Chapter 6 Flattening the Pyramid Cost to manage someone What managers do How many layers are needed? Too few layers Spans of control Stretching spans Rule of 6 Top management clutter Managing staff Confusing status and management DeMatrixing Nurturing innovation
Flattening the Pyramid
Excerpt from Downsizing: Reshaping the Corporation for the Future
By Robert M. Tomasko
"Will we revert to memos that need to be signed by ten layers of authority, which makes a mockery of entrepreneurship?" worried John Welch, General Electric's chairman, as GE pulled out of the early 1980s recession. He shared a concern of many business leaders as the acute need to downsize let up when the economic cycle turned upward. Peter Drucker also raised alarms about the number of middle managers in many companies growing three to four times faster than their sales.
The concern was a very real one. While companies had let go thousands of managers, few had followed these difficult layoffs with any significant redesign of their organizations. They changed the number of managers on board without considering the changes they needed to make in how management work was actually done. The resulting situation is like a Christmas tree with half its ornaments removed. The decorations are gone but the barren branches remain. Without a pruning of some branches one may have an urge to correct the tree's unbalanced appearance by rehanging the missing lights. Similarly, many companies who reduced their management complement without downsizing their management structure can find easy excuses to upsize when their business situation turns more favorable.
Attacking excess management
Such changes in headquarters-field office organization are certainly not limited to the public sector. Sears Roebuck and Co. made a similar move when it eliminated the territorial offices that served as a link between their Chicago headquarters and the two dozen administrative offices that are directly responsible for overseeing Sears' 800 stores. The move was made for reasons similar to the Postal Service change: to speed decision making in response to a fast changing marketplace. The four territorial offices eliminated employed 1,800 people; their functions were either delegated to the administrative offices, done by a new ten-person field management office set up in Chicago, or eliminated. The change was expected to help shore up declining profit margins as well as to insure that the right merchandise got to the right stores at the right times.
Brunswick's Mercury Marine business cut the layer of management just below its division level that was intended to provide sales and marketing support to each division. Instead these functions were consolidated at the Mercury Marine headquarters in a smaller unit that served all divisions. In this instance the primary motivation was cost reduction, an estimated $6 million was saved each year, as Brunswick decided these functions were over-decentralized.
Prudential Insurance Company of America, the nation's largest life insurer, has made several organizational shifts to better prepare it for the financial services industries' revolution. James Melone, Prudential's president, characterized them as a means of "getting the message across that we are going to move with less layers of management and more self-contained business units." These are strong words for an industry that traditionally has rivaled many government agencies in the broad range of management titles used. He viewed a key purpose behind downsizing management layers as getting accountability and responsibility more closely identified with each other.
Merrill Lynch and Co. , another major player in the financial services arena, has given attention to excess mid management staffing. More than 4000 positions have been eliminated, many in areas such as accounting, marketing , planning and product development. Not only headcounts are being reduced. Reporting relations are being changed so each manager has at least six direct reports.
Executives have stopped boasting about having "management in depth." That phrase is no longer an unquestioned positive expression. Instead attention is on minimizing the management structure between chief executive and customer. The hoped for objectives are similar across companies: faster decision making, quicker awareness of market needs and competitor moves, and lower costs. And many companies have found a key side benefit: higher morale among the managers remaining. Why? The decentralization that frequently accompanies de-layering gives them bigger jobs with more real responsibility. Some feel able, for the first time, to really do their work because they are finally in control of their own operations. Middle managers are often the first to be aware that their bosses' job is not designed to require a full time effort, leaving the boss too much time to meddle in their subordinates' work.
In this chapter we will shift attention away from headquarters staff and consider the other half of the downsizing equation: reducing the number of management layers in the corporate hierarchy. In doing so we will examine several key issues:
- How much does it cost to manage each employee?
- What do managers actually do?
- How many layers of supervision does a company need?
- How many people can one manager manage?
- How can a manager's span of control be increased?
After reviewing these simple to ask, but hard to answer, questions we will look at several persistent problem areas. Companies seem to have chronic difficulties managing staff departments, keeping bureaucratic clutter out of their top management structures, and making matrix arrangements work. Finally we will consider how management needs to structure itself when its primary goal is nurturing innovation.
What does it cost to manage
The technique of activity cost measurement described in Chapter Four can help shed light on these issues. In addition to identifying misplaced staff work, it can be used to calculate the actual cost of management in each department. This can allow you to compare differences in management costs among divisions or departments in your company. It can also provide a basis to compare management productivity of companies within an industry, but as with staff ratios these are only rough indicators. One study that compared various companies' activity costs found that, on average, insurance companies pay fifty-five cents in management salaries for every dollar of non-management pay. Banks are just about as costly, while manufacturing based businesses who tend to have more experience managing productivity improvement average only twenty-three cents in management expense per dollar of worker cost.
The real payoff in analyzing the cost to manage each payroll dollar is to use the data to see how much money can be saved if managers' spans of control are increased. One petroleum company averaged spending $0.33 to manage each $1.00 paid to its employees. It did this through an organization structure that had, again on average, a manager for every 5.8 workers. If this oil business were to improve this span of control so each manager, on average, had 7 direct reports, its potential cost savings would be $25 million each year. Researchers who have examined costs to manage employees in a variety of industries have found that companies with spans of control of three spend almost four times as much to manage each payroll dollar as do firms that are able to have average spans of eight.
Measuring the average cost of managing an employee is more difficult than simply dividing the manager's salary and benefits by the number of direct reports. That calculation assumes that 100% of the manager's time goes into managing. This is an assumption that many companies implicitly make, after all "that's what we're paying him to do, aren't we?" But, as many subordinates know, their boss does not really need or use all of his or her time to manage them. An activity cost analysis of where managers' time goes can be enlightening in this regard. This provides a quantification of how much time is spent managing direct reports, people reporting to direct reports, and how much goes to individual contributor tasks. Some studies have identified managers who spend less than half their time actually managing, and have spotted some who spend almost as much time "managing" their subordinates' direct reports as they indicate they spend on the subordinates themselves. This last situation should call into question either the subordinates' capabilities or the manager's sense of what his job really involves.
Identifying opportunities for net reductions in management positions is the primary objective of these company-wide analyses. Multilayered companies frequently have these "part-time managers." As we discussed in Chapter One, corporate compensation systems often encourage the creation of part-manager, part-individual contributor jobs like these. Regardless of their cause, their existence frequently means companies are paying management salaries for non-management work. They suggest opportunities be identified to reassign the non-management work and increase the managers' direct reports. Frequently these "part-time" managers are those with relatively few subordinates. Sometimes this has been caused by an earlier downsizing which eliminated several subordinates' jobs without considering the impact of the elimination on their bosses'.
