Good managers—even great ones—can make spectacularly bad choices. Some of them result from bad luck or poor timing, but a large body of research suggests that many are caused by cognitive and behavioral biases. While techniques to “debias” decision making do exist, it’s often difficult for executives, whose own biases may be part of the problem, to know when they are worth applying. In this article, we propose a simple, checklist-based approach that can help flag times when the decision-making process may have gone awry and interventions are necessary. Our early research, which we explain later, suggests that is the case roughly 75 percent of the time.
Biases in action
In our experience, two particular types of bias weigh heavily on the decisions of large corporations—confirmation bias and overconfidence bias. The former describes our unconscious tendency to attach more weight than we should to information that is consistent with our beliefs, hypotheses, and recent experiences and to discount information that contradicts them. Overconfidence bias frequently makes executives misjudge their own abilities, as well as the competencies of the business. It leads them to take risks they should not take, in the mistaken belief that they will be able to control outcomes.
The combination of misreading the environment and overestimating skill and control can lead to dire consequences. Consider, for instance, a decision made by Blockbuster, the video-rental giant, in the spring of 2000. A promising start-up approached Blockbuster’s management with an offer to sell itself for $50 million and join forces to create a “click-and-mortar” video-rental model. Its name? Netflix. As a former Netflix executive recalled, Blockbuster “just about laughed [us] out of their office.” Netflix is now worth over $25 billion. Blockbuster filed for bankruptcy in 2010 and has since been liquidated.
In retrospect, it is easy to ascribe this decision to a lack of vision by Blockbuster’s leadership. But at the time, things must have looked very different. Netflix was not, then, the video-on-demand business it has since become: there were nearly no high-speed broadband connections of the kind we now take for granted, and widespread use of video streaming would have seemed like a futuristic idea. In Blockbuster’s eyes, Netflix, with its trademark red envelope, was merely one of several players occupying a small (and thus far unprofitable) mail-order niche in the video business.
Furthermore, this was the very time when the dot-com bubble had burst: as the Nasdaq Composite Index quickly collapsed from its March 2000 high, many managers of traditional companies felt vindicated in their belief that investors had grossly overestimated the potential of Internet-based models. Through the lens of the confirmation bias, Blockbuster’s executives likely concluded that the approach Netflix had made to them was evidence of its desperation. And it did not take a lot of overconfidence for them to assume that they could replicate Netflix’s mail-order model themselves, should they ever decide to do so.
The overconfidence and confirmation biases weren’t the only ones at work at Blockbuster, of course, just as in most organizations. But they are important enough to warrant special attention.
An intractable problem?
Fortunately, debiasing techniques can help organizations overcome such biases. These techniques aim to limit the effects of overconfidence by forcing the decision maker to consider downside risks that may have been overlooked or underestimated. And they can mitigate the dangers of confirmation bias by encouraging executives to consider different points of view.
Examples of such techniques include either the systematic use of a devil’s advocate or a “premortem” (individuals project themselves into a future where the decision has failed and imagine, in prospective “hindsight,” what failed and why). Another technique is to organize a formal scenario-planning exercise—expanding the range of assumptions underpinning a plan—or even a war game, in which executives put themselves in their competitors’ shoes. One study of investment decisions showed that when a company uses a range of debiasing techniques, its return on investment rises considerably. For high-impact, repetitive decisions, such as large investments, it is sensible to embed debiasing techniques in a company’s formal decision-making processes.
But this doesn’t solve things for the myriad daily decisions that are the bread and butter of executives. A war game or a scenario-planning exercise entails a significant investment of time; how are senior leaders to know when that is worthwhile? Furthermore, the very nature of biases means that the person driving the decision process generally cannot judge whether further debiasing is needed. Indeed, that executive may be experiencing the confirmation bias and overconfidence at the crucial time. When managers make an ordinary mistake, such as a calculation error, they can learn from their experience and avoid repeating it. But when biases lead them astray, they are not aware of what’s happening, so experience does not help them become better at debiasing themselves, and they cannot “just watch out” to keep their biases in check.
