When executives contemplate strategic decisions, they often succumb to the same cognitive biases we all have as human beings, such as overconfidence, the confirmation bias, or excessive risk avoidance.¹ Such biases distort the way we collect and process information. Even in the rarefied context of the executive suite, judgment can be colored by self-interest leading to more or less conscious deceptions—for example, around the assumptions critical to the valuation of potential capital projects, M&A targets, divestitures, or joint ventures.

CFOs are often the most disinterested parties to such decisions. They seldom chair the relevant meetings, are often highly critical of decision-making dynamics and biases, and can cite examples of past successes and failures. With the technical support of the finance staff, they can also provide hard data to counter the inherent biases of other executives. Yet only a minority of CFOs are fully leveraging their position to change the dynamics of decision making—to promote institutional learning in the interest of better strategic decisions.

To figure out why that might be so—and to look for techniques CFOs can use when playing this critical role—McKinsey’s Bill Huyett and Tim Koller recently talked with Olivier Sibony, a director in McKinsey’s Paris office and a coauthor of numerous articles on the subject of cognitive biases in business decision making.

The Quarterly: Why aren’t CFOs better at using their position to improve the quality of decision making?

Olivier Sibony: CFOs often struggle with a confusion of roles. They’re expected to be both the impartial challenger and an important player in getting things done. They advise the CEO on M&A, but they also drive the discussions with the targets. They have to make sure that the company has the right financing structure, and they’re also supposed to negotiate with the banks. Resolving that tension between roles is where the CFO can do a better job.

The way to do that, I would argue, is for the CFO to view herself not only as the impartial, cool-headed adviser of the CEO, nor just as the executor of the mechanics of a decision, but primarily as the owner of a safe and sound decision-making process—which is a role that no one else plays. And if there is one thing that we take away from the study of behavioral economics, it is that this role is vital. You need to have better processes to make decisions, because people can’t make better decisions alone, but good processes can help if they build on the insights and judgment of multiple people. I’m not saying the CFO is the only person who can build such a process, but she’s in a uniquely good position to build one.

The Quarterly: Why does process matter so much?

Olivier Sibony: Process matters in decision making because we can’t learn from our mistakes the way we think we can. Cognitive biases are everywhere, we all have them, and we pretty much know what they are. We know we’re overconfident, we know we’re susceptible to anchoring, we know we underresearch things that disprove our hypotheses and overresearch things that confirm them, and so on. But these biases are hardwired, and there’s not much we can do about them as individuals. So we can will ourselves to not be overconfident until we’re blue in the face; we’ll still be overconfident.

You can test this yourself. Ask a group of people if they think they are above-average drivers. In the United States, nine out of ten will tell you they’re in the top 50 percent. Now, they all laugh when they get that feedback, but you ask them to do it again and you get the same results. They all think that it’s all those guys around them who are overestimating themselves.

It’s the same in business. We may agree with the proposition that businesspeople in general are overconfident. We may even accept that we’ve been overconfident ourselves in our past decisions, but we always think that this time will be different. Here I’m using the example of overconfidence because it’s easy to demonstrate, but the same is true of other biases. Biases are very deeply ingrained and impervious to feedback.

The Quarterly: So you depend on a multiperson process to control bias?

Olivier Sibony: Exactly. You build a multiperson process where your biases are going to be challenged by somebody else’s perspective. And as CFO, if you manage this process, your goal is to ensure that the biases of individuals weigh less in the final decision than the things that should weigh more—like facts. In other words, you can’t improve your own decision making in a systematic way, but you can do a lot to improve your organization’s decision making through a good process, and that’s what CFOs are uniquely well placed to do.

The Quarterly: It sounds like you’re drawing a contrast between the processes of human interaction and decision making and the more obvious technical systems that the CFO runs—for example, around valuation procedures and merger-management procedures.

Olivier Sibony: There is a contrast and there is also a synergy. The contrast is that CFOs already rely on processes to manage, as you point out, the technical systems. But it’s very easy for people to subvert technical systems to get the answer they want. The typical example of this in M&A is when deal advocates work backward from the price demanded to determine how much in synergies the deal would require to make sense.

What people spend a lot less time thinking about are the interpersonal interactions—the processes of debate—that ensure high-quality decision making. And that is where the synergy lies for CFOs: if you already own the technical processes, you can build on them to improve the quality of debate, for instance by adjusting the agenda, attendees, and protocols of key decision meetings.

The Quarterly: What are some examples of process changes that companies can use?

Olivier Sibony: Let me start with an analogy. Imagine walking into a courtroom where the trial consists of a prosecutor presenting PowerPoint slides. In 20 pretty compelling charts, he demonstrates why the defendant is guilty. The judge then challenges some of the facts of the presentation, but the prosecutor has a good answer to every objection. So the judge decides, and the accused man is sentenced.

That wouldn’t be due process, right? So if you would find this process shocking in a courtroom, why is it acceptable when you make an investment decision? Now of course, this is an oversimplification, but this process is essentially the one most companies follow to make a decision. They have a team arguing only one side of the case. The team has a choice of what points it wants to make and what way it wants to make them. And it falls to the final decision maker to be both the challenger and the ultimate judge. Building a good decision-making process is largely ensuring that these flaws don’t happen.

The Quarterly: How do you build a process that has these features?

Olivier Sibony: My coauthor, Dan Lovallo, and I did some quantitative research on this.² We asked executives to tell us about their investment decisions—which ones worked and which ones didn’t and what practices made the difference—and we reviewed over a thousand of them.

