CHAPTER 5

The Elements of Design

This chapter introduces the elements of design, the knobs that can be turned to structure an organization. We start this description from the ground up, with the design of positions (the cells of the organization’s body): their scope, degree of formalization, and the training and indoctrination they require. Next comes the design of the superstructure (the skeleton of the organization): how these positions are grouped into units, what size these units should be, and how much decision-making power should be decentralized to them. Finally, there is the fleshing out of the superstructure: the systems of planning and control and the lateral linkages to connect all these positions and units together.

Designing Positions: Scope

Positions can be narrow or broad, specialized or general. The pitcher in baseball is highly specialized—in the American League, they don’t even bat—whereas in cricket, many players “bowl” and all of them bat. In his 1776 book The Wealth of Nations, Adam Smith described what has become the most famous example of extensive specialization of labor—in the manufacture of pins.

One man draws out the wire, another straightens it, a third cuts it, a fourth points it, a fifth grinds it at the top for receiving the head; to make the head requires two or three distinct operations; to put it on is a peculiar business, to whiten the pins is another; it is even a trade by itself to put them into the paper.³⁹

Why so hyperspecialized? Because it is efficient. Smith noted that in one pin factory, ten men specialized in their work were able to turn out about forty-eight hundred pins per day, whereas if one had to do all this, the result might not even have been twenty.

Many people don’t relish having narrow jobs. Hence, two centuries after Smith came job enlargement, as in a call center that redesigns the operators’ jobs so that each can address a variety of customer questions instead of being limited to pat answers for a few of them (more about this in Chapter 8).

Designing Positions: Formalization

Organizations formalize work to proscribe discretion in carrying it out, ultimately to predict and control it, whether by specifying the work directly, attaching specifications to the workflow, or establishing rules to control the work across the whole organization.

As the author of this book, my work is minimally formalized: I have considerable control over what I write and how, subject to the rules of proper grammar, of course. But the work of the printers of this book is highly formalized. First, job descriptions specify exactly what each worker does. Second, along with their work comes a docket that specifies the size of the pages, the paper to be used, the binding to put on, and so forth. And third, workers in printing plants usually have to follow a lot more rules than do university professors—for example, when to start work and how much time is allotted for lunch.

If you believe that the pin-making kind of work is gone today, try flipping hamburgers at a fast-food chain. However, if you think that the work of professors and physicians is not otherwise formalized, consider all the rules that are promulgated by their professional associations.

Designing Positions: Training and Indoctrination

Here organization design specifies what skills and knowledge need to be brought to the job as well as what norms have to be assimilated there. You can get a job at McDonald’s and be flipping hamburgers in no time. But don’t try to walk into a hospital and expect to do surgery right away. One job is unskilled, the other, being skilled, requires extensive professional education, followed by considerable on-the-job training. Here is a list of training requirements for some jobs in a retail store: working in a warehouse (on the job); checker (one week at a training center); meat cutter (six weeks, part school, part on the job); butcher (normally two years on the job); and store manager (normally two years on the job, with an occasional week in school).

When an organization hires professionals, it surrenders a good deal of control over these people to the external institutions that have trained them—in other words, that have standardized their skills and knowledge—and to the professional associations that continue to upgrade and enforce these standards. Think of the protocols in medicine and the principles in accounting. Here is where we find true “communities of practice.”

In between skilled and unskilled work is that of craft: not formally trained so much as extensively apprenticed on the job, under the supervision of an expert. Thus, while there are schools for chefs, many of them learn by working under an experienced chef. Likewise, “professional” athletes are generally trained by their coaches.

Organizations that have unique cultures tend to be especially concerned that their people assimilate the culture and internalize the norms. Hence, they are inclined to develop customized programs, in house, to indoctrinate their people, or, if you prefer a less loaded word, socialize them. Think of an army with its boot camp or a company that rotates new hires so they get a sense of the whole place. Of course, some indoctrination is usually built into the professional training too, as when, besides courses such as anatomy, medical students are taught how to behave like proper physicians.

Designing the Superstructure: Grouping into Units

Here we come to the skeleton of the organization, the bones that hold its parts together. We consider how the positions are grouped into formal units, and how these units are grouped into successively larger ones to form the hierarchy of authority, also how large these different units should be.

