Decision making is arguably the most important job of the senior executive and one of the easiest to get wrong. It doesn't have to be that way -- if you look at the process in a whole new light.

LEADERS SHOW THEIR METTLE IN MANY WAYS -- setting strategy and motivating people, just to mention two -- but above all else leaders are made or broken by the quality of their decision. That's a given, right? If you answered yes, then you would probably be surprised by how many executives approach decision making in a way that neither puts enough options on the table nor permits sufficient evaluation to ensure that they can make the best choice. Indeed, our research over the past several years strongly suggests that, simply put, most leaders get decision making all wrong.

The reason: Most businesspeople treat decision making as an event -- a discrete choice that takes place at a single point in time, whether they're sitting at a desk, moderating a meeting, or staring at a spreadsheet. This classic view of decision making has a pronouncement popping out of a leader's head, based on experience, gut, research, or all three. Say the matter at hand is whether to pull a product with weak sales off the market. An "event" leader would mull in solitude, ask for advice, read reports, mull some more, then say yea or nay and send the organization off to make it happen. But to look at decision making that way is to overlook larger social and organizational contexts, which ultimately determine the success of any decision.

The fact is, decision making is not an event. It's a process, one that unfolds over weeks, months, or even years; one that's fraught with power plays and politics and is replete with personal nuances and institutional history; one that's rife with discussion and debate; and one that requires support at all levels of the organization when it comes time for execution. Our research shows that the difference between leaders who make good decisions and those who make bad ones is striking. The former recognize that all decisions are processes, and they explicitly design and manage them as such. The latter persevere in the fantasy that decisions are events they alone control.

In the following pages, we'll explore how leaders can design and manage a sound, effective decision-making process -- an approach we call inquiry -- and outline a set of criteria for assessing the quality of the decision-making process. First, a look at the process itself.

Not all decision-making processes are equally effective, particularly in the degree to which they allow a group to identify and consider a wide range of ideas. In our research, we've seen two broad approaches. Inquiry, which we prefer, is a very open process designed to generate multiple alternatives, foster the exchange of ideas, and produce a well-tested solution. Unfortunately, this approach doesn't come easily or naturally to most people. Instead, groups charged with making a decision tend to default to the second mode, one we call advocacy. The two look deceptively similar on the surface: groups of people, immersed in discussion and debate, trying to select a course of action by drawing on what they believe is the best available evidence. But despite their similarities, inquiry and advocacy produce dramatically different results.

When a group takes an advocacy perspective, participants approach decision making as a contest, although they don't necessarily compete openly or even consciously. Well-defined groups with special interests -- dueling divisions in search of budget increases, for example --advocate for particular positions. Participants are passionate about their preferred solutions and therefore stand firm in the face of disagreement. That level of passion makes it nearly impossible to remain objective, limiting people's ability to pay attention to opposing arguments. Advocates often present information selectively, buttressing their arguments while withholding relevant conflicting data. Their goal, after all, is to make a compelling case, not to convey an evenhanded or balanced view. Two different plant managers pushing their own improvement programs, for example, may be wary of reporting potential weak points for fear that full disclosure will jeopardize their chances of winning the debate and gaining access to needed resources.

What's more, the disagreements that arise are frequently fractious and even antagonistic. Personalities and egos come into play, and differences are normally resolved through battles of wills and behind-the-scenes maneuvering. The implicit assumption is that a superior solution will emerge from a test of strength among competing positions. But in fact this approach typically suppresses innovation and encourages participants to go along with the dominant view to avoid further conflict.

By contrast, an inquiry-focused group carefully considers a variety of options and works together to discover the best solution. While people naturally continue to have their own interests, the goal is not to persuade the group to adopt a given point of view but instead to come to agreement on the best course of action. People share information widely, preferably in raw form, to allow participants to draw their own conclusions. Rather than suppressing dissension, an inquiry process encourages critical thinking. All participants feel comfortable raising alternative solutions and asking hard questions about the possibilities already on the table.

People engaged in an inquiry process rigorously question proposals and the assumptions they rest on, so conflict may be intense -- but it is seldom personal. In fact, because disagreements revolve around ideas and interpretations rather than entrenched positions, conflict is generally healthy, and team members resolve their differences by applying rules of reason. The implicit assumption is that a consummate solution will emerge from a test of strength among competing ideas rather than dueling positions. Recent accounts of GE's succession process describe board members pursuing just such an open-minded approach. All members met repeatedly with the major candidates and gathered regularly to review their strengths and weaknesses -- frequently without Jack Welch in attendance -- with little or no attempt to lobby early for a particular choice.

