What to Read Next
Management is fundamentally about making decisions. In organizations, key decisions are often made by groups — boards, senior management teams, finance committees and so on — rather than by single individuals. Thus, an important role of the person leading a decision-making team is to know how to combine multiple opinions — in effect, to become a decision manager. We have found that how a decision is made can significantly affect the outcome of that decision. Hence, the ideas discussed in this article are applicable to any setting in which one has to “decide how to decide.” For the sake of clarity, we illustrate these ideas in the context of boards of directors.
The board of directors is responsible for a company’s most important decisions. In turn, a key responsibility of the chairman is to lead the board so that, collectively, the board can make the best possible decisions for the company. The chairman guides the discussion among the board members and tries to get the most out of their expertise.
Imagine you are the chairman and that, after many hours of discussion about whether or not to acquire a competitor, three members strongly oppose the acquisition, while you, four board members and the CEO support it. You need the board to make a decision now, because if your company doesn’t buy this competitor soon, some other company is likely to do it. How would you combine the opinions of the nine board members to reach the best possible decision?
You have at least three choices.
- You could argue, “This is a make-or-break decision, so we will only pursue it if there’s unanimity.” There isn’t, so you don’t buy the competitor.
- You could argue, “It seems that we have discussed this thoroughly and have a clear majority, so that should be the decision.” With six out of nine board members in favor, you buy.
- You could attempt to delegate the decision to the CEO, arguing, “The CEO and his team have worked for months on this decision; they are the experts. We should not step into their job, so I suggest we trust them.” This, too, would lead to buying the company.
Thus, by prompting a rule on how the decision will be made — by unanimity, majority or delegation — you can significantly influence what will be decided. So which rule should it be? After all, the arguments for all three make sense, and no one really knows how any acquisition will work out.
It turns out that there’s a little-known but very important statistical result that tells you what decision rule is the most likely to lead to the best decision. The goal of this article is to explain how you can benefit from this result.
Two Types of Errors
We know from the field of statistics that there are two types of errors: commission and omission (also known as Type I and II errors). In business, an omission error means missing a good project, while a commission error means pursuing a bad project. For example, a board makes a commission error if it hires a CEO who subsequently performs very poorly. Omission errors, on the other hand, involve missed opportunity. For example, Samsung Electronics Co., Ltd. made an omission error in 2004, when its board decided not to buy Android Inc. Android Inc. was later bought by Google Inc., and the Android operating system became the most popular smartphone operating system, establishing Google as a dominant player in mobile software.
An understanding of commission and omission errors gives the chairman a powerful lever to improve the board’s chances of making the right decision. This is because the probability of making commission and omission errors depends on the consensus level required to approve a decision. This consensus level, in turn, is something the chairman can influence. Thus, he or she can decrease the chance that the company will make costly errors.
A numerical example will make this clear. Imagine a five-member board that has to make acquisition decisions on 10 companies. Assume that only five of the 10 companies will end up being good investments, but the decision makers can’t tell for sure which acquisitions will work out. Assume also that each board member has a 60% probability of choosing correctly.
If the board only approves decisions by unanimity, then the probability of buying each company is about 8% (= 60% × 60% × 60% × 60% × 60%); that is, each company has about a 92% probability of not being bought. Thus, unanimity would produce many omission errors (in the worst case, the five good companies would not be bought) but very few commission errors. If, instead of using full consensus, the board buys a company as long as any one of its members likes it, then the probabilities would be the other way around: Each company would have about a 92% probability of being bought and about an 8% probability of being passed up. This would lead to very few omission errors (as most of the companies would be bought) but many commission errors (as many of the bad companies would be bought).
In short, a requirement of full consensus minimizes commission errors, while a requirement of lowest consensus minimizes omission errors. A similar analysis shows that decision by majority rule produces levels of omission and commission errors that are in between those of unanimity and lowest consensus.
This relationship between consensus level and decision errors has been tested in the real world. Recently, one of us conducted a study of more than 100,000 stock-picking decisions by mutual funds over 30 months. (See “Related Research” for more information about this study, which was published in Strategic Management Journal.) Mutual funds offer an ideal setting to test the relationship between decision rules and commission and omission errors: In this setting, one can observe the decision rule (in other words, some funds use unanimity, and others delegate decisions to individual fund managers) as well as the type of errors made (an omission error corresponds to missing a good investment, while a commission error corresponds to making an investment that ends up performing poorly). Moreover, one can test the relationship thoroughly, as there are thousands of mutual funds, each one typically making dozens of investment decisions per month.
