Do You Have a Well-Designed Organization?

For most companies, organization design is neither a science nor an art; it’s an oxymoron. Organizational structures rarely result from systematic, methodical planning. Rather, they evolve over time, in fits and starts, shaped more by politics than by policies. The haphazard nature of the resulting structures is a source of constant frustration to senior executives. Strategic initiatives stall or go astray because responsibilities are fragmented or unclear. Turf wars torpedo collaboration and knowledge sharing. Promising opportunities die for lack of managerial attention. Overly complex structures, such as matrix organizations, collapse because of lack of clarity about responsibilities. Most executives can sense when their organizations are not working well, but few know how to correct the situation. A comprehensive redesign is just too intimidating. For one thing, it’s immensely complicated, involving an endless stream of trade-offs and variables. For another, it’s divisive, frequently disintegrating into personality conflicts and power plays. So when organization design problems arise, managers often focus on the most glaring flaws and, in the process, make the overall structure even more unwieldy and even less strategic. What’s been lacking is a practical framework to guide executives through the complexities of organization design. That’s what we aim to provide in this article. We have reviewed the principles of good design, studied the structures of dozens of companies, large and small, and observed how executives go about making design decisions. We have encapsulated our findings into nine tests of organization design, which can be used either to evaluate an existing structure or to create a new one. The first four tests are what we call “fit” tests. They provide an initial screen for design alternatives, revealing whether the structures support the company’s strategy, talent pool, and situation. The next five are “good design” tests. They can help a company refine a prospective design by addressing potential problem areas, including the balance between empowerment and control. This set of tests helps you establish the right amount of hierarchy, control, and process—enough for the design to work smoothly but not so much as to dampen initiative, flexibility, and networking. Many of the tests, and their underlying principles, will sound familiar. Their power stems not from their innovativeness—we’re not trying to promote a new theory of business organization—but from their rigor and completeness. Together, they provide a company’s management with a structured approach for analyzing all the key variables of organizational success. Individual design decisions will still be difficult, often requiring subjective judgments and hard trade-offs, but using the framework will help make the debate more rational, shifting it away from issues of personality and toward issues of strategy and effectiveness. Getting the Fit Right 1. The Market Advantage Test.Does your design direct sufficient management attention to your sources of competitive advantage in each market? In formulating a strategy, a company has to ask itself two fundamental questions: Which markets should we compete in, and how will we gain an advantage over competitors in those markets? It may seem obvious that these questions should also drive the company’s organization design, but many structures end up impeding market strategy rather than furthering it. Some distribute responsibilities in ways that distract the management team’s attention from target customers. Others create divisions among units that make it difficult for them to operate in ways that provide the company with a competitive edge. The penalties of such misalignments can be enormous. The first and most fundamental test of a design, therefore, is whether it fits your company’s market strategy. You should begin by defining your target market segments. The definitions will vary depending on which part of your organization is being evaluated. If GE, for example, were designing its overall corporate organization, it would use broad definitions such as “aircraft engines” or “broadcasting.” But if it were looking only at the design of its financial services unit, it would use much narrower definitions, probably combining particular service lines with particular geographic markets: “aircraft leasing in Europe,” for instance, or “receivables financing in Mexico.” There should be no dispute about the relevant market segments; if there is, you need to do some fresh strategy thinking before you proceed with the design effort. Next, determine whether the design directs enough attention to each market segment. Here’s our rule of thumb: If a single unit is dedicated to a single segment, the segment is receiving sufficient attention. If no unit has responsibility for the segment, the design is fatally flawed and needs to be revamped. Often, the analysis is not so clear; a unit may have responsibility for a number of segments. (This is often the case with small, but rapidly growing, market segments.) You will need to evaluate such situations carefully, making judgments about whether the division of responsibilities will allow sufficient attention to be focused on the segment. It’s also important to determine whether the design supports your key sources of advantage (speedy introduction of products, for example, or low-cost manufacturing) and related operating initiatives (product launches, factory automation). List these sources and initiatives, and check how the design addresses them. In a perfect world, you would have a single unit, or department, dedicated to each source and initiative. In reality, however, market advantages often require coordination across units. For instance, your source of advantage in one segment may be superior new-product development. To achieve that advantage, the business unit responsible for the segment may need to collaborate with a central research function. Or your advantage may be an economy of scale in manufacturing that requires coordinated production across numerous business units. Because collaboration across units is always more difficult to manage than collaboration within units, any source of advantage that requires cross-border links—particularly complex ones—should be a cause for concern. You’ll need to be confident that the design will enable the unit managers to give sufficient attention to maintaining the links. Some compromises may remain; they’ll be further analyzed by the good-design tests below. 