What do managers do when they
If you ask this question of managers the answer will be probably be along the lines: "I'm, uh ah, responsible for many things...." (then they start to list the responsibilities of their subordinates). Or they may say: "Well, I make sure that everything goes all right." True responses, but not very reflective. Unfortunately many academic observers have not provided much better. Few have had actual experience managing and some of the categories they use to describe what a manager does are rather general: plans, organizes, controls, coordinates, motivates and develops subordinates. Some of these tasks, such as planning, controlling and coordinating are responsibilities of non-managers in many companies. A few observers have come a little closer to reality when they call managers "gamesmen," power balancers, and people who put souls in new machines.
But some of the most useful studies of managers are those made by people who closely examine their actual daily activities, rather than trying to force fit what managers are supposed to be doing into some abstract categories. One of these, Henry Mintzberg, spotted ten characteristics of management work:
1. Most managers, because of their formal position, must represent the company as a figurehead.
2. They also devote considerable time to serving as a liaison with other managers and people outside the company, from whom they need both information and assistance.
3. They must be perceived as leaders, setting a direction
4. Through the contacts they make in the first three parts of their job they monitor information about their business and what is happening to it.
5. They also disseminate information to their reports and others in the company.
6. Some managers, formally or informally, serve as a spokesperson representing the company to the outside world.
7. To varying degrees managers serve as entrepreneurs to help initiate innovation and change.
8. Many managers are responsible for handling disturbances which threaten the company's normal course.
9. At least within their area of responsibility, managers serve as the primary resource allocators.
10. Many managers negotiate with employees or outsiders in behalf of the company.
Doing these tasks implies individuals who are more active than analytical, able to quickly shift gears from subject to subject rather than spending days on one activity, and probably more attuned to talking than writing. The nature of these tasks also implies that they will be done by relatively few people. There are limits to the number of entrepreneurs, resource allocators and negotiators a company can contain without going off in too many directions at once. There also are constraints on the number of coordinators, disseminators and spokesmen if the intention is to keep communications relatively free of distortion.
How many management layers
do you really need?
Why is information distortion such a critical issue? Cannot the problem be solved by computers on executives' desks tapped into large, central data bases? Not when you consider some of the types of information managers value most. It is not so much the spreadsheet records of past performance, but the "softer" indications of what future results will be. This information, often qualitative, gives the executive a way to calibrate the likelihood of business forecasts coming true. How confident does the sales manager seem about his forecast? What things are just starting to bother our customers? What novel uses are they making of our products that our market planners never anticipated? Is morale in an old plant high enough to justify our optimism about the results of next week's union election, or should we budget for higher wage costs?
This kind of information is usually communicated verbally and visually. It is the stuff many managers thrive on. It has a great deal of difficulty working its way up a long chain of command, just as instructions about a shift in strategic direction have a long way to go in some companies before they reach the key employees on the firing line.
Too often chief executives on visits to factories or sales offices are amazed to see some of what they thought were clear statements of new company policy being totally, but not necessarily maliciously, ignored. They may find toxic chemicals still being discharged into nearby rivers., or Blacks and women kept from mainstream management jobs. Some are as surprised as John Kennedy was when, in the middle of the Cuban Missile Crisis, he found that outmoded rockets in Turkey he had ordered removed long ago were still there and being used as an excuse by the Soviet Union for its arms in Cuba. They often become very angry, as was Kennedy, at subordinates, but seldom blame the real culprit: their elongated organization structure.
When managers pass orders on to their subordinates they usually accompany them with their explanation of why they are necessary along with a qualifier or amendment. What happens several layers later is that the original directive is distorted and the accumulated qualifiers and explanations have taken on a life of their own. When a subordinate hears two messages, the official policy from "on high," and his immediate bosses' interpretation of it, which message is most likely to be acted on?
Some executives try to go around these problems by using expensive communications campaigns that carry their messages directly to each employee. Some use elaborate videotape systems, others regularly speak over the factory public address system. A few have even taken to public radio or television commercials, not to sell products but to convince employees listening at home that something major has changed back at the office. These efforts, sometimes of heroic proportions, are expensive. They usually require extensive headquarters staff support, and at best they only get around the real problem.
The real problem is more than just too many managers. Reducing the management headcount without modifying the organization structure may save money in the short run, but it will not necessarily improve your company's effectiveness. Doing this requires addressing the number of levels in the company. This is a lesson many Fortune 500 companies have learned. A Michigan State University management professor, Eugene Jennings, noted that since 1980, 89 of the 100 largest U.S. businesses have reduced their number of management layers.
But how many layers can be removed? How short can a company's hierarchy be? If the idea is to eliminate corporate bureaucracy, how can the bureaucracy be identified?
Herbert Rees, the president of Eastman Kodak Co.'s internal ventures subsidiary, defines bureaucracy as the layer, or layers, of management between those with decision- making authority on a project and the highest level person working full-time on it. This may be a useful working definition for those concerned with managing innovation. For those in more steady state businesses, the example of the Roman Catholic church suggests one outer limit for minimizing layers of authority. The church hierarchy places only one level of authority between the Pope and the parish priest, the bishops. And the Pope himself does double duty as the Bishop of Rome. Of course most companies lack the church's sense of mission and impose lesser entry requirements on new recruits. And the Pope has a strong headquarters staff organization and makes considerable use of committees and task forces to help with his management responsibilities.
In the secular world People Express was able to successfully challenge - for a time - the giants of its industry with four levels. Perhaps it set a more relevant outer limit for flatness of the corporate pyramid, although as we will consider in the next section very lean structures seem to only support very lean business strategies.
Another way to get a fix on how many layers are appropriate is to examine the competitors you are up against. At the time of the Bell System breakup, American Telephone and Telegraph Company faced companies like five-tiered Executone and six to seven leveled Rolm with an army of managers arranged in a ten or eleven deck hierarchy. Making these comparisons is a form of management benchmarking can be a useful supplement to that described in Chapter Four to help judge staff size. It can raise a warning flag when the difference is significant. But it still does not provide a precise norm that takes into account all the characteristics of an individual business. And it is not very helpful if your competitors are also over- layered.
This is an area where few management scholars have ventured. Some who espouse the contingency ("it all depends") school of management observe that tall, many layered structures are more common in mature companies in relatively stable businesses while short and flat structures more prevalent in companies operating in more turbulent markets. One other theoretician, though, has gone farther than most in analyzing structure and layers. He is Elliott Jacques, a sociology professor based in England with a Ph.D. from Harvard and an M.D. from Johns Hopkins.
Jacques maintains that even the largest worldwide corporation can be managed with no more than seven managerial levels - including shop floor supervision and the chief executive. Smaller businesses will require correspondingly fewer levels.
Jacques has studied organization structures from ancient China and Greece to modern multinationals. He noticed almost all were configured in a pyramid shape. He believes this form is determined by the distribution of human capacity in large groups of people. If everyone had similar work capacities, organization structure would tend to be very flat. If only a small part of the population had low capacities, with the rest ranking very high, then most structures would be shaped like an inverted pyramid.