Grouping gets so much attention that it has come to be seen as almost synonymous with structuring—hence the obsession with those organization charts. Well, just as there is a lot more to our bodies than our skeletons, so too there is a lot more to our organizations than grouping, even though they can no more exist without such grouping than we can exist without our skeletons.

Why Group?

We group for three reasons:

• To encourage mutual adjustment. By being grouped together, not only physically but also administratively, people are encouraged to communicate and cooperate. For example, a law firm may designate a specialized unit for those lawyers engaged in family practice, to encourage them to share their experience.

• To enable direct supervision. Even specialized physicians who work largely on their own need a chief—for example, to take the lead in hiring new people and to help resolve conflicts among them. Usually the manager can best hold things together in a unit.

• To attain a common result. The retail branch of a bank might put its agents selling insurance, brokerage, and trust services in the same unit to encourage more cross-selling.

Of course, encouraging all this within a unit can discourage it across units. Will those family practice lawyers be inclined to seek out their colleagues in corporate law when they need their advice? See the box about silos and slabs in organizations.

How to Group?

On what basis should positions be grouped into units and these units into larger ones? Here are several common bases for doing this:

• By what they do. The offensive squad in football scores, the defensive squad stops the other team from scoring. Manufacturing companies often group their workers by the different functions they perform, such as purchasing, production, marketing, and sales.

• By how they do it. The internists prescribe, the surgeons cut. The violinists play the strings; the brass play the horns.

• By why they do it, namely to achieve a common result. A diversified company assigns its laptop business to one unit, printers to another, customer service for both to a third.

• By where they do it. Miners work underground, dentists work on teeth, drinks are served at the bar, sales are made in Saskatchewan.

• By for whom they do it. Pediatricians do it for children, geriatricians do it for their grandparents.

• By when they do it. A factory has a day shift and a night shift.

Of course, these categories can overlap. Gynecologists, for example, are grouped not only by whom they treat but also by how they do so, even where they do so.

FIGURE 5.1 Stacking the Bases of Grouping

Some of these bases of grouping fall into two broad categories: by means (what and how) and by ends (why and for whom). Grouping by means favors specialization, so that people can learn from each other, but at the expense of coordination with other specialists. Grouping by ends does the opposite: it encourages coordination across the workflow but at the expense of specialization within it.

An important message in all this is that there is no magic formula for grouping positions and units, just a number of options that trade off one set of advantages for another. This means that some consultant can always find a better way to structure, which can also prove to be worse. Hence, here we have the motto of too many organizations: when in doubt, reorganize! Better still, de-layer—that is, delete a slab or two of those costly managers. Unfortunately, or should I say fortunately, grouping is no panacea for organization design, just one design element among many.

In one sense, however, organizations can have their grouping cake and eat it too. They can structure the way a campfire is built, by stacking the slabs first one way and then another. A manufacturing company can group by business function at the base, then by product line, finally by region (Figure 5.1).

Designing the Superstructure: Sizing the Units

How many angels can dance on the head of a pin? In the Middle Ages, theologians were mocked about asking such questions. Now the question is: how many positions can be grouped under the heels of a manager? And the answer is…five, or six. So claimed the classic management theorist Lyndall Urwick: “No supervisor can supervise directly the work of more than five or, at the most, six subordinates whose work interlocks.”⁴⁰

Well, then, why do one hundred of us at the McGill Faculty of Management whose work interlocks (to create bachelors, masters, and doctoral degrees) report to one dean, or that many workers in a factory to one foreman, or a like number of musicians to a conductor, whereas if you try to play hockey with a line of four forwards instead of three, you may lose the game because of a weakened defense? Must every one of these units be brought down, or rounded up, to five or six members? Obviously not. I imagine that Urwick had in mind VPs reporting to a CEO. But think of all the other arrangements that exist in organizations, indeed sometimes even in the case of those VPs. Enough of generalizing out of context.

The label is the problem. We commonly call this design element span of control, as if it’s all about control by a manager—namely, coordination by direct supervision. It is not. Size of unit is a better label, because other mechanisms of coordination come into play here. All those professors and workers and musicians can report to a single manager because their work is largely controlled by some form of standardization, not direct supervision.