2. The Parenting Advantage Test.Does your design help the corporate parent add value to the organization? Just as parents play varying roles in families, corporate headquarters play varying roles in different companies. The focus of this test is to make sure the organizational design is tailored to support these roles. First, explicitly define and list your company’s “parenting propositions”—the corporate-level activities that provide real value to the overall company. The propositions might involve narrow tasks—for example, managing government relations—or broad coordination roles, such as maintaining strong research capabilities across all units. Or they might entail specific initiatives, such as implementing a company-wide ERP system. (See the sidebar “How Parents Create Value.”) How Parents Create Value To be effective, parent units need to think through the ways in which they can create value or add value to the rest of the organization. We call these sources of added value “parenting propositions,” and, in general, they fall into five categories: Select Propositions The parent unit creates value by acquiring units or people for less than they are worth or disposing of activities for more than they are worth. Build Propositions The parent unit helps units expand their size and scope of activity by, for example, helping with globalization or product extensions. Stretch Propositions The parent unit helps units improve costs, quality, or profitability by, for instance, setting stretch targets or providing benchmarks. Link Propositions The parent unit helps units work together in ways they would find difficult if left to themselves. For example, it might centralize activities or alter incentives. Leverage Propositions The parent unit finds ways to exploit a central resource, such as a brand, a relationship, a skill, or a patent, in new markets or new businesses. Next, determine whether the design gives sufficient attention to these value-adding tasks and initiatives. If, for example, one of the parent’s key roles is encouraging knowledge sharing among a particular group of units, it’s important to ask whether there is a manager in the parent unit focused on that task. You’ll also need to look hard at the organizational links among those units. If the units are located in different divisions, it may make sense to change the design so that they become members of the same division, making collaboration much easier. Sometimes, this test will highlight difficult trade-offs that need to be made. If one of the parenting propositions is to spur high-speed innovation, for instance, you will need to decide whether it makes more sense to centralize R&D in a corporate unit or disperse it in the business units, which are closer to the market. The parenting advantage test can help companies see more clearly the organizational implications of their strategies, as agriculture giant Cargill recently discovered. One of the most important parenting propositions of Cargill’s headquarters was encouraging a greater focus on broad customer solutions rather than on individual products. When top management viewed the organization in this light, it saw that certain fundamental changes were needed. Cargill created new, more market-focused business units, and it grouped them together into broad “platforms” with management teams that could promote a coordinated approach to customer relationships and solutions. The Food Applications platform, for example, brought together all of Cargill’s businesses that sold products to food manufacturers; the businesses that dealt with farmers became the Farmgate platform. The exercise enabled Cargill to more clearly define its parenting propositions and create an organization that supported them. 3. The People Test.Does your design reflect the strengths, weaknesses, and motivations of your people? When an organization doesn’t work right, executives are often quick to blame “people problems.” But that’s wrongheaded. If an organization is not suited to the skills and attitudes of its members, the problem lies with the design, not the people. For this test, first look at your key players—the members of the top management team and other individuals deemed critical to the business. For each, ask whether the design provides the appropriate responsibilities and reporting relationships and wins their commitment. If, for example, your CEO is a marketing type and the design focuses her attention on performance management, you’ve got a problem. If your CFO is a hands-on, detail-oriented guy and your design has the top finance manager in each business unit reporting to the unit head instead of to him, you’re setting yourself up for big conflicts. Now look at the pivotal jobs in the design—the positions that will need to be staffed by highly talented people if the organization is to work well. Typically, these will include the heads of all key business units and the managers of all functions involved in critical cross-unit relationships. Do you have outstanding people to staff these jobs today? Do you have the career paths and development initiatives needed to create and retain new talent for tomorrow? If you had to find replacements outside, would you be able to attract and hire them? A design that cannot be staffed with competent managers should be abandoned. If you’re creating a new structure, you also need to look at the losers—the employees who will forfeit status or power in the revamped organization. All redesigns create losers, and losers can turn cynical and resistant, becoming roadblocks to change. You need to make two difficult judgments. First, determine which of the losers are influential. Then decide how to deal with them—either buying their support through added compensation or neutralizing their influence by changing their roles or letting them go. 4. The Feasibility Test.Have you taken account of all the constraints that may impede the implementation of your design? All companies have constraints on their ability to act. Some constraints, such as laws, are external. Others, such as