His use of the word "capacity" is not intended to be confused with intelligence or what achievement tests try to measure. Instead Jacques is referring to what he calls an individual's "time-span of discretion." This time-span is the maximum amount of time in which a person is given to use his good sense, knowledge, experience and skills to finish an assigned task. While, he believes, people have differing potential abilities to function at greater time spans, a person's time span at work is that implied by the kind of work his boss gives him.
He feels companies have recognized this through their pyramidal organizations. But where he feels many companies have gone wrong is that they have tried break time spans into too many subdivisions by the way they compartmentalize assignments and degrees of authority.
What breakdowns are appropriate? To answer this Jacques has dissected hundreds and hundreds of assignments given by managers in countries throughout the world. In trying to identify target completion times, something many bosses themselves are a little hazy about, he asks many questions about what subordinates are expected to accomplish. Then for each of these he teases out the answer to "how long are you giving them to do it?" The results of these studies surprised him.
He found there were fairly clean breaks between these ranges:
- Up to three months of discretion
- Three to twelve months
- One to two years
- Two to five years
- Five to ten years
- Ten to twenty years, and
- Twenty years and more
Most employees are in jobs with tasks that fit into the first of these categories. They have from a few minutes to three months until their projects must be completed and evaluated. First line managers tend to have work that fits into the three months to a year slot, while second level managers range from one to two years. The discretion allowed to general managers with profit-loss responsibilities ranges, in his studies, from two to five years with more senior executives involved in tasks taking even longer to bring to a conclusion. His data is less complete at the upper end of this spectrum where he suggests that managers in his eighth tier are implementing twenty year strategies. Several executives, such as Thomas Watson, Sr. and Konosuke Matsushita seem to easily fit this level.
Businesses may also be classified using these breakpoints. A one-person show (a door-to-door salesmen or painter) would fit the up to three months span. A second level enterprise might be a small store with an owner-manager and several assistants. The one to two year span may be adequate for an organization of several dozen to several hundred employees since it implies the existence of a middle management tier with a three to twelve months span of discretion .
A fourth rank organization might be a stand-alone small company with several functional managers reporting to the chief executive. These managers would have two levels of management below them. This structure might also be appropriate for a division of a larger enterprise.
The next level of hierarchy ( a company with its leader having responsibilities that cover a five to ten year time horizon) could accommodate up to 6000 employees, Jacques says. Next are holding company-like organizations that involve spans of over ten years for their chief executives. This strategic horizon also characterizes the group or sector structure of some giant corporations. He feels these can include 20,000 or more employees within a six-level management hierarchy, ten to twenty divisions and a group headquarters of less than one hundred people.
Finally, Jacques maintains that a company with an additional management layer should be able to employ hundreds of thousands of workers, assuming that the other levels are appropriately organized and managers given full reign over tasks within their respective time horizons. This seven-level structure should be able to accommodate most of the world's businesses, and public agencies.
While Jacques has advanced a number of psychological explanations for the existence of these categories and break points, what is more interesting is what he observed about their implications for managers.
One of his findings is what happens when managers are perceived by their subordinates as "straw bosses." These are situations when those being supervised feel their immediate manager has only limited ability to make decisions about them or their work. They look toward their boss' boss as their "real" manager. These situations are common in many companies, and usually indicate too many layers of supervision are present. In most of these cases the "straw bosses" occupy the same time-span level as their subordinates. There is not sufficient variation in their degrees of discretion for them to function as real managers of these people.
The U.S. Army has hired Jacques to help them rethink their chain of command, perhaps to help avoid repeating the mistakes described in Chapter One. A Pentagon colonel admitted his theory explains what should work for the army, but also confessed that extra layers in need of weeding out have grown in. While Jacques has been conducting research along these lines for decades his work has only recently started to be noticed in the U.S. In addition to examining organization structure, he has strong views about the inappropriateness of most pay and job evaluation systems. Possibly the controversy around these ideas has limited the exposure of his other, more easily applicable, ones.
His ideas have not been given the attention they deserve by American managers. They should at least provide this challenge to organization planners: if your company has more levels than time-span theory implies are needed, then you should be able to clearly specify the value being added to the business by each additional layer. Determining how many levels you need is a process of triangulation. One reference point is your current situation, another your competitors' structure, and a third is Jacques theory.
Is it possible to have too
People Express, a pioneer low cost, low price air carrier, now folded into the Texas Air Corporation system, received widespread acclaim for its innovative management structure. It hoped to eliminate the customary split between workers and managers by making all employees managers. The conflicts that traditionally emerge between staff and line were dealt with by dividing the work so that the same group of individuals performed both tasks.
At a time when its established competitors such as American, Eastern and TWA operated with seven to eight levels of management between chief executives and pilots or passenger agents, People Express had only four. This was less management hierarchy than the traditionally cost conscious Northwest Airlines (6 layers) or some of the other deregulation spawned carriers such as Air 1 (six layers) or New York Air (seven layers) employed.
The keys to People Express' lean structure were:
- Relying on extensive employee self management. New recruits were selected based on their potential for individual initiative tempered by an ability to work well in team settings. Job candidates showing needs for close supervision and desires to work in structured surroundings were screened out. Each recruit was given the title of manager, well indoctrinated in People Express's operating philosophy, and required to purchase stock in the company to help them feel more like owners than employees.
- Minimizing the number of job categories. Aside from about twenty senior managers, most of the airline's work was divided into three categories: flight management, maintenance management, and customer management. Flight managers did the work traditionally expected of pilots plus, on a rotation basis, much of the headquarters staff jobs of weather forecasting, schedule preparation and planning for new routes. Maintenance Managers kept the planes flying and Customer Service Managers did everything else. Everything else included staffing the departure gates, selling tickets and beverages in-flight, and handling much of the paperwork on the ground. Work that did not fit these categories, such as baggage handling and telephone reservation taking, was contracted-out.
- Sharing the "big picture" with all employees. This was done through an extensive communications program, aimed at the airline's employee-owners, and cross-functional skill training. The training helped people appreciate the requirements that other employees had of them so they could do their jobs properly, and made it possible to move the workforce around easily to quickly respond to business changes.
These practices and others were hailed in the early 1980s as examples of how the company of the future would be managed. They were even documented in a Harvard Business School case. Low fares and frequent schedules helped fill the planes and People Express set off on an ambitious route expansion and acquisition campaign. The carrier soon flew cross country and then to Europe.
