Often, neither direct supervision nor all three forms of standardization are sufficient to achieve the coordination an organization requires. In other words, important interdependencies remain after all the individual posi-tions have been designed, the superstructure built, and the planning and control systems set in place. The organization must then turn to mutual adjustment for coordination. A customer complaint about poor service may, for example, require the sales and manufacturing managers to sit down together to work out new delivery arrangements.
Until recently, this kind of mutual adjustment was left largely to chance; at best, it took place informally, outside the formal organizational structure. But in recent years, organizations have developed a whole set of devices to encourage liaison contacts between individuals, devices that can be incorporated into the formal structure. In fact, these liaison devices represent the most significant contemporary development in organization design—indeed, the only serious one since the establishment of planning and control systems a decade or two earlier.
Since the 1960s, the popular literature of management has heralded each new liaison device as a major discovery. First it was “task forces,” then “matrix structure,” later the “integrators.” But the reader was left in confusion: Were these just different names for the same phenomenon, or was each, in fact, a distinctly new contribution? And if so, did each bear any relation to the others? The writings of Jay Galbraith (1973) have re- solved many of these problems. Galbraith proposed a continuum of these liaison devices, from the simplest to the most elaborate: direct contact between managers, liaison roles, task forces, teams, integrating roles, managerial linking roles, and matrix organization. For purposes of our discussion, Galbraith’s scheme has been reduced to four basic types of liaison devices—liaison positions, task forces and standing committees, integrating managers, and matrix structure.
1. Liaison positions
When a considerable amount of contact is necessary to coordinate the work of two units, a “liaison” position may be established formally to route the communication directly, bypassing the vertical channels. The position carries no formal authority, but because the incumbent serves at the crossroads of communication channels, he emerges as a nerve center of the organization with considerable power. Note that this power is informal, deriving from knowledge, not status. Some liaison positions serve between different line units—for example, the engineering liaison man who is a member of the engineering department but is physically located in the plant, the sales liaison person who mediates between the field sales force and the factory, or the purchase engineer who sits between purchasing and en- gineering. The latter are “instantly available to provide information to engineers whenever they need help in choosing components. They assist in writing specifications (thus making them more realistic and readable) and help expedite delivery of laboratory supplies and material for prototype models” (Strauss, 1962-63:180-81). Other liaison positions join line and staff groups; for example, the personnel specialists and accountants who counsel line departments while remaining responsive to their technocratic homes.
2. Task forces and standing committees
The meeting is the prime vehicle used in the organization to facilitate mutual adjustment. Some meetings are impromptu; people bump into each other in the hall and decide to have a “meeting.” Others are sched- uled on an ad hoc basis, as required. When the organization reaches the point of institutionalizing the meeting—that is, formally designating its participants, perhaps also scheduling it on a regular basis—the meeting may be considered to have become part of the formal structure. This hap- pens when extensive and fairly regular contact—at least for a period of time—is required between the members of various units to discuss com- mon concerns. Two prime liaison devices are used to institutionalize the meeting. The task force is a committee formed to accomplish a particular task and then disband. In contrast, the standing committee is a more permanent interdepartmental grouping, one that meets regularly to dis- cuss issues of common interest. Many standing committees exist at middle levels of the organization, and others are formed at the strategic apex, a common one being the executive committee.
3. Integrating managers
When more coordination by mutual adjustment is required than liaison positions, task forces, and standing committees can provide, the organi- zation may designate an integrating manager—in effect, a liaison position with formal authority. A new individual, sometimes with his own unit, is superimposed on the old departmental structure and given some of the power that formerly resided in the separate departments. Integrating man- agers can include brand managers in consumer-goods firms, responsible for the production and marketing of particular products; project managers in aerospace agencies, responsible for integrating certain functional ac- tivities; unit managers in hopsitals, responsible for integrating the activities of doctors, nurses, and support staff in particular wards, and so on.
The formal power of the integrating manager always includes some aspects of the decision processes that cut across the affected departments, but it never (by definition) extends to formal authority over the depart- mental personnel. (That would make the person department manager in- stead of integrating manager.) To control their behavior, therefore, the integrating manager must use his decisional authority and, more impor- tant, his powers of persuasion and negotiation. Galbraith lists three stages in the extension of the decisional power of the integrating manager. First, he can be given power to approve completed decisions—for example, to review the budgets of the departments. Second, he can enter the decision process at an earlier stage—for example, to draw up in the first place the budget that the departments must then approve. Third, he can be given control of the decision process, as when he determines the budget and pays the departments for the use of their resources.