A process characterized by inquiry rather than advocacy tends to produce decisions of higher quality -- decisions that not only advance the company's objectives but also are reached in a timely manner and can be implemented effectively. Therefore, we believe that leaders seeking to improve their organizations' decision-making capabilities need to begin with a single goal: moving as quickly as possible from a process of advocacy to one of inquiry. That requires careful attention to three critical factors, the "three C's" of effective decision making: conflict, consideration, and closure. Each entails a delicate balancing act.

Critical thinking and rigorous debate invariably lead to conflict. The good news is that conflict bring issues into focus, allowing leaders to make more informed choices. The bad news is that the wrong kind of conflict can derail the decision-making process altogether.

Indeed, conflict comes in two forms -- cognitive and affective. Cognitive, or substantive, conflict relates to the work at hand. It involves disagreements over ideas and assumptions and differing views on the best way to proceed. Not only is such conflict healthy, it's crucial to effective inquiry. When people express differences openly and challenge underlying assumptions, they can flag real weaknesses and introduce new ideas. Affective, or interpersonal, conflict is emotional. It involves personal friction, rivalries, and clashing personalities, and it tends to diminish people's willingness to cooperate during implementation, rendering the decision-making process less effective. Not surprisingly, it is a common feature of advocacy processes.

On examination, the two are easy to distinguish. When a team member recalls "tough debates about the strategic, financial, and operating merits of the three acquisition candidates," she is referring to cognitive conflict. When a team member comments on "heated arguments that degenerated into personal attacks," he means affective conflict. But in practice the two types of conflict are surprisingly hard to separate. People tend to take any criticism personally and react defensively. The atmosphere quickly becomes charged, and even if a high-quality decision emerges, the emotional fallout tends to linger, making it hard for team members to work together during implementation.

The challenge for leaders is to increase cognitive conflict while keeping affective conflict low -- no mean feat. One technique is to establish norms that make vigorous debate the rule rather than the exception. Chuck Knight, for 27 years the CEO of Emerson Electric, accomplished this by relentlessly grilling managers during planning reviews, no matter what he actually thought of the proposal on the table, asking tough, combative questions and expecting well-framed responses. The process -- which Knight called the "logic of illogic" because of his willingness to test even well-crafted arguments by raising unexpected, and occasionally fanciful, concerns -- was undoubtedly intimidating. But during his tenure it produced a steady stream of smart investment decisions and an unbroken string of quarterly increases in net income.

Bob Galvin, when he was CEO of Motorola in the 1980s, took a slightly different approach. He habitually asked unexpected hypothetical questions that stimulated creative thinking. Subsequently, as chairman of the board of overseers for the Malcolm Baldrige National Quality Program, Galvin took his colleagues by surprise when, in response to pressure form constituents to broaden the criteria for the award, he proposed narrowing them instead. In the end, the board did in fact broaden the criteria, but his seemingly out-of-the-blue suggestion sparked a creative and highly productive debate.

Another technique is to structure the conversation so that the process, by its very nature, fosters debate. This can be done by dividing people into groups with different, and often competing, responsibilities. For example, one group may be asked to develop a proposal while the other generates alternative recommendations. Then the groups would exchange proposals and discuss the various options. Such techniques virtually guarantee high levels of cognitive conflict. (The exhibit "Structuring the Debate" outlines two approaches for using different groups to stimulate creative thinking.)

But even if you've structured the process with an eye toward encouraging cognitive conflict, there's always a risk that it will become personal. Beyond cooling the debate with "time-outs," skilled leaders use a number of creative techniques to elevate cognitive debate while minimizing affective conflict.

First, adroit leaders pay careful attention to the way issues are framed, as well as to the language used during discussions. They preface contradictory remarks or questions with phrases that remove some of the personal sting ("Your arguments make good sense, but let me play devil's advocate for a moment"). They also set ground rules about language, insisting that team members avoid words and behavior that trigger defensiveness. For instance, in the U.S. Army's after-action reviews, conducted immediately after missions to identify mistakes so they can be avoided next time, facilitators make a point of saying, "We don't use the `b' word, and we don't use the `f' word. We don't place blame, and we don't find fault."

Second, leaders can help people step back from their preestablished positions by breaking up natural coalitions and assigning people to tasks on some basis other than traditional loyalties. At a leading aerospace company, one business unit president had to deal with two powerful coalitions within his organization during a critical decision about entering into a strategic alliance. When he set up two groups to consider alternative alliance partners, he interspersed the groups with members of each coalition, forcing people with different interests to work with one another. He then asked both groups to evaluate the same wide range of options using different criteria (such as technological capability, manufacturing prowess, or project management skills). The two groups then shared their evaluations and worked together to select the best partner. Because nobody had complete information, they were forced to listen closely to one another.