But as one newspaper account of its just-before-bankruptcy purchase by Texas Air noted "success doomed People Express." Its competitors, already experienced in managing nation-wide operations, imitated its prices, but not its organization structure. They lowered costs through threats of bankruptcy, union give backs, and low wages for new hires. In addition to these rebounded competitors, People Express' expansion did it in, organizationally and financially. It went so deeply in debt to finance expansion that annual interest payments exceeded $60 million - far more than yearly operating revenue. And it found that its unique minimally layered management structure worked far better for a medium sized airline than a large one.
Rapid expansion meant rapid influxes of new hires, more than could be thoroughly socialized into People Express' ways. The company became less homogeneous, and more difficult to run by philosophy instead of direct supervision. And as the competition became more sophisticated in its responses, People Express' staff generalists were not able to do as good a job analyzing profit yields and allotting capacity to new routes as were its rivals' full time specialists. One securities analyst commented: "They organized themselves with a very lean organization with no fat in it, but not much muscle either." He stressed that there are some important staff functions that just must be performed. While People Express did attempt to expand its management structure, the changes took place well after it was implementing its high growth strategy. This upsized strategy was not facilitated by a severely downsized organization.
This is an issue that has plagued Fortune 500 companies as well, including some most admired for management innovativeness. Both Minnesota Mining and Manufacturing and Hewlett-Packard have faced, and addressed, problems of missing management layers.
3M has been famous for its abilities to turn home made inventions into quick profits. But as many of its markets matured and copy-cat versions of 3M products appeared more quickly, the formula for money making became more dependent on concentrated marketing efforts than rapid product development. Also office customers were increasingly interested in buying complex, integrated systems not virtuoso stand alone devices. Responding to these demands required a reorganization of 3M's operating units, heretofore based on the concept of the small, semi-autonomous product-oriented division.These groups gave 3M much of its entrepreneurial energy and style, but at a cost of only loose coordination and communication among them.
Hewlett-Packard faced similar problems. Its historical decentralized management philosophy assigned the design and building of products to small divisions, but gave responsibility for their sale to other decentralized marketing units. The result, as described by the head of one of their customer associations, was "three or four companies that don't seem to talk to each other." Products overlapped, important new markets were approached on a piecemeal basis, and technologies developed in one division were sometimes slow to migrate to others. Push came to shove when HP needed to build unified computer systems to compete in the marketplace against IBM and other more integrated manufacturers. Then it became apparent how its loose organization design could work against developing families of products.
Both companies realized the necessity for change but were very concerned about keeping the spirit of entrepreneurship alive by simulating a small company environment. 3M first tried to use a task force on office systems to coordinate across division lines. HP initially designated an executive as "program manager" with broad authority to tap whatever divisions were necessary to provide equipment or software for a new computer line. But these half steps met with limited success.
HP turned to a regrouping of its dozens of product divisions under sectors that are focused on markets rather than product types. A second new management layer was created when the position of chief operating officer was created and a centralized sales force was based in a new corporate market division. 3M made similar moves. A sector organization was set up putting their ten business groups (each with three to six small divisions) into one of four new sectors. This allowed the span of control of the top executives to be cut in half, freeing them to concentrate on 3M's long term strategies.
These moves were made very cautiously by each top management. 3M involved many managers in task forces to plan the details of the reorganization and HP has used employee surveys and communications programs to ensure that top management and workers are not overly distanced from each other because of the new layers.
We have considered these three companies' experiences to emphasize that minimizing management layers is not the right prescription for every situation. The popularity of operating lean and mean should not cause companies to downsize when they need to expand management. The starting point of an effective streamlining effort must be the strategic plan. Both 3M and HP took theirs into account before tampering with organization structures that had received wide acclaim. At People Express, unfortunately, an admirably lean structure could not keep pace with an expansive business strategy. Perhaps all the favorable external publicity it received for its slim structure made it harder to modify the organization. In spite of the airline's problems, the People Express innovations are still worthy of consideration by many companies. They suggest ways to minimize bureaucracy when economic considerations indicate a smaller scale operation will be most successful. They also give an indication of just how far management hierarchy can be pruned. In Chapter Eight we will return to this issue of what recently slimmed companies must do to operate safely with fewer layers and staff.
How many people can one manager
Many management theoreticians and practitioners have tried to specify the optimum point. It was also an issue of concern to nineteenth century militarists. Napoleon felt five reports was the greatest desirable span; Karl von Clausewitz, advisor to the Prussian army, maintained ten was more appropriate, perhaps based on a view that German soldiers were more disciplined than their French counterparts. Henri Fayol, a French manager and early management consultant, was more flexible in his judgments. He suggested fifteen direct reports for lower level supervisors but only four for senior executives.
Other management observers tried to suggest more of a theoretical basis for these seat of pants judgments. Some referred to studies that indicate that, on average, humans' span of attention is limited to six or seven things at once. From this they reasoned that managers should not be responsible for keeping track of more than that number of direct subordinates. The most quoted of these, an Englishman Lydal Urwick who combined a military and management background, stated as a principle of management: "No supervisor can supervise directly the work of more than five or, at the most, six subordinates whose work interlocks."
Sears, Roebuck, either in ignorance or disregard of these theoreticians, managed the post-World War II growth spurt with its headquarters merchandising vice president having forty-four senior managers reporting directly to him. Their store managers also coped, somehow, with supervising about forty department heads each.
More recently academics have tended to favor the Sears model. As Theory Y and participative management became in favor the importance of organization structure was deemphasized. More attention was given to a by-product of the Sears, Roebuck experience, deliberately giving key executives so many subordinates so they could not exercise close supervision over them if they wanted to. Ironically in the 1960s and 70s, just as organization observers seemed to lose interest in classical issues like span of control, many American companies went on a binge of adding managers and layers. Only recently have they started to shed these.
Despite these many years of pendulum swinging academic inquiry, the most honest answer to the question "how many can one manager manage?" is the not so helpful: it all depends.
Depends on what? This is an issue in which research and consulting experience can be more helpful. There are a series of factors which characterize a manager, his or her subordinates, the work they do, and the organization setting in which they operate that can serve as a yardstick to help evaluate each managerial situation. This last point is the significant one; each manager's situation must be evaluated individually. While there are across-the-board actions a company can take to broaden spans of control (such as investing in a computerized information network), determining the exact number of new reports each manager can cope with must be done one-by-one.
Here are twelve of the more important factors with which each situation can be judged.
1. Nature of the work being done. What exactly are the subordinates doing? Is it simple and mechanistic or complex and creative? How critical are their jobs to the overall well being of the company? How routine are the responsibilities? How many of them are being done by the subordinates for the first time? By anyone for the first time? What is the time frame involved: does the cycle from start to finish of a typical assignment last a day or a year? This comes back to Elliott Jacques' key concern: what is the maximum time a subordinate continues to exercise discretion without the results being reviewed by an overseer?