Consider the brand manager in a consumer-goods firm. He is a kind of mini-general manager, responsible for the success of a single product. His performance is measured by how well it does in the marketplace. He must understand purchasing, manufacturing, packaging, pricing, distribu- tion, sales, promotion, advertising, and marketing, and must develop plans for the brand, including sales forecasts, budgets, and production schedules. But the brand manager has no direct authority over the market- ing or manufacturing departments. Rather, along with all the other brand managers of his firm, he negotiates with manufacturing to produce his brand and with marketing to sell it. If, however, he controls the budget for his brand, and has discretion in the use of it—for example, to contract its manufacture to different plants—he may have considerable power.
Whereas the brand manager is concerned with an existing or ongoing product, the project or program manager is concerned with bringing a new or embryonic undertaking to fruition—say, a new product or new facility. In both these cases, integrating managers with market orientations have been superimposed on functional structures to achieve work-flow coordi- nation. But integrating managers with functional orientations can also be superimposed on market-based structures to encourage specialization, as when a manager concerned with the quality of programming is overlaid on a data-processing department formally organized on a project basis.
The job of integrating manager is not an easy one, the prime difficulty being to influence the behavior of people over whom he has no formal authority. The brand manager, for example, must persuade the manufac- turing department to give priority to the production of his product and must encourage the sales department to promote his brand over the oth- ers, and the programming manager must encourage the programmers who report formally to project managers to increase the quality of their work. As Galbraith notes, what the integrating manager has at his command are contacts, information gained from serving at the crossroads of different channels, and the capacity to build up confidence and to encourage more effective decision making because of his broader perspective. The effective integrating manager appears to require a high need for affiliation and an ability to stand between conflicting groups and gain the acceptance of both without being absorbed into either.
Matrix structures
No single basis for grouping can contain all the interdependencies. Func- tional ones pose work-flow problems, market-based ones impede contacts among specialists, and so on. Standardization achieved through formaliza- tion of behavior, training and indoctrination, or planning and control sys- tems can sometimes alleviate the problem, but important interdependen- cies often remain.
In our discussion to this point, we have seen at least three ways in which organizations handle this problem. These are shown in Figure 4-3. The first is to contain the residual interdependencies at the next higher level in the hierarchy; the second is to deal with the residual interdepen- dencies in staff units (a dual structure is built—one line with the formal authority to decide, that contains the main interdependencies, the other staff, which advises on the residual interdependencies, as when market researchers or financial analysts advise the different product managers to help them coordinate their activities functionally); the third is, of course, to use one of the liaison devices already discussed, the organization in effect preserving its traditional authority structure but superimposing, say, task forces to deal with the residual interdependencies.
Figure 4-3. Structures to deal with residual inter- dependencies
Each one of these solutions favors one basis of grouping over an- other. Sometimes, however, the organization needs two (or even three) bases of grouping in equal balance. For example, an international firm may not wish to favor either a geographical or a product orientation in its structure, or a data-processing department or advertising agency may not wish to make a choice between a project orientation and an emphasis on specialization. Galbraith cites the case of the high-technology company whose products were undergoing continual change. Some managers ar- gued for product divisions to deal with the complex problems of schedul- ing, replacing, and managing the new products, but others objected. The engineering manager felt that this would reduce the influence of his people just when he was experiencing morale and turnover problems. Manage- ment needed a product orientation as well as an improvement in the mor- ale of the key specialists, both at the same hierarchical level. In these cases, organizations turn to the ultimate liaison device—matrix structure.
