Analyzing Failure

We are often called upon to analyze failures: the customer’s system was down for an hour; it took a month to diagnose a technical problem; the expected contract renewal did not happen. And many times we try to reduce the answer to a simple answer: a particular person made a specific mistake. But in the complex environments where we work, I think we need a more sophisticated approach.

Enters Amy Edmondson and her latest book, Right Kind of Wrong: The Science of Failing Well which talks, at length, about how wonderful failure is, in the right setting–but I want to highlight here its taxonomy to classify failure.

• Sabotage: an individual chooses to violate a prescribed process or practice (this is rare, at least I hope so for your context)
• Inattention: an individual inadvertently deviates from what should be done (it’s an error, not a willful misdeed)
• Inability: an individual lacks the knowledge, attitude, or skills required to perform the task
• Challenge: the task is too difficult for an individual to reliably perform every time
• Uncertainty: events are unpredictable and even though people take reasonable actions, it doesn’t work. Edmonson’s examples include emergency departments and battlefields. I would include complex escalations here, too!
• Experimentation:  an experiment conducted to expand knowledge and investigate a possibility just does not work

The taxonomy is organized from blameworthy (at the top) to praiseworthy (at the bottom), and she further highlights that we must observe both the overall system and context and the behavior of individual actors. We tend to jump too quickly to individual blame, when in fact the issue may be found in the process. For instance:

• An individual did not respond to an incoming customer case in time (inattention) because we are so understaffed that it’s impossible to attend to all customer requirements in a timely manner.
• We don’t train team members well enough to do the work they are expected to do (inability).
• We designed a system that expects individuals to get to root cause within 30 minutes, but realistically it takes an hour (challenge).

Would a structured taxonomy help your retrospective efforts? Please share.

And if you need help setting up retrospectives, we can help.