2. How interdependent are the subordinates. Here the principle is simple. The more closely related the work of one subordinate is to another, the more likely their manager will be involved in coordinating workflow and integrating work products. The more the manager is actively doing these things the fewer of these subordinates he or she can easily manage. The opposite also holds. To the extent the subordinate's work is relatively self contained and can be done independently of the manager's other direct reports, the less of an integrating role needs to be played and the more subordinates can directly report to the manager.
Peter Drucker expressed this idea well when he commented that the real issue is not how many people report to a manager, but how many of them that have to work with each other report to a manager. A former Wharton professor, Jay Galbraith, has elaborated on this concept by characterizing three possible types of interdependence among boss and subordinates. Each has different span of control implications:
- POOLED EFFORTS. Here the work of each subordinate is relatively independent of that of the others. They rely on their common manager to set a general common direction for their efforts and to provide the pool of budgetary and other resources they need. This arrangement is common in staff units where an executive may oversee the work of many specialist departments in unrelated fields (such as legal, personnel, and public relations). This executive may still be kept very busy making sure each of these departments is serving the best needs of the company as a whole, but this is a task that is potentially more easily delegated than that of internal coordination.
- SEQUENTIAL EFFORTS. In these the output of one subordinate is the input of another. Here the situation creates greater likelihood that the manager becomes involved in coordinating, scheduling and resolving conflicts among subordinates. This is typical of offices that have some of the characteristics of an assembly line operation.
- RECIPROCAL EFFORTS. Here the work of one group becomes the input of a second and the second's output cycles back to become further input to the work of the first group. Some complex manufacturing processes are characterized by this type of interaction. So are the relations between a product design group and a marketing group when they are working together to produce an innovative product breakthrough. Then the overall manager may become deeply involved in keeping the work steadily progressing. Often in new product development situations this manager is the one to announce when innovating time is up and the time to start selling the product is at hand. In these reciprocally interdependent supervisory situations the manager is most likely to be forced into the details of the subordinates' work.
In general, reciprocal situations tend to favor narrow spans of control. Pooled efforts can tolerate relatively wide spans, and sequential work comes out somewhere in the middle.
3. How well trained the subordinates are. The more highly skilled subordinates are at doing their jobs and coordinating their work with others, the more of them a manager can manage. The converse is also true. In many large corporations the work of a manager's subordinate will also be management. Spans of some upper mid managers are narrower than the work itself seems to dictate because they are compensating for some of their direct reports' deficiencies by supervising their reports' reports.
4. How well trained and capable the manager is. The more a manager is still learning the ropes, or is basically a better individual contributor than a manager, the fewer direct reports he or she can safely handle. Management expert Arch Patton has been concerned for many years that the most able managers have been attracted to staff work rather than line management. He feels that if better skilled people held line jobs, fewer staff positions would be needed and the span of control of the line managers could be broadened.
5. Manager's tenure. It also seems reasonable that the longer a manager stays put the more likely he will develop ways to effectively deal with his current span of control. Over time this "learning curve" effect should permit long tenured managers to gradually increase their number of direct reports. Unfortunately many companies move managers too frequently for this effect to take hold. In 1985 the average U.S. corporate manager stayed put for only 4.5 years.
6. Subordinate turnover. The more rapidly the positions reporting to a manager turn over, the more time is needed to orient and train the newcomers and the less available to handle a broad span of control. The longer the tenure of a manager's subordinates the greater is the likelihood that trust and understanding of each other's needs and capabilities will make the supervisory job less difficult. As with managers, high employee turnover is a problem in many businesses. The average U.S. employee moved from one job to another less than every four years, according to a 1985 study.
7. Subordinate rotation. One type of subordinate turnover can facilitate broader spans of control. This involves an organized system of lateral rotation where subordinate A may swap jobs with B, C with D and so on. These moves must be made carefully to avoid management by musical chairs. They cannot be done too frequently and their success often depends on the investment made in cross training and career planning. But they can have a downsizing payoff by increasing each subordinate's understanding of each other's work and thus allowing informal coordination and mutual accommodation to replace some top down direction and conflict resolving. Where the nature of the work requires sequential or reciprocal efforts, well planned rotation can help counteract the tendencies of this kind of work to necessitate relatively narrow spans.
8. Strength of lateral communication channels. The faster and more accurately information flows horizontally in a company or department the less of it needs to move up and down. A lot of important management time is spent "bee buzzing" : learning something useful from subordinate A and telling it to subordinate B. As long as the manager maintains some way of staying informed about the critical activities of his or her people, it is usually more efficient for B to learn directly from A. Subordinate job rotation can help increase this kind of interaction; so can well placed coffee machines and open plan offices. In general, one way to free managers to manage more people is to relieve them of part of the department town crier role.
9. Investment in automated information systems. Unfortunately creative architecture and coffee pot placement cannot carry the entire burden of communication. In many situations geography and information detail require computer based systems. To the extent these are designed to facilitate information exchange among subordinates, rather than only increasing the level of detail their manager is exposed to, these can open possibilities for broader spans. Too frequently the investment companies make in systems serves to allow managers to better second guess their staff - in effect, to do their work - rather than to increase management productivity. In the pre-computer era a management theorist once advised that managers should be limited to "seven or so" subordinates because the traditional management tools of the time (yellow pads of paper and wall charts) would not allow them to cope with any more direct relationships and interrelationships. Maybe so then, but computer and telecommunications networks provide tools with greater potential for control and communications.
10. Number of performance measures needed. The fewer measures of performance a manager needs in order to know subordinates are on course, the more subordinates can be overseen. In some retail corporations it is possible for one executive to manage 20-25 individual store managers because a single number (profit contribution) can be calculated frequently (weekly) for each store. And in this business this number is frequently the most critical indicator of good performance. These executives practice a form of corporate triage. They classify stores as exceeding profit targets, meeting them or losing money. The first and last categories get most of the executive's attention, either through positive recognition or turnaround assistance. Most stores are in the middle group. In a de facto narrowing of the executive's span of control, these stores' managerial needs are primarily met through several staff service groups.
The converse is also true: the more numbers or non-quantifiables are needed to measure good performance, the fewer managers can be watched. This consideration often works in tandem with the investment made in information systems. The chief executive of one of Europe's largest retailers admitted he was comfortable giving his store managers a great deal of what appeared to them as operational autonomy because the automated information system he had in place served as a safety net to allow headquarters to quickly spot weakened performance
11. Use made of job enrichment. Job enrichment (sometimes mislabeled job enlargement) is the term given by industrial psychologist Frederick Herzberg to efforts made to consciously build self motivating factors into the way jobs are designed. When a manager redesigns a subordinate's job with the hope of enriching it ways are found for the subordinate to:
- Assume more responsibility
- Receive greater recognition for accomplishments
- Feel that something concrete is being achieved
- Develop skills and talents
- Feel advancement in the company will come as a result of strong performance
While most managers intellectually agree these factors are important, relatively few actively manage their subordinates in a way to bring these abstractions to life. Doing this usually involves more delegation on the part of the manager, which can set the stage for broader spans. Unenriched jobs (the ones most of us occupy) tend to involve fragmented work done by marginally motivated people. They seem to require relatively close supervision.
