If at first you don’t succeed, then go ahead and fail, but do it intelligently. Failure, in fact, not only should be dissected and analyzed, but should be planned for—and, if done well, celebrated. That’s because failure, according to Amy C. Edmondson, not only is inevitable in complex systems (such as a busy hospital emergency department), but is a key innovation tool. The tragedy occurs when organizations fail to learn from failure, and most of them do, she writes in the April 2011 issue of Harvard Business Review.¹
As health care embarks on a period of great experimentation, the lessons that Edmondson imparts are well timed for this market. The author reviews the role and spectrum of failure (from deviance to exploratory testing), but she also addresses ways to limit the liabilities associated with failure.
Edmondson lays the responsibility for the inability to learn from failure squarely at the feet of leaders, who are invariably guilty of viewing failure as something bad (though not all failures are) and are unable to embrace the lessons of failure due to cultural beliefs and stereotypical notions of success. “In organizational life it is sometimes bad, sometimes inevitable, and sometimes even good,” she says.
Begin by understanding how the blame game gets in the way of learning from mistakes. Executives Edmondson has interviewed in a range of organizations, including hospitals, worry that without the fear of blame, people will not do their best work. The author identifies a spectrum of reasons for failure to explain why a culture in which it is safe to admit blame can—and in some organizations, must—exist in an organization that adheres to the highest standards.
On the blameworthy end of the spectrum, it begins with deviance, followed by inattention, lack of ability, and process inadequacy. Task challenge is at the center of the spectrum, followed by (and progressing toward the praiseworthy) process complexity, uncertainty, hypothesis testing, and exploratory testing, the most praiseworthy reason for failure.
Very few of the reasons for failure on this spectrum are purely blameworthy, and in fact, most executives acknowledge that the percentage of blameworthy failures is in the low single digits. Because 70% to 90% are treated as blameworthy, however, many failures go unreported and therefore are likely to recur.
Failures fall into three broad categories: preventable in predictable operations, unavoidable in complex systems, and intelligent at the frontier. Preventable failures are to be avoided and usually involve deviations from specified processes in high-volume or routine operations. Training, checklists, and systems that adapt processes (based on what is learned from small mistakes) are measures that are successful in preventing such failures.
Unavoidable failures in complex systems can usually be attributed to an unpredictable combination of factors that might never have occurred simultaneously before in a complex environment, such as a hospital emergency room, a battlefield, or a fast-moving startup. Such failures are inevitable, Edmondson believes, and the best approach is to identify them rapidly, so that they do not trigger consequential, larger failures. “Most accidents in hospitals result from a series of small failures that went unnoticed and unfortunately lined up in just the wrong way,” she notes.
Intelligent failures occur when experimentation is required, providing an organization with valuable new knowledge. At the frontier, managers hope for experiments that fail quickly, thereby preventing the failure of a larger-than-necessary experiment, which is what Edmondson calls an unintelligent failure.
The Leader’s ROle
Tolerating the failures that provide knowledge requires leadership that can shift the organization away from blame and toward a culture of learning—in which managers seek the reason for a failure, rather than the perpetrator. All organizations learn from failure through three activities: detection, analysis, and experimentation.
Detection, however, can be tricky. Big failures are easy to spot, but the small failures that are unlikely to cause serious harm stay hidden, in many organizations. Edmondson relates an anecdote about Alan Mulally, president and CEO of Ford Motor Company (Dearborn, Michigan), who instituted a new system for detecting failure by color coding reports green for good, yellow for caution, and red for problems. Mulally grew increasingly