Third, leaders can shift individuals out of well-grooved patterns, where vested interests are highest. They can, for example, ask team members to research and argue for a position they did not endorse during initial discussions. Similarly, they can assign team members to play functional or managerial roles different from their own, such as asking a lower-level employee to assume the CEO's strategic perspective.

Finally, leaders can ask participants locked in debate to revisit key facts and assumptions and gather more information. Often, people become so focused on the differences between opposing positions that they reach a stalemate. Emotional conflict soon follows. Asking people to examine underlying presumptions can defuse the tension and set the team back on track. For instance, at Enron, when people disagree strongly about whether or not to apply their trading skills to a new commodity or market, senior executives quickly refocus the discussion on characteristics of industry structure and assumptions about market size and customer preferences. People quickly recognize areas of agreement, discover precisely how and why they disagree, and then focus their debate on specific issues.

Once a decision's been made and the alternatives dismissed, some people will have to surrender the solution they preferred. At times, those who are overruled resist the outcome; at other times, they display grudging acceptance. What accounts for the difference? The critical factor appears to be the perception of fairness -- what scholars call "procedural justice." The reality is that the leader will make the ultimate decision, but the people participating in the process must believe that their views were considered and that they had a genuine opportunity to influence the final decision. Researchers have found that if participants believe the process was fair, they are far more willing to commit themselves to the resulting decision even if their views did not prevail. (For a detailed discussion of this phenomenon, see W. Chan Kim and Renee Mauborgne, "Fair Process: Managing in the Knowledge Economy," HBR July-August 1997).

Many managers equate fairness with voice -- with giving everyone a chance to express his or her own views. They doggedly work their way around the table, getting everyone's input. However, voice is not nearly as important as consideration -- people's belief that the leader actively listened to them during the discussions and weighed their views carefully before reaching a decision. In his 1999 book, Only the Paranoid Survive, Intel's chairman Andy Grove describes how he explains the distinction to his middle managers: "Your criterion for involvement should be that you're heard and understood.... All sides cannot prevail in the debate, but all opinions have value in shaping the right answers"

In fact, voice without consideration is often damaging; it leads to resentment and frustration rather than to acceptance. When the time comes to implement the decision, people are likely to drag their feet if they sense that the decision-making process had been a sham -- an exercise in going through the motions designed to validate the leader's preferred solution. This appears to have been true of the Daimler-Chrysler merger. Daimler CEO Jurgen Schrempp asked for extensive analysis and assessment of potential merger candidates but had long before settled on Chrysler as his choice. In fact, when consultants told him that his strategy was unlikely to create shareholder value, he dismissed the data and went ahead with his plans. Schrempp may have solicited views from many parties, but he clearly failed to give them much weight.

Leaders can demonstrate consideration throughout the decision-making process. At the outset, they need to convey openness to new ideas and a willingness to accept views that differ from their own. In particular, they must avoid suggesting that their minds are already made up. They should avoid disclosing their personal preferences early in the process, or they should clearly state that any initial opinions are provisional and subject to change. Or they can absent themselves from early deliberations.

During the discussions, leaders must take care to show that they are listening actively and attentively. How? By asking questions, probing for deeper explanations, echoing comments, making eye contact, and showing patience when participants explain their positions. Taking notes is an especially powerful signal, since it suggests that the leader is making a real effort to capture, understand, and evaluate people's thoughts.

And after they make the final choice, leaders should explain their logic. They must describe the rationale for their decision, detailing the criteria they used to select a course of action. Perhaps more important, they need to convey how each participant's arguments affected the final decision or explain clearly why they chose to differ with those views.

Knowing when to end deliberations is tricky; all too often decision-making bodies rush to a conclusion or else dither endlessly and decide too late. Deciding too early is as damaging as deciding too late, and both problems can usually be traced to unchecked advocacy.

Deciding Too Early. Sometimes people's desire to be considered team players overrides their willingness to engage in critical thinking and thoughtful analysis, so the group readily accepts the first remotely plausible option. Popularly known as "groupthink," this mind-set is prevalent in the presence of strong advocates, especially in new teams, whose members are still learning the rules and may be less willing to stand out as dissenters.