12. Existence of a shared culture. A corporation's culture is a control mechanism as much as it is anything. The existence of strong, widely shared values and beliefs, reinforced by common management practices and ways of relating to each other, can all make a company more manageable: "We can get away with a lot of decentralization here; everybody thinks alike anyway."
Everyone "thinking alike" is not something that happens by itself. Building a strong corporate culture requires a well thought out approach to recruitment and selection, investments in periodic employee training and retraining, consistent managers, and reinforcement from rewards and punishments. To the extent such a shared culture exists among subordinates, and between manager and subordinates, it is more likely they will know what to expect of each other. They will be more likely to trust each other ("because I know where he's coming from"). This allows some forms of explicit control - such as close supervision and staff police functions - to be replaced by implicit, trained-in controls. All of which open possibilities of operating with fewer managers and broader management spans. Mergers, restructurings and executive turnover, unfortunately, limit companies' abilities to build and maintain stable cultures. Strong cultures often depend on the continuity of direction that long service chief executives provide, but the average U.S. CEO in 1985 remained in place only four years.
Each of these twelve factors alone is a potentially dangerous generalization. They have more validity when taken together than considered individually. It is unwise to make span of control decisions based on only one or two or three of these because each proposition is true only to the extent everything else is held constant. But in the real world nothing ever stays constant. Careful planning to broaden spans must start with a consideration of all these factors together.
A potential 13th factor, give the manager an assistant-to or some other form of increased staff support to help him oversee subordinates, has been consciously omitted . While it is generally true these can help a manager broaden span of control, they also slow decision making, potentially increase confusion about who is really in charge, and add to costs. This is an issue where tradeoffs have to be made. Harold Geneen justified his strong staff by saying they helped him eliminate layers of management, and they got him closer to actual facts instead of layers of managers' opinions. For an individual with his unique talents this might be true, but apparently at a cost of building a machine only one person could operate.
The previous chapters urged movement of some staff functions to line managers. While this is a good way to downsize some headquarters units, too much migration of staff work can overburden managers and limit their time available for supervising additional subordinates. A balance needs to be reached. The strategic approach to managing staff activities described in the previous chapter suggests delegating only staff work that is primarily service-providing in nature and is relatively far along in its life cycle.
Too much staff-line interdependence can also narrow a manager's span. The need to closely work with several materials managers, staff design engineers, or accountants to get day to day work done adds the equivalent of one or several subordinates to a manager's roster. So downsizing the number and scope of these units may allow the manager to oversee more direct reports.
Tactics for stretching spans
Then consider the twelve factors above. Which of them might be applied company-wide to increase the overall average span?
Look carefully at staff departments for examples of small spans. Consider some of the alternative ways to organize them discussed later in this chapter.
Also look carefully at the line or operating management structure. Two key reference points are the corporation's chief executive and the first tier of operating general managers with profit and loss responsibility (division heads in many companies). Make sure that there is a clear reason for each layer of management between the operating manager and the first line supervisors. While many companies have reasonably broad spans in this part of their organization, it is still important to insure these key managers do not have too much bureaucracy between them, customers and employees. Then consider the number of layers between the division heads and the chief executive. Is value clearly added by other levels of general managers, group or sector heads and subsidiary officers? This is where the pyramid narrows in many companies. Are there any powerful staff groups or formal committees that, in effect, add additional layers? Can their contributions be easily justified?
Then move from the organization chart to whatever data has been collected about how managers spend their time. Consider converting part time management positions into full time ones by increasing their number of direct reports.
Any company-wide effort to broaden spans of control must eventually consider each manager's situation individually. This involves more than a quick redraw of the organization chart. As these twelve factors suggest there is no one appropriate span for every company. Many of these factors are people-specific, or they relate to the mission and nature of a department's work at a specific time. It is almost impossible to specify appropriate spans for every personnel department, auto manufacturing plant or sales force. For one manager, five direct reports may be too many; for another twelve may be too light a load. Data bases which purport to indicate the appropriate span for every quality control department or accounts payable unit are susceptible to burying truth in the data averaging.
Too many companies have implemented span-broadening efforts without considering many of these factors and how they interact. The result is often solutions that either suboptimize (moving from three direct reports to five when the manager and her situation could have accommodated eight or nine) or overextend the manager (who may be given what is for him an unsustainable load of seven others to supervise when the complexity of their work and his inexperience as a manager suggested that five subordinates was more prudent).
This approach is used by some consultants in their organization planning assignments. It can also be done in-house. An internal planning team of a human resources professional along with the manager of the manager whose job is being examined may be useful. Individual managers can also use this list of factors to suggest ways to improve their own situations.
While this template approach is reasonably comprehensive some managers may find it helpful to add other factors (such as geography) to their analysis. They may find their experience with other successful reorganizations indicates that for their company some of these factors should be given more weight than others. But in general most will find that reducing management layers by broadening a manager's number of direct reports involves some, if not most, of these:
- Maximizing the number of full time managers.
- Reducing the complexity and uniqueness of the work they supervise.
- Having subordinates able either to function independently of each other or to easily accommodate to each other with intervention from above.
- Providing extensive training to both managers and subordinates.
- Minimizing turnover.
- Encouraging well planned job exchanges among subordinates.
- Establishing strong communication channels to enable subordinates to coordinate with each other.
- Investing in automated information technology.
- Using relatively few performance measures.
- Enriching jobs.
- Maintaining a relatively homogeneous corporate culture.
Obviously these may not be possible for all situations. The nature of the work being done may be unchangeable. Rotation may be out of the questions for some employees. Turnover may be difficult to control. Building a homogeneous corporate culture takes many years. Setting up a microcomputer communication network may be too costly.
Changing any of these factors requires some investment. There is no free lunch. The issue to consider when planning a reorganization is: what level of benefit is expected to result and will it justify the cost?
How many is right: the Rule
- more than six layers between chief executive and first line supervisor and/or
- fewer than six direct reports per manager,
may have too much management. Determining if it does or not will take more detailed examination along the lines suggested above. The Rule of Six is intended to set an outer limit. It is a wide screen. Just because it has not been violated does not imply a company is appropriately organized. Many small companies, or large ones operating in swiftly changing markets, will rapidly approach bankruptcy , or severe loss of market share, with six levels of hierarchy.
Major problem areas
There are several chronic problem areas where this approach to span stretching needs to be supplemented with some creative organization redesign. This especially needs to be done when top management groups, headquarters staff organization, matrix structures, and promoting innovation are considered.