In contrast, the small number of players on a forward line in hockey is explained by their reliance on mutual adjustment for coordination—convenient, frequent, informal communication. The greater the need for this, the smaller the unit has to be. Dozens of people in a unit can hardly communicate casually. (So go explain how lacrosse, similar in some ways to hockey, was played by up to a thousand indigenous Americans.) Hence, while a hundred professors reporting to a dean can be fine when it comes to teaching courses (with “area coordinators,” who are not managers, overseeing who teaches what), six can be too many when it comes to doing research in teams. And bear in mind that although such teams may designate a manager (although hockey lines do not), the work of that person may be more about linking the team out—for example, serving as its spokesperson—than leading it in.

Designing the Superstructure: Untangling Decentralization

After a century of discussion about the terms centralization and decentralization in the management literature, they remain as confusing as ever. But what else can we expect from the label de-centralization, as if centralization is the default position?

Consider an automobile company with five business units, one for each model of car that it makes, with the manager of each unit having most of the decision-making power. As we shall discuss at greater length later, General Motors in the 1920s was the famous case of this. Its CEO labeled this “decentralization,” in reference to the managers of the Chevrolet, Pontiac, Buick, Oldsmobile, and Cadillac businesses.⁴¹ But can a company that had so many thousands of employees, five of whom held the lion’s share of the power, be called decentralized? Maybe, compared with a company whose CEO at the headquarters had that power, but otherwise?

When all the power for decision making rests with a single individual, an organization can obviously be called centralized; when that power is dispersed more or less equally among everyone, it can obviously be called decentralized. On one side is an autocratic regime in government, on the other, a traditional kibbutz in Israel where the members share most everything (more on this later). Between these two extremes, however, is where decentralization gets interesting.

Decision-making power can be delegated vertically, down the hierarchy, more or less, or it can be dispersed horizontally, to nonmanagers, whether analysts, support staffers, or operators at the base. And this power can go partially or comprehensively, for example, for hiring decisions only, or else for most of the decisions pertaining to a unit.

With this is mind, we can consider various versions of decentralization. The decentralization just described at General Motors was vertical and comprehensive, but seemingly limited to the managers of the five divisions. When staff analysts have power over, say, the budgets that control the spending of everyone else, this can be described as decentralization that is horizontal, selective, and also limited. Alternatively, when professionals in the operating core control many of the decisions that affect them—physicians in hospitals, for example— this can be described as comprehensive horizontal decentralization.

Considering the coordinating mechanism in these terms, direct supervision is the most horizontally centralizing whereas mutual adjustment is the least, with the forms of standardization—first of the work, then of the outputs, next of the skills, and finally of the norms—falling in between, in that order.

Fleshing Out the Superstructure: Systems of Planning and Control

Once positions have been established and grouped into units, with their size and decisional power determined, there remains the need to put some flesh on these bones and flow all this together, as do the nerves of our bodies. Systems of planning and control coordinate by standardizing outputs across positions and units; liaison devices coordinate by encouraging mutual adjustment.

You are organizing COP77 in 2077, to deal with climate change . . . finally. There’s an awful lot to do, so you lay it all out on a spreadsheet, as a set of action plans: to choose the speakers, choreograph the events, cater the coffee breaks, coddle the lobbyists. You will establish one or more units to do each and decentralize considerable power to them to get it done. But for each unit, you need to establish performance controls of various kinds: schedules to meet, budgets to spend, and so on.

Action plans delineate intended targets, or outputs—what is to be achieved and when, but not how—while performance controls measure how successfully these targets have been met. Action planning is essentially top down, beginning, in theory at least (later we shall discuss practice), with the formulation of strategies by the senior management, which are deconstructed into specific projects, programs, budgets, schedules, and other operating plans for implementation by everyone else. Performance controls measure the after-the-fact results of these actions, from the bottom up the hierarchy (although these controls are usually designed near the top of it, by the analytical staff who report to the senior management).

For a graphic example of how targets can influence the way people behave, consider how scoring is tabulated in hockey. The player who scores a goal gets a point. But so too does the player who passed the puck to that scorer, and even the one who passed the puck to that passer. These are called assists, and they count for as much as goals. How’s that for encouraging teamwork? Compare this with singling out the CEO’s remuneration in a corporation, based on increases in its share price. How’s that for encouraging narcissism?