Two tests of decision readiness
Since executives won’t get very far by focusing directly on biases, they should consider instead whether safeguards against them have been used. In other words, leaders should ask about the process used to develop the proposal, not about the proposal itself or the degree of confidence it inspires. Exhibit 1 suggests questions for evaluating the process in the context of the two main categories of biases described earlier:
- The first set of questions (“Consideration of different points of view”) aims to determine whether the confirmation bias has been kept in check. These questions focus on the sources of assumptions and the diversity of opinions expressed. A broad set of sources (including outside views) or a diverse set of opinions is a good indicator that the initial assumptions of the decision process have not gone unchallenged.
- A second set of questions (“Consideration of downside risk”) asks whether the possibility of negative outcomes—including company-, industry-, and macro-level downsides—has been thoroughly evaluated. Such an evaluation can act as a safeguard against overconfidence.
On each dimension, the questions are designed to be flexible, so that the circumstances of the decision at hand can be taken into account. Once the questions have been answered (with a simple yes or no), the responses can be transcribed on a matrix (Exhibit 2). This scoring will place the proposed decision in one of four quadrants, leading to different courses of action:
- Decide. This quadrant represents the most favorable outcome: the process that led to such a decision appears to have included safeguards against both confirmation bias and overconfidence.
- Reach out. Proposals that fall in this quadrant have been tested for their resilience to downside risks but may still be based on overly narrow assumptions. Decision makers should consider techniques that broaden their perspectives and help them generate meaningful alternatives. One such technique is the vanishing-options test: executives force themselves to generate new ideas by imagining that none of the proposals on the table are available.
- Stress-test. Decisions in this quadrant reflect a variety of viewpoints but, nevertheless, may not have been sufficiently challenged and could therefore be tainted by overoptimism. Executives should consider a thorough outside review of the possible risks—for instance, by conducting a premortem or asking an outside challenger to play the role of devil’s advocate.
- Reconsider. When a decision appears in the bottom-left quadrant, the process has probably not been comprehensive. Decision makers should therefore follow a dual strategy that generates both new perspectives and new reviews of risks.
By using this decision-screening tool, a company can learn if it needs to expand its focus and options in the strategy process. We recently applied a version of the tool together with 26 senior executives of European corporations from a variety of industries, ranging from construction to manufacturing, services, and retail. We asked these executives to analyze a strategic-decision proposal that a project team within their own organization (but not the participants) had recently made.
Only just over a quarter of the proposals, it emerged, were truly decision ready. The bar for readiness on each dimension (three positive answers out of six questions) was relatively low. Yet a striking 73 percent of the respondents judged that the decisions they were reviewing did not pass these tests. They then used the prescriptions of the matrix to revisit the decisions.
How to use the decision-screening tool
A key question is who answers the questions in the tool. Since individuals developing a recommendation will not be aware of their biases, they cannot be expected to assess their own decision readiness. The answers must therefore come from the outside: not the executive who has driven the decision process, but others who have a more neutral view.
In practice, decision makers will be in one of two situations. In the first, and easiest, they reviewed recommendations prepared by others but had minimal involvement in developing them. In that case, decision makers are well placed to address the screening tool questions themselves.
But in the second and more frequent case, the decision makers were actively involved in studying decisions that have now reached the final stage. In this case, they no longer have an outside view of the process and will need to seek out answers from informed observers: staff members, such as the CFO; colleagues from other parts of the organization; or outside advisers. Some companies will wish to define this role in advance and make it a formal part of their decision-making process, to avoid having a respondent who shares the decision maker’s point of view.
In an environment of change and disruption, many leaders fear—rightly—that their companies do not take enough risks or will fall prey to “analysis paralysis” and let opportunities slip away. Hence the popularity of start-ups as role models of fast, iterative decision making. As Reid Hoffmann’s often retweeted quote goes, “If you are not embarrassed by the first version of your product, you’ve launched too late.”
While this “better sorry than safe” mind-set characterizes many successful start-ups, it may not be the best inspiration for the strategic decisions of mature companies. Some risks are worth taking: those taken knowingly, in pursuit of commensurate rewards. But some risks are taken recklessly because the risk takers are blind to their own overconfidence or have failed to consider alternative viewpoints.
The disciplined use of decision aids such as this screening tool offers a way to spot bad decisions before they happen, without significantly slowing down the decision process. Executives who adopt this approach will free up resources for value-creating projects—and improve their chances of keeping the names of their companies off the roll call of organizations that made notorious blunders.