By using matrix structure, the organization avoids choosing one basis of grouping over another; instead, it chooses both. “In the simplest terms, matrix structure represents the effort, organizationally speaking, to ‘have your cake and eat it, too'” (Sayles, 1976:5). But in so doing, the organization sets up a dual authority structure. As a result, matrix struc- ture sacrifices the principle of unity of command. As shown in Figure 4-3(d), formal authority comes down the hierarchy and then splits, creat- ing joint responsibilities and doing away with the notion of an unbroken chain of authority. To the classical writers, dual authority was anathema; it violated the principles and destroyed the neatness of the structure.2 But as Galbraith notes, dual authority is hardly foreign to us: “Almost all of us were raised in the dual authority system of the family . . .” (1973:144). Similarly, in the matrix structure, different line managers are equally and jointly responsible for the same decisions and are therefore forced to recon- cile between themselves the differences that arise. A delicate balance of power is created. To return to our example of the advertising agency, if the specialists need to be oriented to projects yet insist on being evaluated by their own kind, then matrix structure would have the evaluation decision made jointly by project and functional managers.
This balance of formal power is what distinguishes matrix structure from the other means of handling residual interdependencies, including the other liaison devices. It is one thing to have four product managers, each with a manufacturing, marketing, engineering, and personnel man- ager reporting to him, or four integrating managers, each seeking to coor-dinate the work of four functional managers with the line authority, or even to combine the latter into market-based task forces; it is quite another thing to force the product and functional managers to face each other, as in Figure 4-3(d), with equal formal power.
Nevertheless, Sayles (1976) notes in his review of matrix structure that in many contemporary organizations, the alternatives to it are simply too confusing:
There are just too many connections and interdependencies among all line and staff executives—involving diagonal, dotted, and other “informal” lines of control, communication, and cooperation—to accommodate the comfort- able simplicity of the traditional hierarchy, be it flat or tall. . . .Many companies, in fact, tie themselves in semantic knots trying to figure out which of their key groups are “line” and which “staff.” (pp. 3, 15)
Sayles goes on to suggest that matrix structure is for organizations that are prepared to resolve their conflicts through informal negotiation among equals rather than recourse to formal authority, to the formal power of superiors over subordinates and line over staff. In effect, he seems to be telling us—picking up on Galbraith’s point about the family—that matrix structure is for grown-up organizations. In fact, he believes that a great many organizations have already adopted some form of matrix structure, even if not in name.
Two kinds of matrix structures can be distinguished: a permanent form, where the interdependencies remain more-or-less stable and so, as a result, do the units and the people in them; and a shifting form, geared to project work, where the interdependencies, the market units, and the people in them shift around frequently. An example of permanent matrix structure can be found in the administration of some cities, where the functional citywide departments of parks, police, health, and so on, coor- dinate with the administrators of specific wards, and the two are jointly responsible for ensuring the quality of services to the city population. Some international companies have also moved toward this type of struc- ture, typically putting the managers of geographical regions face to face with the managers of worldwide product lines. Reporting to both is a regional product manager, to whom in turn the functional managers re- port, as shown in Figure 4-4. A characteristic of the permanent matrix structure, evident in Figure 4-4, is that the chain of authority, once split, may reunite again, so that while one manager reports to two above him, his own subordinates report only to him.
The shifting matrix structure is used for project work, where the out- puts change frequently, as in aerospace firms, research laboratories, and consulting think tanks. In these cases, the organization operates as a set of project teams (in effect, temporary market-based units) that draw their members from the functional departments, which serve various “house- keeping” purposes. The National Aeronautics and Space Administration (NASA) has been a well-known user of such a structure. A fundamental characteristic of the teams used in the shifting matrix structure is that their leaders are full-fledged managers (of the market units), with formal author- ity (jointly shared with the managers of the functional units) over their members. That is what distinguishes them from the leaders of the task forces and the integrating managers described earlier. Those liaison de- vices were superimposed on a traditional line structure. This structure is matrix precisely because the task-force leaders take their place alongside the functional managers, sharing power equally with them.
Figure 4-4. A permanent matrix structure in an interna- tional firm
Matrix structure has its share of problems. Although it seems to be a most effective device for developing new activities and for coordinating complex multiple interdependencies, it is no place for those in need of security and stability. Dispensing with the principle of unity of command creates considerable confusion, stress, and conflict, and requires from its participants highly developed interpersonal skills and considerable toler- ance for ambiguity. There is also the problem of maintaining the delicate balance of power between the different sorts of managers. A tilt in one direction or the other amounts to a reversion to a traditional single-chain hierarchy, with the resulting loss of the benefits of matrix structure. How- ever, a perfect balance without cooperation between the different manag- ers can lead to so many disputes going up the hierarchy for arbitration that top management becomes overloaded. Then there is the problem of the cost of administration and communication in these structures. “The sys-tern demands that people have to spend far more time at meetings, dis- cussing rather than doing work, than in a simpler authority structure. There simply is more communicating to be done, more information has to get to more people . . .” (Knight, 1976:126). Moreover, as we shall soon see, matrix structure requires many more managers than traditional struc- tures do, thereby pushing up the administrative costs considerably.