information systems, are internal. Because they can impede or even block certain organizational changes, such constraints need to be identified and assessed early in any design effort. Look, in particular, at four categories: Government regulations are rarely a dominant influence on organizational structure, but they can preclude certain design elements. In some countries, for instance, it is impossible to do business without setting up a separate joint venture with a local partner. And in the utilities industry, regulators often insist on keeping regulated and nonregulated business activities in separate units. Companies with international operations need to be particularly cognizant of legal issues. The interests of a company’s stakeholders should be considered carefully. Large shareholders often need to sign off on major organizational moves, and stock markets may impose rules—such as restrictions on minority investors—that have organizational implications. In some companies, other stakeholders, such as trade unions, will demand a voice in decisions. A company’s information systems may prevent certain organizational changes. You may, for example, want to move from a country-based to a product-based structure, but if your systems are unable to report performance by product, you will either have to retool them, at considerable cost, or seek another design option. Corporate cultures can limit the feasibility of design choices. Executives frequently discover this in postmerger integration situations, as was the case at DaimlerChrysler. Try to identify the root causes of cultural constraints. For example, if your company boasts of a strong performance culture but has poor interunit coordination, the cause may be an incentive system that provides no rewards for cooperation. If your redesign relies on cross-unit processes, you’ll have to change the incentive system for it to work. Refining the Design 5. The Specialist Cultures Test.Does your design protect units that need distinct cultures? In most companies, there are certain units that should maintain distinct cultures. They need to think and work in ways that are different from the prevailing organizational norms. Examples might include new-product development teams, e-business groups, or functional service units. In evaluating the strength of an organization design, you need to make sure that such “specialist cultures” are sufficiently insulated from the rest of the organization. Once you’ve identified the specialist cultures in your company, ask yourself whether any of them is in danger of being dominated. Look, in particular, at sister units and the parent unit to which the specialist culture unit reports. If these other units share a culture that’s different from the specialist culture, you can assume that the specialist culture is at risk of being “contaminated” by the dominant culture. Let’s say that a big chemical company has housed a specialty chemicals unit in a division with many bulk chemicals units, and the divisional headquarters is staffed mainly by bulk chemicals executives. That’s a potential problem to which a solution must be found. Dow Chemical, for example, realized that its specialty chemicals units were in danger of being dominated by the bulk chemicals cultures of Dow’s large integrated sites. So when the company reorganized, it grouped all the specialty chemicals units in a new division, separating them from the bulk chemicals units. When you find a specialist culture that is at risk, first look for ways to protect it without changing the basic structure. You might, for example, put a high-ranking corporate executive in charge of the unit, giving it the standing necessary to resist external influences. Or you might grant the unit greater autonomy, freeing it, say, from corporate human-resources policies. Or you might try to solve the problem through communication—educating the rest of the company about the unique goals and requirements of the specialist unit. If you can’t come up with protective measures, you’ll need to alter the design, possibly by rethinking the way you organize your business units. 6. The Difficult-Links Test.Does your design provide coordination solutions for the unit-to-unit links that are likely to be problematic? However you define your units, some collaboration among them is likely to be necessary. (See the sidebar “How Units Connect.”) We have found that the large majority of these links are best handled through self-managed networking among the units. Whenever possible, top management should leave this up to the units rather than impose top-down coordination processes. But this may not be sufficient for links that pose particular challenges. Such “difficult links,” as we call them, arise for many reasons. Managers in different units may not perceive the benefits of collaborating, they may have no incentives to work together, or they may simply lack the skills and resources to make the necessary cooperation happen. (For a fuller discussion, see “Desperately Seeking Synergy,” by Michael Goold and Andrew Campbell, HBR September—October 1998.) Whatever the cause, difficult links call for specially designed solutions, such as a clearly defined arbitration process for resolving disputes. How Units Connect Some of the biggest organizational challenges involve coordination among units. In evaluating a design, you need to pay close heed to unit-to-unit links, which take six basic forms: Shared Know-How Links Sharing best practices, leveraging expertise in functional areas, pooling knowledge about how to succeed in specific geographic regions, or sharing product or market know-how Shared Tangible Resources Links Creating economies of scale and eliminating duplicated effort through the sharing of physical assets (such as an R&D lab) or resources (such as people) Pooled Negotiating Power Links Generating economies of scale or other benefits through common purchases or joint negotiation with stakeholders (such as customers, governments, and universities) Coordinated Strategies Links Aligning the strategies of two or more units by, for example, coordinating responses to a new competitor Vertical Integration Links Coordinating the flow of products or services from one unit to another in order to reduce inventory costs, enhance product development, increase