently
puts pressure on them to produce. For the long-term good of both the people involved and
the organization, top managers need to keep such pressures from becoming too strong.
Because it is perfectly possible to get too much as well as too little pressure, a creative
tension is sought.
DiagnosisSinking may occur for two reasons. Either senior management has not
understood or been able to implement the matrix concept as well as lower level managers,
or the matrix has found its appropriate place. For example, if a company sets up a matrix
between its basic functional and product groups, the product managers never truly
relinquish their complete control, and the matrix fails to take hold at the corporate level.
But, say, one or two of the managers find the idea to be useful within their divisions. Their
own functional specialists and project leaders can share the power they delegate and the
design can survive within subunits of the corporation. For example, Dow Chemicals
attempt to maintain the product/geography balance at the top failed, but the
function/product balance held within the geographic areas for several years.
PreventionIf the corporate top management thinks through which dimensions of the
company it must balance, and at what level of aggregation, it can keep the matrix from
sinking. For example, top managers should ask themselves if all the business units need to
be balanced by central functional departments. If the answer is no, then some business
units should operate as product divisions with the traditional pyramid of command, while
others share functional services in a partial matrix. However, sinking is not always bad and
should be prevented only when it indicates that an appropriate design is disintegrating.
The question of size is a great concern for many managers who ask, in effect, That sounds
great for a $250-million company with a few thousand employees, but can it work for a $2billion or $3-billion company with 50,000 employees? Its entire company is the size of one
of our divisions. Our experience indicates that matrix management and organization
seems to function better when no more than 500 managers are involved in matrix
relationships. But that does not rule out the $2-billion to $3-billion company. In a company
of 5,000 only about 50 managers are likely to be in the matrix; so in a company with 50,000
employees only about 500 may need to be involved in dual reporting lines. With that
number, the people who need to coordinate regularly are able to do so through
communication networks that are based on personal relations.
Whatever the size unit in which the matrix operates, the important thing is for
management to have reasoned carefully from an analysis of the task to the design of the
organization. Then, if settling occurs, it should be seen not as a pathology but as a selfadjustment that suggests the organizations capacity to evolve with growth.
Uncontrolled layering
Matrices which lie within matrices which lie within matrices result frequently from the
dynamics of power rather than from the logic of design.
DiagnosisSometimes matrices not only sink but also cascade down the organization and
filter through several levels and across several divisions. This layering process may or may
not be pathological. In fact, it may be a rational and logical development of the matrix, but
we include it briefly here because it sometimes creates more problems than it solves. In
terms of the metaphor we have used in this article, layering is a pathology only if the
matrix begins to metastasize. When this occurs, organization charts begin to resemble
blueprints for a complex electronic machine, relationships become unnecessarily complex,
and the matrix form may become more of a burden than it is worth.
Prevention and treatmentThe best remedies for uncontrolled layering are careful task
analysis and reduced power struggles. We have seen a few cases where one dimension of a
matrix was clearly losing power to the other, so, adapting an if you cant beat em, join
em philosophy, it created a matrix within its own dimension. A product unit, for
example, developed its own functional expertise distinct from the functional units at the
next level up. The best defense was a good offense, or so it seemed.
In two other cases, the international divisions of two large companies each created its own
matrix by adding business managers as an overlay to its geographic format, without
reconciling these with the managers who ran the domestic product/service groups. In each
case, adequate conceptualization by top managers would probably have simplified the
organization design and forestalled the layering, which occurred because of power
maneuvers. Management can treat this unhealthy state best by rebalancing the matrix so
that no manager of one dimension is either too threatened or pushed too hard toward a
power goal.
Matrix design is complex enough without the addition of power struggles. A wellconceptualized matrix is bound to be less complex and easier to manage than one that is
illogically organized.
Navel gazing
Managers in a matrix can succumb to excessive internal preoccupation and lose touch with the
marketplace.
The navel gazers are not at all lethargic; rather they are involved in a heated fraternal
love/hate affair with each other. This inward preoccupation is more common in the early
phases of a matrix, when the new behaviors are being learned, than in matrices that have
been operating for a few years.
TreatmentIf the managers in the matrix are navel gazing, the first step in the treatment is
to make these managers aware of the effects. Are customers complaining a lot, or at least
more than usual? Managers need to confront internal conflict, but also to recognize that
confrontation is secondary to maintaining effective external relationships. Navel gazing
generally occurs when the matrix has been fully initiated but not yet debugged. People
accept it, but they are engrossed in figuring out how to make it work.
The second step is to treat the inward focus as a symptom of the underlying issue: how to
institutionalize matrix relationships so that they become familiar and comfortable
routines, and so that people can work through them without becoming obsessed by them.
Finally, it must always be remembered that any form of organization is only a means and
should never become an end in itself.
Decision strangulation
Decision strangulation
Constant clearingIn one company we know of, various functional specialists who reported
to a second boss, a product manager, picked up the idea that they had to clear all issues
with their own functional bosses before agreeing to product decisions. This meant that
every issue had to be discussed in at least two meetings, if not more. During the first
meeting, the specialists and the product manager could only review the facts of the issue,
which was then tabled until, at the second meeting, the specialists cleared the matter with
their functional bosseswho by this process were each given a de facto veto over product
decisions.