The danger of groupthink is not only that it suppresses the full range of options but also that unstated objections will come to the surface at some critical moment -- usually at a time when aligned, cooperative action is essential to implementation. The leader of a large division of a fast-growing retailer learned this the hard way. He liked to work with a small subset of his senior team to generate options, evaluate the alternatives, and develop a plan of action, and then bring the proposal back to the full team for validation. At that point, his managers would feel they had been presented with a fait accompli and so would be reluctant to raise their concerns. As one of them put it: "Because the meeting is the wrong place to object, we don't walk out of the room as a unified group." Instead, they would reopen the debate during implementation, delaying important initiatives by many months.

As their first line of defense against groupthink, leaders need to learn to recognize latent discontent, paying special attention to body language: furrowed brows, crossed arms, or curled-up defiance. To bring disaffected people back into the discussion, it may be best to call for a break, approach dissenters one by one, encourage them to speak up, and then reconvene. GM's Alfred Sloan was famous for this approach, which he would introduce with the following speech: "I take it we are all in complete agreement on the decision here. Then I propose we postpone further discussion of the matter until our next meeting to give ourselves time to develop disagreement and perhaps gain some understanding of what the decision is all about."

Another way to avoid early closure is to cultivate minority views either through norms or through explicit rules. Minority views broaden and deepen debate; they stretch a group's thinking, even though they are seldom adopted intact. It is for this reason that Andy Grove routinely seeks input from "helpful Cassandras," people who are known for raising hard questions and offering fresh perspectives about the dangers of proposed policies.

Deciding Too Late. Here, too, unchecked advocacy is frequently the source of the problem, and in these instances it takes two main forms. At times, a team hits gridlock: Warring factions refuse to yield, restating their positions over and over again. Without a mechanism for breaking the deadlock, discussions become an endless loop. At other times, people bend over backward to ensure evenhanded participation. Striving for fairness, team members insist on hearing every view and resolving every question before reaching a conclusion. This demand for certainty -- complete arguments backed by unassailable data -- is its own peculiar form of advocacy. Once again, the result is usually an endless loop, replaying the same alternatives, objections, and requests for further information. Any member of the group can unilaterally derail the discussion by voicing doubts. Meanwhile, competitive pressures may be demanding an immediate response, or participants may have tuned out long ago, as the same arguments are repeated ad nauseam.

At this point, it's the leader's job to "call the question." Jamie Houghton, the longtime CEO of Corning, invented a vivid metaphor to describe this role. He spoke of wearing two hats when working with his senior team: He figuratively put on his cowboy hat when he wanted to debate with members as an equal, and he donned a bowler when, as CEO, he called the question and announced a decision. The former role allowed for challenges and continued discussion; the latter signaled an end to the debate.

The message here is that leaders -- and their teams -- need to become more comfortable with ambiguity and be willing to make speedy decisions in the absence of complete, unequivocal data or support. As Dean Stanley Teele of Harvard Business School was fond of telling students: "The art of management is the art of making meaningful generalizations out of inadequate fact."

Unfortunately, superior decision making is distressingly difficult to assess in real time. Successful outcomes -- decisions of high quality, made in a timely manner and implemented effectively -- can be evaluated only after the fact. But by the time the results are in, it's normally too late to take corrective action. Is there any way to find out earlier whether you're on the right track?

There is indeed. The trick, we believe, is to periodically assess the decision-making process, even as it is under way. Scholars now have considerable evidence showing that a small set of process traits is closely linked with superior outcomes. While they are no guarantee of success, their combined presence sharply improves the odds that you'll make a good decision.

Multiple Alternatives. When groups consider many alternatives, they engage in more thoughtful analysis and usually avoid settling too quickly on the easy, obvious answer. This is one reason techniques like point-counterpoint, which requires groups to generate at least two alternatives, are so often associated with superior decision making. Usually, keeping track of the number of options being considered will tell if this test has been met. But take care not to double count. Go-no-go choices involve only one option and don't qualify as two alternatives.

Assumption Testing. "Facts" come in two varieties: those that have been carefully tested and those that have been merely asserted or assumed. Effective decision-making groups do not confuse the two. They periodically step back from their arguments and try to confirm their assumptions by examining them critically. If they find that some still lack hard evidence, they may elect to proceed, but they will at least know they're venturing into uncertain territory. Alternatively, the group may designate "intellectual watchdogs" who are assigned the task of scrutinizing the process for unchecked assumptions and challenging them on the spot.

Well-Defined Criteria. Without crisp, clear goals, it's easy to fall into the trap of comparing apples with oranges. Competing arguments become difficult to judge, since advocates will suggest using those measures (net income, return on capital, market presence, share of mind, and so on) that favor their preferred alternative. Fuzzy thinking and long delays are the likely result.