capacity utilization, or improve market access New-Business Creation Links Fashioning new businesses by combining know-how from different units through teams, internal joint ventures, or other alliances One example of a difficult link is collaboration on pricing decisions between product-based units and customer-based units. In Citibank’s Corporate Banking Group, for example, customer-based units, defined around industry groupings, must work together with product specialists in areas such as cash management, foreign exchange, and structured finance. The customer units may want to discount certain products to reinforce customer relationships, while the product units’ motivation is to preserve product-specific margins and profitability. A conflict of this sort can only be resolved by designing detailed coordination solutions. For example, a process for reaching agreement on pricing decisions, specifying who has ultimate authority, can be established. Alternatively, incentives and performance measures can be amended to align the interests of managers in the product and customer units more closely, or a group-level manager can be given authority to arbitrate and resolve disputes. Refinements to the basic structure may not always be sufficient to resolve a difficult link. A typical example is when the link will lead to hard-to-compensate losses for one or more of the units (as in a facilities rationalization program). Substantial redesign, such as bringing the units involved into the same division or setting up a new dedicated unit, may then be needed. IBM, for example, found it necessary to set up a Global Services Division in order to achieve sufficient integration in the provision of services to its customers. Taken together, the specialist cultures test and the difficult-links test can guide top management’s judgments about how narrowly or broadly to define unit responsibilities. They allow the design to strike a suitable balance between specialization and coordination. 7. The Redundant-Hierarchy Test.Does your design have too many parent levels and units? Big companies can have many parent units, including corporate headquarters and various groups that oversee line divisions, corporate functions, and geographic regions. Some of these units may be very small, consisting of just a line manager and a finance executive. Others may be large and complex, encompassing many staff members. In evaluating an organization design, our basic presumption is in favor of decentralizing decisions to frontline units and retaining decisions at upper levels only if those levels can add value. Therefore, it’s important to determine whether each parent level is needed and, if so, whether it has the resources necessary to do its job. First, identify each level and unit in your corporate hierarchy above the operating units. Then ask yourself whether each has clear and distinct parenting propositions (as described in the parenting advantage test). If a level’s propositions echo those of the level above or below it, one of the levels may be redundant, and you should think hard about removing it. To cover the inevitable costs and drawbacks of an extra layer, we believe it needs to be able to improve the performance of the units reporting to it by at least 10 % . That rule of thumb can make it easier to determine whether levels are worth keeping. It’s a powerful argument against having many layers: If there are, say, three layers above the business units, the total parenting added value needs to be at least 30 % to justify the layers! You now need to determine whether every level with compelling parenting propositions has access to the skills and resources it needs. Let’s say, for example, that you have a division responsible for all businesses in Europe, and its parenting proposition is to coordinate manufacturing and customer service and integrate back-office functions across countries. To succeed, the division will require deep functional expertise in manufacturing, major-account management, and information systems. If it lacks those skills, you will need to develop them or alter some roles and responsibilities. For instance, the coordination of information systems may need to be pushed down to the country-level business units, which could together organize a pan-European council of systems managers. This test does not require exhaustive analysis or fine-grained judgments. The object is to spot major problems, not minor ones. But even when conducted at a high level, the redundant-hierarchy test can provide powerful insights. Many companies that have used it have found ways to cut out layers of management, shrink corporate and divisional functions, and refocus corporate and divisional managers so that they add greater value to the company. 8. The Accountability Test.Does your design support effective controls? In decentralized organizations, accountability for performance is important. The purpose of this test is to ensure that every unit has appropriate controls over its performance—controls that suit its responsibilities, are economical to implement, and motivate managers. In assessing accountability, focus particularly on two common sources of problems. First, look at any units with shared responsibilities, especially if their collaboration is mandatory. A company with both global business units and national operating companies, for example, may require global product heads to reach consensus with operating company executives on issues such as major investments and profit targets. Or business units may be required to use the services provided by corporate IT or HR departments. Whenever shared responsibilities are imposed by the corporate parent in this way, it dampens accountability. It becomes easy for units to blame performance problems on one another and difficult for senior executives to determine who’s at fault. Second, focus on any units whose performance is difficult to measure, either because there are no objective outcomes for comparison or because of the expense of collecting performance data. A corporate unit doing basic research is a typical example of a unit for which it’s hard, if not impossible, to come up with clear-cut, easy-to-collect performance measures. Where possible, remedies should be put in place for units with blurred responsibilities or fuzzy measures. Clearer performance measures should be developed, and greater clarity should be provided for overlapping responsibilities. Often, however, full solutions are not possible. In such cases, parent managers will have no choice but to rely on subjective judgments about performance. This usually makes the control process unsatisfactory unless the parent managers have a deep operating knowledge of the units they oversee. If such managers are unavailable, you may need to modify the design. 