One of the practices that we found made the most difference was having explicit discussions of the irreducible uncertainties in the decision. Notice the difference between that kind of conversation and the one elicited by the typical slide in a PowerPoint presentation, with the title “Risks we identified and risk-mitigating actions we will take.” That’s the way you frame it if you want to look like a confident presenter and want the meeting to go smoothly: you suppress the discussion of uncertainties. Instead, you should be emphasizing them to make sure you have a debate about them.

Executives reported some other things making a big difference—for example, whether the discussion included points of view contradictory to those of the person making the final decision. In other words, did anyone voice a point of view that was contrary to what the CEO wanted to hear or to what they thought he wanted to hear? And did the due-diligence team actually seek out information that would contradict the investment hypothesis, as opposed to simply building a case for it? These types of things can be hardwired into the process to make sure that they happen, and some companies do this routinely.

The Quarterly: Let’s talk about specific techniques. Take M&A as an example—does it help to assign people ahead of time to argue either side of a decision, regardless of what they actually believe?

Olivier Sibony: When evaluating an acquisition, there is of course the issue of impartiality—as Warren Buffett said, relying on one investment bank to tell you if you should do a deal is like asking your barber if you need a haircut. And there is the more subtle issue of motivated error: even people who sincerely believe that their assessments are objective are in fact often biased in the direction of their own interests.

So in this case, it can help in some settings to field two deal teams, at least at some stage in the process: one to argue for the deal and a second to argue against it. In other settings, if companies find that people avoid the direct confrontation that two deal teams imply, managers might prefer to ask the same people to argue both sides of the case or to make the uncertainties explicit. There are many different techniques to foster debate.

The Quarterly: What other techniques come to mind as effective in M&A situations?

Olivier Sibony: Another technique we find useful addresses the overconfidence bias. It is the “premortem,” invented by psychologist Gary Klein, whom we interviewed in 2010.³ In a premortem, you ask people to project themselves into the future and to assume that a deal has failed—not to imagine that it could fail, but to assume it already has. Then you ask them to write down, individually and in silence, the three to five reasons why it failed. And that forces people to speak up about the risks and the uncertainties that they’ve kept to themselves for fear of appearing pessimistic, uncommitted to the success of the proposal, or disloyal to the rest of the deal team.

A third technique is, at some point in the process, to write a memo explaining why the CEO should not do a deal, including the things the CEO would need to believe to not do it. Because by the time companies get to the actual decision meeting, everybody has forgotten about those reasons. So unless they’ve actually been recorded, no one’s left to argue the negative case. Everyone’s framing the positive case, and all the reasons you used to be worried about the deal have disappeared.

Here’s an example: when one company did a retrospective analysis of a deal that went wrong, it looked at a series of memos from the deal team to the investment committee, two months, one month, and two weeks before the deal was actually approved. The firm found that the top three things on a long list of worries in the first memo fell to the bottom of the list in the next memo and in the final memo had completely disappeared. Apparently, those concerns had been resolved to the team’s full satisfaction. But when the deal was done and the acquirers prepared to take possession of the company, guess what were the top priorities on their agenda: the same three things that had been swept under the rug in order to do the deal in the first place. This illustrates the dynamics of deal frenzy: when you sense that everybody around you wants to do a deal, you’re very prone to suppressing evidence that might lead you to not do it.

Another technique we’ve used is to develop a taxonomy of deals and a checklist for each type of deal. Companies that do a lot of deals, especially private-equity companies, tend to function by association and by pattern recognition and to look at a deal and say, “Oh, this one is just like this or that previous deal.” But usually the deals they’re reminded of are not the failures but the great successes. And once they latch onto that pattern recognition, it’s very difficult to see the broad range of things that actually can make the analogy irrelevant.

What you can do to remedy this bias is to use techniques such as multiple structured analogies or reference class forecasting.⁴ The names sound complicated, but the techniques are actually simple to apply. Essentially, they are ways of making sure that you look at a range of examples, not just one, and to explicitly analyze what makes those examples relevant and what could make them less relevant.

If you do enough deals so that you can actually recognize the different patterns, the way to use this technique is to identify the different types of deals and the things that matter for each. For instance, the things that we need to check in a deal where we acquire complementary product lines are not the same ones that we needto check for when we are doing a cross-selling kind of deal or a geographic-expansion kind of deal. So we will have different deal processes and different due-diligence checklists.

The Quarterly: What advice do you have for CFOs who want to incorporate these techniques into their decision-making processes?

Olivier Sibony: The crucial thing to keep in mind is that there isn’t one magic technique that will strip out all biases. This is more about putting in place a process that includes techniques to correct for the biases to which you’ve been susceptible in the past: probably not 20 techniques but 2 or 3 that you can use to help you avoid those biases in the future.

And once you put a process in place, it’s only valuable if it’s used consistently. First, because you’re going to learn and become better at using the process. Second, because it is precisely when you’re about to make a big mistake that you’re likely to have made an exception. The temptation, when you have a decision-making process, is always to say that for a really exceptional, difficult decision, we’re going to bypass the process, since the decision is an unusual one.

That’s precisely what you want to avoid. That’s why you need a process and the habit of following it, not just a tool kit of practices that you use from time to time. That’s why in areas where we don’t tolerate failure, we have routines. If you fly an aircraft, you don’t say, “The weather is really bad and we’re already behind schedule, so let’s skip the takeoff checklist.” You say, “This is a flight like every other one, and we’re going to use the checklist—that isn’t negotiable.”