Matrix organizations can help companies outperform their competitors. But they are often hard to run, and attempts to fix them frequently fail. At the root of these problems - which relate less to the structure of the matrix itself and more to the way the parts work together - is how we think about power.
Here is what happens: one dimension of the matrix (projects, functions, geographic units, or customer segments, for example) has too much or too little power, so we reallocate it. Or we change the coordination mechanisms (such as dual reporting lines, hierarchical links with informal overlays, or solid versus dotted lines). People caught in the middle of these matrices end up getting lost, and performance drops.
For example, many organizations initially fail to give enough power to project managers. Eventually they recognize their error, so they "empower" these managers by giving them authority, in whole or in part, to evaluate team members and determine promotions, raises, and other rewards. But their new power comes at the expense of line managers, who now control only a part, or none, of the team's rewards. Because the power of line managers is only a fraction if what it was, they now achieve only a fraction of what they used to accomplish in mobilising the team toward line objectives (innovation and deploying new standards, for instance). The company wins on one front (the short-term horizon of ongoing projects), only to lose on another (the longer term).
In this scenario, the shifting of power is a zero-sum game. But that dynamic of intrinsically contradicts the intent of matrix design: channeling the behaviour of employees along multiple dimensions that reflect different performance requirements. To achieve that end, each dimension needs power - in the form of motivating elements, constraints, or other types of influence - to make people do what they would not do spontaneously. When we add new dimensions to the structure, we need more overall power. The inability to move from a zero-sum game to a positive-sum game by increasing the pool of power is at the heart of malfunctioning matrix organizations.
The Origin of Power
Where does power come from? Power is based on the ability of one party to influence those things (the stakes) that matter to other parties to a degree that causes those other parties to eventually do something they would not have done without the first party's actual or implied intervention. Despite popular belief, power is not particularly related to an imbalance of information available to the parties. Instead, the asymmetry relates to the terms of exchange in a relationship: the reciprocal possibilities of action. The imbalance - thus the power - issues from A's potential to make a greater difference regarding stakes that matter to B than the reverse. On another issue or as stakes change, B may have power over A. Power is not an attribute of people. Rather, it stems from a relationship tied to a situation.
To increase the total amount of power available in an organization, management must introduce at least one new stake that matters to people and to the company. Those who can influence that stake will get the power they need, without taking power away from others. In one case, the added stake was competence development, which became a new condition for promotion. Until then, promotions were based mainly on employees' contributions to projects. Line managers became responsible for assessing engineers on a variety of competence criteria, thus gaining more power, while project managers retained their power to assess and reward teams on project performance.
Empowering one dimension without diminishing others empowers the organization as a whole. Think of it as increasing the number of cards in a game: you then increase the possibility of exchanges. Each dimension can now have the critical mass of cards needed to risk playing in an open way: to take initiatives with others, to accept transparency about performance, or to question one's own choices to benefit the common good. In other words, each dimension can actively participate in the cooperative mix that is the lifeblood of matrix organizations.
Misconceptions About the Matrix
What often gets in the way of making the matrix work is a series of misconceptions.
Misconception 1: the matrix must be balanced so that each dimension has the same weight. Strict equality is not the key. What is important is to avoid a concentration of power that causes dominated dimensions to withdraw from the cooperative effort as they realize that they always lose out in the trade offs inherent to the interdependencies of a matrix. The issue is not one of balance but of enriching the game of adding stakes (the cards in the cooperation game) to increase the possibilities of exchange.
Misconception 2: the matrix approach doesn't work, because people can be held accountable only for results that depend on factors under their sole authority (such as teams and equipment). The problem with this principle, as well as with its complement about the importance of a single boss, is that is is less and less applicable, given that the interdependencies that most organizations must now master in order to be competitive. If it were applied to distribution, the only result for which supermarket managers could be accountable would be the cleanliness of their shops. Everything else, from the product mix to the IT-supported checkout process, depends substantially on others. The "single boss" and "accountability equates authority" principles make sense only when power is a zero-sum game. They do not apply when increasing the total amount of power allows each dimension to obtain from the others the cooperation it needs.
Misconception 3: The matrix needs "guardians" to serve as arbitrators among its dimensions. This approach results in over escalation- worse yet, an over escalation that delivers a weak punch. Real power, the kind needed to make other parties accept trade offs, is not an attribute one automatically inherits from a title or from a position in the hierarchy. Power is intransitive: just because A, the guardian, controls something important for B, and B for C, it does not necessarily follow that A controls something important for C. Which is why guardians are never found "high" enough, unless they are CEOs, who then become mired in making operational trade offs. The better practice is to avoid needing guardians as a compensatory mechanism by instead creating the conditions for cooperation among dimensions.
Misconception 4: What matters most is the clarity of rules and structure" The clarification of decision rights is indeed crucial for matrix organizations. Problems arise however when a company relies mainly on the formalization of rules and boxes to make the matrix work. That effort always results in anemic leadership. The reason is simple: power invariably requires some room to maneuver, some area for discretion. When there is a rule for everything, one depends on the rule, not on one's management. Then the only people who can make a difference - who have real power - are those who can change the rules. These are players at the top of an organization or collective bodies (unions, for instance) that are far from operations. An over reliance on formalization always results in a dependency that takes the form of an inverted hierarchy. Managers depend on teams to achieve their objectives, but teams do not depend on their managers. The consequences are counterproductive whenever field management is supposed to play a real role.
As companies become increasingly multidimensional or even network like, issues about power become even more pervasive, subtle, and critical than in the basic hardwired form of traditional vertical-and-horizontal matrices. Understanding and managing power are now at the heart of sound organization design.