9. The Flexibility Test.Does your design facilitate the development of new strategies and provide the flexibility required to adapt to change? A well-designed organization is flexible for the future as well as fit for the present. It provides ways for a company to pursue innovation and allows for adaptability to changing circumstances. Of course, it takes more than a good organization design to ensure innovation and flexibility—it also requires flexible minds, deep talent, and robust resources. This test is therefore aimed not at determining whether the company has all the capabilities required to innovate and adapt, but whether there will be any major organizational roadblocks along the path to the future. Start by assembling a group of managers from across the company who have deep knowledge of products and markets. Ask them to create a list of ten or so future opportunities, including a couple of off-the-wall ones, that are not anticipated in the current strategy but that are representative of the opportunities the company is likely to encounter in the future. Now look at your design and see whether it would support or impede the pursuit of each opportunity. This is, of course, a tough task. It’s not easy to predict how groups of people will act in the face of hypothetical challenges. But it is possible to get a rough sense of whether the organization will be supportive, neutral, or obstructive and to think of changes that will reduce any obstructions. For example, in the mid-1990s, Emap, the magazine publisher, listed new media as one of its future opportunities. Recognizing that its magazine-focused business units would be reluctant to cannibalize their existing franchises, top management created business development functions at the division level to ensure that these opportunities were not overlooked. If you find that your design could obstruct the pursuit of opportunities, explore possible modifications to it. You might, for instance, set aside seed capital for new products or revamp your strategic planning process to force managers to spend time thinking about potential new businesses. Keep in mind, however, that the opportunities are speculative. You don’t want to make changes that will render your organization less able to fulfill its current business imperatives. Follow a similar approach in examining flexibility. First, identify five or ten major organizational changes that may be required over the next three to five years. Then, identify any parts of the organization that would be resistant to the changes. Pockets of resistance can often be found in the top management layer, with its rigid personal loyalties and entrenched fiefdoms. It can also occur because of tightly integrated sets of units, with complex relationships and policies that are difficult to untangle. If push came to shove, would you be able to mitigate these obstacles—by, for example, redistributing top managers’ responsibilities or moving to more independent and self-contained business units? If you find that the risk of inertia is too high, you’ll need to think about altering the design. An Iterative Process Once you’ve gone through all the tests, you’ll probably have made a number of minor, and perhaps some major, changes to your design. (For a summary of the problem-solving process, see the sidebar “Getting the Bugs Out.”) You’ll want to run through the tests again to ensure that the changes made to pass one test haven’t caused the design to fail any of the others. Let’s say, for example, that a company finds a difficult link between its business units and one of its shared corporate functions. To fix the problem, it decides that the function should be run as an informal joint venture operated by all the business units. But while going back through the tests, the company realizes that the business units share a single culture that’s very different from the one in the corporate function. As a result, the design now fails the specialist cultures test. The company needs to find another way to solve the difficult-link problem. Getting the Bugs Out When you identify a problem with your design, first look for ways to fix it without substantially altering it. If that doesn’t work, you’ll have to make fundamental changes or even reject the design. Here’s a step-by-step process for resolving problems: Steps not involving major design change Modify without changing the units. Refine the allocation of responsibilities (for example, clarify powers and responsibilities). Refine reporting relationships and processes (for example, define new parenting propositions). Refine lateral relationships and processes (for example, define coordination mechanisms). Refine accountabilities (for example, define more appropriate performance measures). Redefine skill requirements and incentives. Modify criteria for selecting people. Redefine skill development needs. Develop incentives. Shape informal context. Clarify the leadership style needed. Define norms of behavior, values, or social context. Steps involving major design change Make substantial changes in the units. Make major adjustments to unit boundaries. Change unit roles (for example, turn functional units into business units or shared services). Introduce new units or merge units. Change the structure. Change reporting lines. Create new divisions. The iterative nature of the tests is one of their great strengths. Organizational decisions are inevitably complex, and tweaking one area may produce unanticipated consequences in a very different area. To get the best design, you need to take the broad view, working step-by-step through the myriad tradeoffs. It’s not an easy process, but it is a manageable one. And the alternative—waiting for design flaws to turn into disasters—is far worse.

Do You Have a Well-Designed Organization?

Partner Center