A continuum of the liaison devices
Figure 4-5 summarizes our discussion of these four liaison devices—liai- son positions, task forces and standing committees, integrating managers, and matrix structure. Again, the idea is borrowed from Galbraith and then modified. The figure forms a continuum, with pure functional structure at one end (that is, functional structure as the single chain of line authority) and pure market structure at the other. (Again, any other basis for group- ing could be put at either end.) The first and most minor modification to either of the pure structures is the superimposition of liaison positions on it. Such positions generate a mild market orientation in the functional structure or a mild functional orientation in the market structure, thereby reducing slightly the informal power of the line managers (as shown by the diagonal line that cuts across the figure). A stronger modification is the superimposition of task forces or standing committees on either of the pure structures; the strongest modification, short of dispensing with the princi- ple of unity of command, is the introduction of a set of integrating manag- ers. As we have seen, such managers are given some formal decisional power—for example, control of important resources—and acquire consid- erable informal power. But the other managers, whether functional or market, retain their traditional line authority, including that over the per- sonnel. Finally, standing midway between the two pure structures of Fig- ure 4-5 is matrix structure, which represents an equal balance of power between the two. Dual authority replaces unity of command.
4. The liaison devices and the other design parameters
At a number of points, our discussion has hinted at the relationships between the liaison devices and the design parameters we have already discussed. Now let us focus on these relationships, looking first at the superstructure and then at the individual positions.
It is clear that the liaison devices can be used with any basis for grouping, since they are designed to override the limitations of using only a single one. Nevertheless, a review of the examples in the literature sug- gests that these devices are most often superimposed on functional group- ings to introduce an orientation to markets.
As for unit size, as we saw earlier, liaison devices are the tools tcencourage mutual adjustment by informal communication, and as we noted in Chapter 3, such communication requires face-to-face work groups of small size. Hence, we would expect that the greater the use of the liaison devices, the smaller the average size of organizational units. This should be especially pronounced for task forces and standing committees, as well as for temporary matrix structures, where the essential work is carried out in groups. Were we to consider the number of managers in- stead of unit size, the effect of the liaison devices should be even more pronounced. The addition of integrating managers ups the proportion of managers to nonmanagers significantly; the switch to matrix structure means the doubling of managers, more or less,3 since many employees now have two bosses. So certain of the liaison devices, especially matrix structure, result in a proliferation of the managers in the organization.
Turning to the design of the individual positions, we would expect the liaison devices to be used where the organization cannot standardize its behaviors but must instead rely on mutual adjustment to coordinate its activities. In other words, there is less need for informal communication in bureaucratic structure, which means that the liaison devices are tools pri- marily of organic structures. They are flexible mechanisms to encourage loose, informal relationships. No doubt the milder liaison devices—liaison positions, task forces, and standing committees, those near the ends of the Figure 4-5 continuum—are sometimes superimposed on bureaucratic structures to reduce their inflexibility in places. But the use of the stronger liaison devices—integrating managers and matrix structure—so upset the traditional patterns of formalized behavior that the resulting structure can no longer be thought of as bureaucratic.
The liaison devices are generally used where work is, at the same time, (1) horizontally specialized, (2) complex, and (3) highly interdepen- dent. If the work were not both horizontally specialized and interdepen- dent, close coordination would not be necessary and the liaison devices would not be used. And if the work were not complex, the necessary coordination could be achieved largely by direct supervision or the stan- dardization of work processes or outputs. Complex work can, of course, be coordinated by standardizing the skills used to do it—but only as long as the interdependencies are not great. Past some point of interdependence among specialized complex tasks, mutual adjustment is mandatory for coordination, and so the liaison devices are called upon to coordinate them.