To avoid the problem, the team should specify goals up front and revisit them repeatedly during the decision-making process. These goals can be complex and multi-faceted, quantitative and qualitative, but whatever form they take, they must remain at the fore. Studies of merger decisions have found that as the process reaches its final stages and managers feel the pressure of deadlines and the rush to close, they often compromise or adjust the criteria they originally created for judging the appropriateness of the deal.

Dissent and Debate. David Hume, the great Scottish philosopher, argued persuasively for the merits of debate when he observed that the "truth springs from arguments amongst friends." There are two ways to measure the health of a debate: the kinds of questions being asked and the level of listening.

Some questions open up discussion; others narrow it and end deliberation. Contrarian hypothetical questions usually trigger healthy debate. A manager who worked for former American Express CEO Harvey Golub points to a time when the company was committed to lowering credit card fees, and Golub unexpectedly proposed raising fees instead. "I don't think he mean it seriously," says the manager. "But he certainly taught us how to think about fees.

The level of listening is an equally important indicator of a healthy decision-making process. Poor listening produces flawed analysis as well as personal friction. If participants routinely interrupt one another or pile on rebuttals before digesting the preceding comment, affective conflict is likely to materialize. Civilized discussions quickly become impossible, for collegiality and group harmony usually disappear in the absence of activate listening.

Perceived Fairness. A real-time measure of perceived fairness is the level of participation that's maintained after a key midpoint or milestone has been reached. Often, a drop in participation is an early warning of problems with implementation since some members of the group are already showing their displeasure by voting with their feet.

In fact, keeping people involved in the process is, in the end, perhaps the most crucial factor in making a decision -- and making its stick. It's a job that lies at the heart of leadership and one that uniquely combines the leader's numerous talents. It requires the fortitude to promote conflict while accepting ambiguity, the wisdom to know when to bring conversations to a close, the patience to help others understand the reasoning behind your choice, and, not least, a genius for balance -- the ability to embrace both the divergence that may characterize early discussions and the unity needed for effective implementation. Cyrus the Great, the founder of the Persian Empire and a renowned military leader, understood the true hallmark of leadership in the sixth century BC, when he attributed his success to "diversity in counsel, unity in command."

Legend for Chart:

A - Advocacy
B - Inquiry

               A
               B

Concept of decision making

               a contest
               collaborative problem solving

Purpose of discussion
               persuasion and lobbying
               testing and evaluation

Participant's role

               spokespeople
               critical thinkers

Patterns of behavior

               strive to persuade others
               present balanced arguments

               defend your position
               remain open to alternatives

               downplay weaknesses
               accept constructive criticism

Minority views

               discouraged or dismissed
               cultivated and valued

Outcome

               winners and losers
               collective ownership

By breaking a decision-making body into two subgroups, leaders can often create an environment in which people feel more comfortable engaging in debate. Scholars recommend two techniques in particular, which we call the "point-counterpoint" and "intellectual watchdog" approaches. The first three steps are the same for both techniques:

Legend for Chart:

A - Point-Counterpoint
B - Intellectual Watchdog

A
               B

The team divides into two subgroups.

               The team divides into two subgroups.

Subgroup A develops a proposal,
fleshing out the recommendation,
the key assumptions, and the
critical supporting data.
               Subgroup A develops a proposal, fleshing out the
               recommendation,  the key assumptions, and the
               critical supporting data.

Subgroup A presents the proposal
to Subgroup B in written and oral
forms.

                Subgroup A presents the proposal  to Subgroup B
                in written and oral forms.

Subgroup B generates one or more
alternative plans of action.

                Subgroup B develops a detailed critique of these
                assumptions and recommendations. It presents
                this critique in written and oral forms.
                Subgroup A revises its proposal based on this
                feedback.

The subgroups come together to
debate the proposals and seek
agreement on a common set of
assumptions.

                The subgroups continue in this
                revision-critique-revision cycle until they
                converge on a common set of assumptions.

Based on those assumptions, the
subgroups continue to debate
various options and strive to
agree on a common set of
recommendations.

                Then, the subgroups work together to develop a
                common set of recommendations.

~~~~~~~~

By David A. Garvin and Michael A. Roberto

David A. Garvin is the Robert and Jane Cizik Professor of Business Administration at Harvard Business School in Boston. His most recent HBR article is "Leveraging Processes for Strategic Advantage" (September-October 1995)

Michael A. Roberto is an assistant professor at Harvard Business School.