Studying mutual funds confirmed the logic presented here: Funds whose managers used unanimity tended to err by missing good investment opportunities; those whose fund managers could act independently tended to err by acquiring stocks that ended up performing poorly.
The fact that consensus level affects the type of errors made naturally raises the question, “What is the best consensus level?” The traditional belief, particularly prevalent at the board level, is that high consensus is always good. Many groups proudly say, “All our decisions are unanimous.” But the scientific evidence does not support the traditional view.
In reality, since the consensus level changes the probabilities of making commission and omission errors in predictable ways, the best consensus level depends on the relative cost of the errors. (See “Consensus Level and Errors of Commission and Omission.”) For example, a company like Google, Microsoft or Kodak — for which missing the next technology breakthrough could prove fatal — should use a low consensus level, as this minimizes omission errors; a company for which, say, trying a new formulation can expose it to huge liabilities, such as Procter & Gamble, Coca-Cola or Johnson & Johnson, should use a high consensus level, as this minimizes commission errors.
Consensus Level and Errors of Commission and Omission
In a five-member group, requiring higher levels of consensus for a decision decreases the likelihood of commission errors but increases the likelihood of errors of omission.
A company may need to use different consensus levels for different decisions. Typical business decisions that have a higher cost of omission than commission include choosing R&D investments, investing to comply with regulation and investing in whistle-blowing channels.
In turn, decisions with a higher cost of commission than omission include decisions such as hiring a new CEO, approving a large loan or deciding on a merger.
Applying This Knowledge
If the size and composition of the board are fixed, the chairman faces an inescapable trade-off: Decreasing one type of error will tend to increase the other. The chairman must therefore decide which type of error would do the most harm and promote the decision rule that minimizes that type. Of course, the chairman often has no direct say over that; it may be written into the corporate bylaws. But that doesn’t mean he or she has no influence.
To apply the ideas discussed here, a chairman can start by openly discussing with the board the relationship between the level of consensus and the likelihood of commission and omission errors — a relationship the board is probably unaware of. The board could then analyze the decisions at hand, define the relative costs of commission and omission errors, and collectively agree on the best decision rule for each decision. If the chairman foresees resistance to changing the decision rule, he or she can use the following tactics to nudge the decision process toward the optimal.
If commission errors are clearly costlier than omission errors, we have seen that the optimal choice is to use unanimity (full consensus) — in other words, to give veto power to anyone opposing. To nudge the board toward making a decision based on unanimity, the chairman can try the following: (1) encourage the dissenting members to fully express their opinions; (2) suggest creating a committee to study the opposing arguments in more depth; or (3) suggest postponing the decision, which gives the dissenting members enough time to better support their position.
There’s a little-known but very important statistical result that tells you what decision rule is most likely to lead to the best decision.
As protection from omission errors, the chairman may want to nudge the board to delegate some decisions. For example, in an area where omission errors are the primary concern, the chairman could urge the board to delegate a decision to the CEO, even if there is no majority on the board. To help a board avoid omission errors, the chairman can use the following strategies: (1) debate where the line between board and CEO decisions lies and analyze whether this particular decision could be delegated to the CEO; (2) identify concerns before the meeting and give well-prepared supporters the chance to speak first (to anchor the discussion) and for longer; and (3) suggest delegating the decision to a board committee that includes the strongest supporters.
If commission and omission errors seem to be equally costly, the chairman may want to nudge the board toward using majority rule. Steering the board toward this rule should not be difficult, as this is the default decision rule in many situations. To move the board in this direction, the chairman can (1) highlight that time is of the essence and that spending time trying to reach full consensus may exact a large opportunity cost or (2) explore individuals’ positions, then subtly imply that if the issue is voted, there will be winners and losers. This may compel the minority to save face by yielding.
Shifting Gears As Needed
After the corporate scandals of the past decade, much advice has been directed at improving boards’ compliance with regulations and fiduciary duties — all important and necessary issues. However, we believe that a key point has been forgotten: how boards and other groups make decisions and how these decisions can be improved. While this article uses the context of boards of directors to illustrate these ideas, these insights can be used in other decision-making contexts, as well.
Like a car, a decision-making group needs to shift into different gears depending on the situation. Our research shows that these groups can now rely less on instinct and tradition, and more on the science of decision making to match the required consensus level with the desired outcome of avoiding the most costly type of decision error. Although our method cannot guarantee that every particular decision will be optimal, it will improve the group’s overall decision making — which will be of no small benefit.