Of course, specialized complex tasks are professional ones, and so we should find a relation between professionalism (as well as training) and the use of the liaison devices. Indeed, many of our examples in this chapter have come from organizations that rely on professional expertise—aero- space agencies, research laboratories, and the like. Earlier it was suggested that there could be two kinds of professional organizations, one where the professionals function independently as individuals, and the other where they work together in groups. Now we see that the liaison devices are key design parameters in this second type of professional organization.
As for the relation between liaison devices and planning and control systems, to some extent at least, the use of these two lateral linkages is apt to be mutually exclusive. Unable to contain task interdependencies by the design of both individual positions and the superstructure, the organiza- tion would rely either on the standardization of outputs or the use of the devices of mutual adjustment. Consider, for example, how Sayles de- scribes the organization that uses matrix structure. Its introduction of mul- tiple sources of authority presupposed that its decisions “cannot be made by a well-programmed computer or small, expert planning groups” (1976:15); its “goals are, at once, multiple and conflicting and changing” (p. 16); the nature of its work interdependencies are such that “no account- ing model” (p. 15) can balance the range of forces present in it. Rather, “the matrix forces decision making to be a constant process of interchange and trade-off, not only between the overall system and its specialized components and interest groups, but also between and among the special- ists in the interest groups themselves” (p. 17). Clearly, planning and con- trol systems cannot flourish in such an organization. In particular, perfor- mance control systems inappropriately require stable goals and units with only pooled interdependencies. And although some action planning may be feasible to deal with unit interdependencies, it must be general enough to allow for considerable adaptation through mutual adjustment. NASA used action planning to lay out the general schedule of the Apollo project, but so much additional coordination and adaptation were required that the space agency emerged from the project as a leader in the use of the liaison devices.
5. Liaison devices by part of the organization
The liaison devices appear to be best suited to the work carried out at the middle levels of the structure, involving many of the line managers as well as staff specialists. A standing committee may meet weekly to bring together the plant superintendent, sales manager, and head of purchasing; an engineer may be designated to a liaison position between a staff group in research and the line marketing department; a task force may be created, drawing middle-level members from the accounting, manufacturing, en-gineering, and purchasing departments, to investigate the feasibility of purchasing new equipment. And matrix structure, especially of the perma- nent kind, is commonly used where the power of middle-line managers representing two different bases for grouping must be balanced.
In general, given the nature of the work of middle managers—largely ad hoc but somewhat amenable to structure—we would often expect the set of liaison devices to be a most important design parameter of the mid- dle line. At the very least, meetings abound in this part of the organization, many of them bringing together task forces and standing committees. Sim- ilarly, within staff units doing specialized, complex, and highly interde- pendent work—both in much of the technostructure and the upper levels of the support staff—we would expect the set of liaison devices to be a prime design parameter. Task forces and shifting matrix structure are es- pecially well suited to the project work that often takes place in the tech- nostructure. For example, a management science department may base its specialists in homogeneous groups (cost analysts, statisticians, econo- mists, and so on) but deploy them in project teams to do their studies. And as we shall see later, organizations with many staff groups in close contact with middle-line units make such heavy use of the liaison devices that the staff/line distinction can break down and their three middle parts emerge as one amorphous mass of mutual-adjustment relationships.
As noted in earlier chapters, work in the operating core is coordinated primarily by standardization, with direct supervision as the backup coordi- nating mechanism. But in cases where the operating core is manned by professionals whose work interdependencies require them to function in teams—as in research centers and creative film companies—mutual adjust- ment is the key coordinating mechanism, and task forces and shifting matrix structures key design parameters.
Some use is also made of the liaison devices at the strategic apex. As we have seen, standing committees are common among senior managers; task forces are also used sometimes to bring them together with middle- line managers as well as senior staff personnel; likewise, liaison positions are sometimes designated to link the strategic apex to other parts of the organization, as when a presidential assistant is designated to maintain contact with a newly acquired subsidiary. But wider use of the liaison devices at the top of the organization is probably restricted by the very fluid and unprogrammed nature of the work there. Even the flexible liaison devices are simply too structured. As I have found in my own research, top managers often seem to prefer the informal telephone call or the im- promptu meeting to the task force with its designated membership or the standing committee that meets on a regular basis.
Source: Mintzberg Henry (1992), Structure in Fives: Designing Effective Organizations, Pearson; 1st edition.