This impossible clearing procedure represented, in our view, a failure of delegation, not of
the matrix. One needs to ask why the functional specialists could not be trusted to act on
the spot in regard to most product decisions in ways that would be consistent with the
general guidelines of their functional departments? Either the specialists were poorly
selected, too inexperienced and badly informed, or their superiors were lacking in a
workable degree of trust of one another. Regardless, this problem, and its prevention and
treatment, needs to be addressed directly without making a scapegoat of the matrix.
to the CEO for resolution without creating the ultimate in information overload. So, they
think, will not the inevitable disagreement lead to a tremendous pileup of unresolved
conflict?
Certainly, this can happen in a malfunctioning matrix. Whether it does happen depends
primarily on the depth of understanding that exists about required matrix behavior on the
part of managers in the dual structure. Let us envision the following scene: a manager with
two bosses gets sharply conflicting instructions from his product and his functional bosses.
When he tries to reconcile his instructions without success, he quite properly asks for a
session with his two bosses to resolve the matter. The three people meet, but the
discussion bogs down, no resolution is reached, and neither boss gives way.
The two bosses then appeal the problem up a level to their respective superiors in each of
the two chains of command. This is the critical step. If the two superiors properly
understand matrix behavior, they will first ascertain whether the dispute reflects an
unresolved broader policy issue. If it does not, they know their proper step is to teach their
subordinates to resolve the problem themselvesnot to solve it for them. In short, they
would not let the unresolved problem escalate, but would force it back to the proper level
for solution, and insist that the solution be found promptly.
Often, conflict cannot be resolved; it can, however, be managed, which it must be if the
matrix is to work. Any other course of action would represent managements failure to
comprehend the essential nature of the design.
Unilateral styleA third possible reason for decision strangulation in a matrix system can
arise from a very different sourcepersonal style. Some managers have the feeling they are
not truly managing if they are not in a position to make crisp, unilateral decisions.
Identifying leadership with decisive action, they become very frustrated when they have to
engage in carefully reasoned debates about the wisdom of what they want to do.
Such a manager is likely to feel frustrated even in regard to a business problem whose
resolution will vitally affect functions other than his own, such as in a company that is
experiencing critical dual pressure from the marketplace and from advancing technology.
A matrix that deliberately induces simultaneous decision making between two or more
perspectives is likely to frustrate such a person even further.
If managers start feeling emasculated by bilateral decision making, they are certain to be
unhappy in a matrix organization. In such cases the strangulation is in the eye of the
beholder. Such people must work on their personal decision-making style or look for
employment in a nonmatrix organization.
At Last, Legitimacy
We do not recommend that every company adopt the matrix form. But where it is relevant,
it can become an important part of an effective managerial process. Like any new method it
may develop serious bugs, but the experiences that many companies are acquiring with
this organization form can now help others realize its benefits and avoid its pitfalls.
The matrix seems to have spread despite itself and its pathologies: what was necessary was
made desirable. It is difficult and complex, and human flexibility is required to arrive at
organizational flexibility.
But the reverse is also true; success has given the form legitimacy, and, as the concept
spreads, familiarity seems to reduce the resistance and difficulties people experience in
using the matrix. Managers are now beginning to say, It isnt that new or different after
all. This familiarity is a sign of acceptance, more than of change or moderation of the
design.
For generations managers lived with the happy fiction of dotted lines, indicating that a
second reporting line was necessary if not formal. The result had always been a sort of
executive mnage trois, a triangular arrangement where the manager had one legitimate
relationship (the reporting line) and one that existed but was not granted equal privileges
(the dotted line).
As executives develop greater confidence with the matrix form, they bring the dotted line
relationship out of the closet, and grant it legitimacy.
Each time another organization turns to the matrix, it has a larger and more varied number
of predecessors that have charted the way. The examples of wider applicability suggest
that the matrix is becoming less and less an experiment and more and more a mature
formulation in organization design. As more organizations travel the learning curve, the
curve itself becomes an easier one to climb. Similarly, as more managers gain experience
operating in matrix organizations, they are bound to spread this experience as some of
them move, as they inevitably will, into other organizations.
We believe that in the future matrix organizations will become almost commonplace and
that managers will speak less of the difficulties and pathologies of the matrix than of its
advantages and benefits.
1. From the forthcoming article by M.R. Weisbord, M.P. Charns, and P.R. Lawrence,
Organizational Dilemmas of Academic Medical Centers, Journal of Applied Behavioral
Science, Vol. XIV, No. 3.
2. C.J. Middleton, How to Set Up a Project Organization, HBR MarchApril 1967, p. 73.
A version of this article appeared in the May 1978 issue of Harvard Business Review.
Stanley M. Davis is professor of organization behavior, business policy, and international business at Boston
Universitys School of Management. Before that he was on the faculty of Columbia University and the Harvard
Business School. He is a director of the Management Analysis Center in Cambridge, Massachusetts and the
author of numerous journal articles and books.
Comments
Leave a Comment
POST
0 COMMENTS
POSTING GUIDELINES
We hope the conversations that take place on HBR.org will be energetic, constructive, and thought-provoking. To comment, readers must
sign in or register. And to ensure the quality of the discussion, our moderating team will review all comments and may edit them for clarity,
length, and relevance. Comments that are overly promotional, mean-spirited, or off-topic may be deleted per the moderators' judgment.
All postings become the property of Harvard Business Publishing.