Fleshing Out the Superstructure: Lateral Linkages

All the design elements so far discussed can take structural design only so far, because much that matters in organizations requires lateral linkages of a less formal nature, to encourage mutual adjustment across the silos and the slabs. A whole set of these have found increasing application in recent times.

Most simple are liaison positions, designated to sit between two units and connect them together. A purchasing engineer can link engineering with purchasing, to keep the cost of components down. These positions have no formal authority, just the responsibility to find ways to get the two sides to cooperate.

Integrating managers take this a step further, by having some formal authority, for instance, over the resources available to the units. Thus a brand manager in a consumer-goods firm, responsible for a particular product, can use control of its budget to negotiate with engineering to design the product and manufacturing to produce it.

Meetings, standing committees, teams, and task forces also foster mutual adjustment. We know about meetings, which bring people together to discuss common issues, share information, and perhaps make joint decisions. Think about all such meetings needed to organize that COP77. Plus many meetings are impromptu: people just bump into each other, and discuss what’s on their minds. Other meetings are scheduled on an ad hoc basis—one time only—or else regularly, perhaps with designated membership, in which case they become standing committees, as part of the formal structure. Such a committee for COP77 might bring together, every Tuesday at 9 a.m., representatives of the teams that are dealing with various aspects of the event.

The board of directors of an organization is, in effect, a standing committee, with regular meetings to review its overall performance. Even the players of a hockey team can be thought to constitute a standing committee when they meet in the dressing room between periods to review the game so far.

These days, of course, many meetings take place virtually, over the internet. This may seem less formal, but it can be as orchestrated as a concert. Instead of a baton, the conductor has a mute button. Moreover, on Zoom people don’t bump into each other at the coffee machine. Where’s the key to press for serendipity? How to click for spontaneity? This wonderful technology can wreak havoc upon mutual adjustment.

A team or task force brings a group of people together temporarily, to accomplish a particular project—say, to develop a product or restructure an organization—and then disband, at which point the members return to their home units. A research team in a university might bring together professors from geography, geology, and geophysics to test some new theory.

A team can, of course, also bring in members from outside the organization—say, in this last example, a geologist from a mining company—or even be made up of people from various organizations, as is common in filmmaking, which makes substantial use of independent freelancers. A joint venture brings together people from two or more organizations in a partnership—for example, to develop a new airplane. (In Chapter 20, we discuss “Organizations Outward Bound,” namely, the propensity in recent years to break down the traditional boundaries of organizations in these and other ways.)

Matrix Structure

Earlier we discussed stacking up different levels of grouping in an organization the way campfire logs are stacked one way and then another, because a single one cannot provide all the coordination that is necessary. Further to this, a line and staff structure can help, as can the lateral linkages just discussed. But further steps may be necessary to enhance collaboration, especially across the silos. This is when organizations turn to matrix structure. It encourages collaboration by violating one of the sacred principles of management, called unity of command, which means that everyone must report to a single boss. In matrix structure, people report to two or more bosses, which increases ambiguity for the sake of collaboration.

A matrix structure may be permanently built into the superstructure or else set up temporarily to execute a project. As an example of the former, those purchasing engineers discussed earlier can report to both the manager of purchasing or the manager of engineering, to help bridge the gap between the two units. Likewise, a regional sales team can report to both the VP of sales and the VP of the region. The organization lives with the ambiguity in order to sustain a balance of power. As an example of a temporary matrix, consider a team of designers, engineers, and marketers working together on a project to develop a new product. Each may report to both the manager of the project and the manager of their functional home unit.

Such matrix arrangements are more common than you might think, even if not by that name. After all, most of us have been raised in a matrix structure called a family. Ambiguity of authority is a fact of life there, so why not in our organizations? Resolving conflicts through informal negotiations among equals, rather than having to rely on the formal power of a “superior” over “subordinates” sounds pretty grown-up to me!

To summarize, all these lateral linkages can be seen along a continuum, with pure functional structure (grouped by means) at one end and pure market structure (grouped by ends) at the other, while the liaison positions, integrating managers, meetings, teams, task forces, and standing committees fall in between, and matrix structures in the middle.⁴² Add all this up and you can appreciate why in recent years there has been such a proliferation of people with “manager” in their titles.