About ten years ago, I worked with a mid-sized metal fabrication shop whose business had outgrown its physical facility. Our goal was to improve flow through the shop by rethinking layout and workplace organization. During this effort, a young leader pulled me aside to discuss relocating a large sheet metal rack. He had thought carefully about the move but encountered resistance from others in the organization.

As he explained his rationale—the current state, the proposed change, and its potential benefits—it became clear that he was seeking validation from me, the “lean expert.” I couldn’t give it to him. The truth was, he understood the situation better than I did. After listening, I simply said, “Why not? The worst that can happen is that it doesn’t work, and we move it back.”

Moving a rack loaded with thousands of pounds of metal is no small task. But if it failed, the business would not collapse. More importantly, if the change made the work easier, it would be a clear success. And if it didn’t, the team would learn something valuable: failure was manageable, reversible, and not catastrophic.

The rack was relocated and resulted in a modest improvement in workflow; however, it became clear that a more comprehensive reorganization of the area was required to achieve the desired results. The most significant change, however, was not operational but behavioral. The leader’s attitude toward experimenting with improvement ideas shifted noticeably, along with a greater acceptance of failure as a natural part of the improvement process.

We are often taught that experiments have only two meaningful outcomes—success, or failure accompanied by some profound lesson. This framing creates pressure to succeed and, paradoxically, pressure to extract deep meaning from failure. That pressure can produce unhealthy behaviors. Organizations may label marginal results as “big wins,” or worse, continue investing time and resources into initiatives that should be abandoned, simply to avoid admitting failure. This raises an important question: does this mindset discourage experimentation altogether? Is there a third outcome—“that didn’t work as expected”—where we adjust and move on without drama or blame?

Do individuals and organizations that consistently succeed think differently about failure?

Over the past decade, extensive research has focused on psychological safety. In 2012, Google conducted a major internal research initiative, code-named Project Aristotle, to identify what makes teams effective. The most important factor was psychological safety.

Amy Edmondson, a leading researcher in this field, defines psychological safety as “a belief that one will not be punished or humiliated for speaking up with ideas, questions, concerns, or mistakes.” More broadly, it is a shared sense that interpersonal risk-taking—candor, vulnerability, and admitting errors—is safe.

Psychological safety is essential. Knowing that failure will not be punished creates space for experimentation. But even in psychologically safe environments, most people—including myself—still don’t want to fail.

Daniel Kahneman’s Prospect Theory, described in “Thinking, Fast and Slow,” may help explain why. People experience losses far more intensely than gains; a loss hurts roughly twice as much as an equivalent gain feels good. As a result, people often avoid risks that are statistically favorable, simply because the possibility of loss feels too painful.

So what needs to change? How do we build a healthier relationship with failure in order to sustain continuous improvement?

First, we must embrace uncertainty and doubt. Failure is often defined as the absence of success, and success as achieving a specific goal. Historically, however, success had a broader meaning. In the 16th century, the term—derived from the Latin successus, meaning “to follow” or “to turn out”—referred simply to an outcome, whether good or bad. Only later did it become synonymous exclusively with positive results.

Scientific thinking reinforces this broader view. It asks us to form hypotheses about what we expect to happen, knowing we may be wrong. Learning occurs in the uncomfortable space between knowing and not knowing. The longer that we are in that space, the more we learn. If all our hypotheses are correct, we are likely not learning very much. As Voltaire observed, “Doubt is not a pleasant condition, but certainty is absurd.”

Second, organizations can benefit from adopting a venture capitalist mindset toward improvement. Venture capitalists invest in portfolios, knowing that most investments will fail or produce minimal returns. A few will generate moderate gains, and one or two will deliver outsized returns that justify the entire portfolio. Overall returns of 10–30% or more are common, despite frequent failures.

I have seen a similar approach succeed in a large manufacturing organization producing high-volume, make-to-order products. Teams within manufacturing cells are encouraged to develop improvement ideas aligned with the company’s strategic goals—making work easier, better, faster, and cheaper. Teams submit a one-page description of their idea, and the engineering department designs and builds the required fixtures or equipment, often at significant cost.

Notably, there is no formal approval process beyond the team itself. Engineering implements every idea it receives. Many of these initiatives fail outright or require substantial rework. Yet the organization recognizes that rejecting ideas early risks discouraging future innovation—including ideas with significant upside. Teams also own the outcomes of their ideas, which builds confidence when initiatives succeed and resilience when they do not. Taken together, the total return across all ideas—including failures—is strongly positive.

In contrast, many organizations rely on complex vetting processes designed to ensure that every initiative delivers a guaranteed return. While this may appear prudent, it sends a clear message: failure is not acceptable. This slows improvement and suppresses initiative. While experimentation should not be reckless—and even venture capitalists apply investment criteria—a better approach is often to encourage many small improvements, provide clear boundaries and guidance, and give teams the authority and responsibility to act. This requires leaders to tolerate failure, trusting that the net gains will outweigh the losses.

Finally, organizations should encourage continuous improvement as a daily practice. Like a portfolio strategy, many small experiments can produce substantial cumulative gains. Some ideas will fail, most can be reversed, and little harm will be done. Over time, this builds failure tolerance and sustains engagement, allowing valuable ideas to surface that would otherwise remain hidden.

Ultimately, the challenge may not be reframing failure as something positive or heroic. Failure is uncomfortable, and it likely always will be. Perhaps the real goal is simpler: to accept that failure is survivable, that progress does not require constant success, and that we can move forward—whether or not we extract a profound lesson every time. Building tolerance for failure may be the most practical path to continuous improvement.

Continuous improvement depends not on avoiding failure, but on developing tolerance for it. When organizations embrace uncertainty, encourage small experiments, and view improvement efforts as a portfolio rather than a series of guaranteed wins, learning accelerates and innovation increases. Psychological safety creates the conditions for people to try, but true progress comes when leaders accept that many efforts will not work as intended—and that this is not only acceptable, but necessary. By normalizing reversible, low-risk failure, organizations unlock sustained improvement and long-term success.