Downsizing staff units involves more than reducing headcount. As we considered earlier, if only this is done the result may be staff managers with fewer direct reports and more limited management responsibilities. Their management skills may be wasted unless the department's supervisory structure is reconfigured. Often difficult problems arise when companies try to streamline staff departments. We will consider these along with the special problems related to innovation and matrices below.
Determining how many senior executives are necessary and how they should be configured poses some special difficulties. Let's consider them first.
Top management clutter
Apart from personality and politics, there are some structural problems that have long plagued executive suites. These include:
- Too much bureaucracy at the top.
- An extra executive layer between the chief executive and operating management.
- Line and staff coming together too high in the structure.
Several companies have found faster decision making results from consolidating the jobs of chief executive officer and chief operating officer. Brunswick and Celanese eliminated the latter of these. Observers have even suggested the Celanese President and Chief Operating Officer voluntarily resigned because a reorganization that gave more power to the operating divisions in effect worked him out of a job. At Brunswick, the group vice president positions were also eliminated and all division presidents reported directly to the chief executive.
Crown Zellerback went even further in its downsizing decentralization. Cutbacks at the top of the company left a great deal of power in the hands of sixteen managers of large plants. They were given complete accountability for plant performance with each controlling credit practices and labor policies as well as marketing and sales. The overhead reductions this reorganization made possible have reduced some plant break even points by 30%.
Two giant chemical companies have also streamlined their top management structures to allow them to move quickly in spite of their size. Du Pont's executive committee members, its most powerful management group, now each have direct management responsibility for individual business lines. Previously many of them operated in more of an advisory capacity. Britain's Imperial Chemical Industries used a top level reorganization to help restore it to profitability. A layer of senior executives was removed and the board's size cut. Use of multiple management committees was curtailed and investment decisions that used to require months of deliberation are now made in days. This quickened response time reportedly allowed it to finalize a decision in one day to purchase Beatrice's chemical division.
A solution to the problem of the overburdened chief has caused problems of its own for a number of companies that tried it. Offices of the President, for a time a popular way to manage, are falling into disfavor in many businesses. About half the 50 major corporations which had chairmans' or presidents' offices ten years ago have eliminated them. Some of these formalized previously useful "kitchen cabinets," but Harvard Business School professor Abraham Zaleznik describes many of these attempts as "deserts." Harry Levinson, a management psychologist, has also been critical of these structures which have sometimes taken on a life of their own, leaving the rest of the company adrift. He looks at them as devices to avoid or postpone formal designation of powers. As gimmicks for often political purposes, Levinson has never seen them work well as executive tools.
In some corporations the most frustrating job is that of group or sector head. Originally a way to help top management oversee diverse technologies, products and markets, this is becoming less relevant as companies restructure and focus their operations. One group executive complained he was "little more than a high-priced courier between the chief executive and the operating units." These top managers continually walk a tightrope between allowing division heads entrepreneurial elbow room, and staying sufficiently informed and controlling to keep their boss comfortable.
For corporations with too many divisions for one chief executive to oversee, the group position may be the only alternative. The issue then is to maximize its usefulness while keeping it from becoming a second tier headquarters. Most companies tend to place similar divisions together, sometimes giving the group executive a marrow span of control. It may be possible to broaden this span, and increase the contribution of this extra layer, by grouping divisions by common strategic mission rather than similar products or markets. This way all start ups would be under one executive, all businesses oriented toward niche marketing and manufacturing under another, and all focused on aggressive market share acquisition under a third. What is right for one company may be inappropriate for another. This is an issue that needs to examined on a company-by-company basis.
Top managements are sometimes cluttered because too many types of executives report in to the same person. Operations that should have been coordinated at lower levels fill agendas in the executive suite. Some of this can be dealt with by better delegation and more cross-divisional task forces, but the most difficult involve coordinating staff and line executives. There are often significant status and salary differences between senior line executives responsible for half a billion dollar divisions and the officers who head the human resources, public affairs or legal activities. But in many companies these all report to the same person.
Several companies have adopted novel top level reporting relations that slim down layers while forcing executives to closely coordinate with each other. Exxon's top management committee long combined line and staff responsibilities, giving each committee member some of each. For example, at one point the senior vice president who watched over Esso Middle East also had company-wide responsibility for the legal, medical and public affairs staffs. The senior vice president dealing with employee relations also was concerned with Esso Inter-America and Exxon's mining and minerals businesses.
Alfred "Freddie" Heineken kept his worldwide brewing businesses tightly coordinated by giving his three principal deputies charge of staff functions that are vital to the success of each other's line operations. Each is responsible for results in a major geographic area of the world as well as for some worldwide staff functions. So while one person must sell as much beer as possible in the Western Hemisphere he must deal with someone equally concerned with sales in a different geographic territory because that person has charge of marketing and advertising. Clark Equipment Company has set up a system of four group vice presidents along these same lines as a way to prune back what even its chief executive called an over layered company.
Structures such as these require a close functioning top team to work, but they can provide important payoffs in developing executive talent and unifying staff and line operations. They also help develop and test successors for the chief executive's job.
The most glaring problem is one that often creeps up on large, relatively mature companies who have gone through periods of significant growth. Their management structures frequently have four or more significant levels: headquarters, group or sector managements, divisions, and plants. In the worst case of multiple staff layers each of these levels has fully developed staff departments that mirror those on the higher and lower levels. As Peter Drucker pointed out, layering such as this distorts communications making it difficult to maintain consistent staff policies up and down the organization. Decision making is slowed as more experts get into the act and overall lines of authority confused as staff department heads must balance the demands of their line manager boss with the oversight requirements of their counterpart one layer above. Some companies try to deal with these difficulties through "dotted line" reporting or some combination of functional and administrative definitions of accountability - but these usually only add difficult complexity to an already cluttered structure.
Other businesses, such as Harold Geneen's ITT, have tried to clear reporting channels by having all controllers report to headquarters, not to their immediate line manager. But this "solution" puts tremendous pressure on headquarters to do all the coordinating and takes away potentially useful support from the line managers.
Previously we suggested that companies carefully question the need for sector or group management. They also need to be sure this questioning does not lead to the circular reasoning that justifies that level primarily because of the staff units that exist there. These staff are most likely to be redundant in companies where it is possible to decentralize their functions.
Two approaches can be taken to deal with these layers on layers of staff. First, all operating responsibilities can be delegated to units at the plant or divisional levels. Then one level above these, either the group or corporate headquarters may maintain a small policy setting or monitoring function. In some well managed companies this can be done by one person. There may be instances, where the group is the logical operating level, with no staff based at the division level. Second a more radical delegation may make sense by selecting one of the four levels at which to base each staff responsibility, depending on the nature of the staff work being performed. Staff responsibilities would be then taken away from the other three levels. Both these approaches have to be planned on a company-by-company and staff-by-staff basis.
