Monday, November 5, 2012

Celebrate Failure? [Part 1]

Earlier this year I wrote about Agile's perspective on failure. In that blog I indicated that my brother-in-law (a psychologist) sent me some of the latest research on failure. In particular, there were two fascinating studies that helped me understand why failure is indeed a cause for celebration. This is the first in a two part series.

"We find that organizations learn more effectively from failures than successes" 

In a 2010 study (more details at the bottom of this post) on the effects of failure and success on organizational learning, a team of researchers found that failure was a crucial ingredient for longer term success.

In order to find suitable organizational data to support their research they searched for and found an ideal candidate - the Orbital Launch industry. This industry was ideal for the following factors:
  • Every launch had a high incentive to succeed due to the high cost of failure.
  • Placing objects in space is a relatively new activity so data is available for all launches ever attempted.
  • Because it is a high profile industry the records were relatively easy to find.
  • The sample data contained 4663 launch attempts, 443 failures, 36 organizations, and 9 countries. The data starts with the launch of Sputnik 1 on October 4, 1957 and ends March 2004.
(Note: The Orbital Launch industry is responsible for sending rockets into space to place a payload into orbit around the earth.)

As researchers their goal was to look at the causes of improved organizational performance. Did success drive improvements? What part did failure play in future success? Here are a few of their key findings:
  1. Organizations learn more effectively from failures than from successes. Success causes organizations to be complacent in the belief that they have figured it all out. On the other hand, failure increases the desire to learn and challenge existing beliefs. Those organizations that fail end up being more successful in the end.
  2. Success breeds complacency and overconfidence and reduces the incentive to learn. Organizations who regularly succeed may in fact suffer from this experience in the long term. 
  3. Organizations learn more from large failures than from smaller ones. As part of a community that believes in failing (learning) fast, this one is the hardest for me. But at the least it supports the notion of celebrating failure when it occurs in the larger or the small.
  4. Organizations should embrace failure so that they can learn from it. One response to failure is to punish those involved or hide the failure. Organizations that are open about their failures have an increased chance of learning and then improving because of it.
As many people like Luke have pointed out - failure isn't the goal. However, when it happens we can either run from it or celebrate it. The evidence from the Orbital Launch industry strongly suggests that celebrating failure is in order. For even more reason to celebrate, take a look at part 2.

Subscribe to Winnipeg Agilist by Email

You can get the full research from your neighbourhood grad student or PhD but here are a few more details on the studies mentioned above:

Failing to Learn? The Effects of Failure and Success on Organizational Learning in the Global Orbital Launch Vehicle Industry

Abstract: "It is unclear whether the common finding of improved organizational performance with increasing organizational experience is driven by learning from success, learning from failure, or some combination of the two. We disaggregate these types of experience and address their relative (and interactive) effects on organizational performance in the orbital launch vehicle industry. We find that organizations learn more effectively from failures than successes, that knowledge from failure depreciates more slowly than knowledge from success, and that prior stocks of experience and the magnitude of failure influence how effectively organizations can learn from various forms of experience"

Discussion and Conclusion: "These findings do not imply that organizations that fail in period t are more likely to succeed in period t + 1 than are organizations that succeed in period t. But they do imply that organizations that fail in period t improve their own likelihood of succeeding in period t + 1 (relative to their likelihood of success in period t) more than do organizations that succeed in period t. Our definition of learning is inherently self-focused - change in organizational performance as a result of prior experience. ... Consequently, the result presented here suggest that experience with failure allows organizations to improve their performance relative to their own previous baseline, but that experience with success does not generate similar levels of improvement. 

Indeed, this study not only yielded strong evidence that organizations learn by observing their own and others' failures, but also failed to uncover evidence of significant learning from observation of their own or others' successes. In the full models, coefficients estimating the effect of success experience on future performance are indistinguishable from zero. We don not interpret these results as evidence that organizations cannot learn from success to improve performance. But the fact that launch vehicle organizations (which face significant incentives to learn from success as well as failure) did not experience demonstrable learning from success suggests that organizational learning from success is far from an automatic process."

Implications for Practice: "Failure is often difficult for organization members to cope with. Because failures - and those that appear to be involved in them - are often stigmatized, organization members frequently refuse to acknowledge failure, refrain from communicating about it, or redefine it as success (March et al., 1991). Indeed, in Vaughan's (1996, 2005) analyses of the Challenger and Columbia disasters, she noted that the most significant organizational antecedent to both tragedies was the institutionalized practice of ignoring failures. 

Nonetheless, given failure's central role in organizational learning shown here, organizations that stigmatize failure may be depriving themselves of major opportunities for improvement. Consequently, the most significant implication of this study of practice is that organization leaders should neither ignore failures nor stigmatize those involved with them; rather, leaders should treat failures as invaluable learning opportunities, encouraging the open sharing of information about them. Indeed, this suggestion dovetails with existing evidence that members of organization that treat failure nonpunitively report more errors, but experience fewer serious failures, than member of organizations that seek to assign blame for failures (Edmondson, 1999; Weick, Sutcliffe, & Obstfeld, 1999)."