While reducing staff layers is more troublesome than just redrawing the organization chart, a more difficult problem lies in the management of the remaining staff departments. Advancing significantly in the line or general management hierarchy usually requires demonstrated management skill. Geneen's abilities were more in the skilled use rather than production of staff work. However, the staff manager counterparts of many line managers tend to have less supervisory experience. They often are promoted to supervisory jobs to reward and recognize their contributions as skilled analysts and professionals. In many companies they are given management titles and salaries so they will be perceived by their line management clients as of "equal" status. This bureaucratic practice is often ineffective and usually costly.
Many staff managers are expected to do staff work in addition to overseeing it. Staff managers themselves are not necessarily to blame for this situation. It is often driven by companies' job evaluation methods and reinforced by senior executives' expectations. More than one Vice President for Human Resources has been personally expected to counsel a retiring executive on how to handle his pension savings (although a member of the compensation staff three levels down in the hierarchy may be just as able to handle this). Considerations of status too often make this officer the "personnel rep" for the senior executive group. Heads of Information Services units, in addition to running their expanding departments, are called upon to personally counsel senior managers on personnel computer purchases. The double-pronged nature of many of these jobs makes it hard for their incumbents to work full time at management. And very frequently the number of direct reports they and their subordinate managers have is relatively low to accommodate this.
This is an area where companies can learn from the experience of professional practices, such as large law and architectural firms. Their partnership form of organization allows many to distinguish high status from management. Frequently they have a managing partner who is not necessarily the best lawyer or most talented architect or even not the highest paid person in the firm. Decisions about the direction of the practice are made collectively by the managing partner and the senior practitioners. With some modification this model could be appropriate for some corporate staff departments. Its use should allow the full time department manager to have more direct reports and less management hierarchy in the department.
Reducing the management hierarchy within staff departments is often a priority of downsizing projects. Staff groups are frequently a company's worst offenders regarding under spanned managers and unneeded layers of supervision. Some of this is due to their professional specialization and corporate willingness to tolerate accountants' wanting supervision only from other accountants, etc. But a lot of the layering results from an approach to organization that confuses status with place in the company management hierarchy.
Confusing status with management
We have already considered where these managers come from. What more streamlined alternatives are there? One way out of these problems is a structure with fewer, but full-time, management positions. Each has a span of control considerably wider than the part- time managers. Relative rank of the staff professionals is indicated by their job titles and pay. It is now possible to be a senior member of the department without having supervisory responsibilities.
Work is done in this streamlined alternative either by individual contributors or through teams. A team structure helps interlink the staff professionals as well as insuring that work is done by the minimum number of staff required. Some teams may be semi-permanent for ongoing responsibilities, others may be set up to complete a one-time project. Some may have changing memberships as a task goes through different stages toward its completion. Keeping this overall system of project management smoothly working is a key responsibility of the department manager. The various types of teams offer a variety of leadership opportunities for department members.
In addition to its potential for saving costs by helping to balance changing workloads, this structure allows department members to broaden the scope of their professional work outside their subspecialities through membership in multiple teams. This overlapping membership, if well orchestrated, can lead to better integrated staff products for the rest of the company. For example, it makes it easier to develop and maintain human resource policies that send a consistent message to all employees as some staff working on compensation and benefits may also contribute to developing the company's training policy.
Many companies have used such a project management approach to doing work that crosses traditional department lines, especially for technical development and new product introduction projects. But relatively few have tried this as a way to organize work within individual staff units. Today's pressure to downsize headquarters staffs may encourage more to experiment with this as a way to improve productivity and effectiveness.
Some Xerox executives felt at one point that their new product development efforts were strangled because of their matrix organization. Geography compounded the problems with key design and manufacturing departments based in Rochester, New York, reporting to separate executives at Xerox's Stamford, Connecticut, headquarters. The matrix led to endless debates over design issues and it did not designate any one executive with the priority of getting the new products into the market place. At times the only person who could resolve the conflicts this structure built into the development process was the president.
While matrices can be a useful way-station when a company is on the path from one strategic orientation to another (most frequently when changing from a functional to a divisional organization), it is usually a poor permanent choice. Its attempt at balancing opposing strategic perspectives is not a viable alternative to clearly deciding how the company wants to compete and how it should organize to do so. The conflicts inherent in two-boss systems can, over time, blunt a firm's strategic focus.
Two business school professors who helped popularize matrix organizations have also warned about their negative consequences. Stanley Davis of Boston University and Paul Lawrence of Harvard Business School suggest companies consider matrices only when nothing else works. Walter Wriston, former Chairman of Citicorp - a bank that has made widespread use of global matrix structures - admits "There are, indeed, easier ways to manage." The two Boston academics suggest only using this structure when, simultaneously (1) tightly coordinated decision making is necessary in two or more of a company's most critical activities, and (2) the work to be done is extremely complex and must be done amid conditions of great uncertainty, and (3) several resources vitally needed to do the work are extremely scarce. These conditions characterized the preparations for the first manned flight to the moon, but most business situations are less complex and intense.
When management consultants scan their clients' structures to identify opportunities for streamlining, the widespread use of matrices serves an early warning that changes may be needed. Alternatives frequently recommended include using the project management approach described above, instituting policies that encourage resource sharing, and setting up small self-contained divisions. Abolishing matrices is often recommended. When Xerox converted its copier business matrix into four strategic business units they achieved a 10% productivity gain. Eventually they were able to cut design time for some products in half.
Other companies, such as Exxon in its attempts to build an office systems business, have found that trying to grow new enterprises from the hierarchy appropriate for their mature base business is very difficult. Some companies have tried to get around these problems by setting up special organization units where some of the staff-driven standard operating procedures are suspended and hierarchy is kept at a minimum. Some, such as Lockheed with its famous skunkworks, keep these innovation oriented ventures far away from the headquarters. Others such as Air Products and Chemicals, Allied's New Ventures Group and ASEA's Innovation unit limit their management layers but keep them close enough to home to allow for easy transfer of their discoveries to the parent.
Managing innovation offers special challenges to the downsizer. While many innovation champions welcome the easier access to top management that layer reduction can bring, some also find that more visibility leads to greater expectations. They have fewer places to hide their "bootlegging" in radically streamlined operations. The successful new computer development documented in Tracy Kidder's The Soul of a New Machine duplicated an officially sanctioned project being conducted simultaneously at another Data General location. Vigorous cost cutting and elimination of all redundant activities might have also eliminated Data General's first 32-bit computer. To the extent the downsizing is motivated by short term cost cutting considerations, they may find far fewer dollars left for development projects. It is important to consider these potential downsides of downsizing for the innovator. While stripping away excess management can potentially make a company more hospitable to innovation, this will not happen just by changing the structure. And if the reductions involve major layoffs, the surviving managers may feel too insecure to deviate from the corporate norm for some time.
The bottom line
© Robert M. Tomasko 1987, 1990, 2002