The Report Isn’t the Lesson: How Leaders Turn Setbacks into System Upgrades

Imagine heading off on a work trip expecting to be back within a fortnight. You pack light and leave a few loose ends, because you’ll be home soon.

Then one technical problem becomes three. The plan changes, the return date slips, and weeks become months.

Now stretch that scenario to 286 days.

That’s what happened to NASA astronauts Butch Wilmore and Suni Williams. They launched on 5 June 2024 aboard Boeing’s Starliner for what NASA described as an eight-to-14-day crewed test flight to the International Space Station.

Instead, propulsion anomalies stretched the Starliner mission to 93 days while NASA assessed the risk. NASA ultimately decided not to return the astronauts on the Starliner, and the capsule returned uncrewed in September 2024. Wilmore and Williams didn’t return to Earth until March 2025, aboard a SpaceX Crew-9 craft 286 days after launch. Now, that’s an over-extended business trip!

Earlier this year (19 February 2026), NASA released the investigation report¹ into the mission. This 300-plus-page document’s central finding is sobering because it’s so recognisable: the risk didn’t come from one flaw.

Investigators describe an interplay of hardware failures, qualification gaps, leadership missteps and cultural breakdowns. NASA formally classified the event as a Type A mishap, its highest level.

You don’t need to run a space program to recognise that pattern.

Swap “propulsion anomalies” for product defects, missed compliance signals or a project that quietly drifted out of scope. The labels change, but the underlying dynamic remains the same. When the same risk categories and issues show up repeatedly, despite reviews, fixes and so-called ‘lessons learned’, the question needs to shift from ‘What went wrong?’ to ‘What does real learning require?’.

Most organisations are not short on incident reviews, post-mortems, audit findings or end-of-project retrospectives. They are, however, short on reliable mechanisms that convert insight into changed behaviour, systems and outcomes.

A thorough investigation is not the same as learning. Learning is what changes things next time; i.e. the decisions, checks, thresholds and, ultimately, organisational culture.

Why failure doesn’t teach by default

We are told (endlessly) that failure is the best teacher. And sure, sometimes it is. Yet, more often, it only teaches us if we treat it like data and make the learning deliberate.

Kellogg University’s research² captures the misconception nicely. Across multiple studies, people consistently overestimated the likelihood that someone will succeed after an initial failure.

We like the redemption story and assume that pain produces insight and leads to change.

The researchers argue this rosy view has consequences: when we assume failure naturally builds character, we’re more likely to underinvest in the support and structure people need to recover and improve.

This matters in organisations because the same logic shows up after a setback. For example:

• We hold a debrief because that’s what capable organisations do.
• We document findings because action needs evidence.
• We assign owners or point figures because accountability matters.

And then the calendar, incentives or politics win, and the learning from the event is soon forgotten. You don’t have to look too far across industries and governments to see examples of that playing out in real time.

We’re good at studying failure. We’re less good at taking those learnings and getting proportionately better at preventing future occurrences.

Not all failures are equal

Professor Amy Edmondson argues³ that leaders need to distinguish between different kinds of failure, because each requires a different response.

She outlines three types:

• Preventable failures in predictable operations: These are the ‘we should have known better’ category of failures. Errors, missed steps, unclear handovers, and avoidable rework. The fix is usually better training, simpler processes, clearer standards, and fewer workarounds.
• Unavoidable failures in complex systems: Multiple small breakdowns line up at once, for example, handoffs, timing, competing priorities, weak signals, and ambiguous ownership. There are multiple interacting factors, making it harder to identify one cause. Consequently, the fix is rarely a single root cause; it’s about strengthening the system’s ability to detect, escalate, and adapt.
• Intelligent failures at the frontier: Well-designed experiments where you’re genuinely learning something you could not know in advance. The failure is the price of information, and the goal is to make that price small and the learning high.

When leaders don’t separate these categories, two predictable things happen. Firstly, people get punished for intelligent failures, so experimentation goes underground and innovation becomes theatre. Secondly, preventable failures get excused as learning, so standards slip, and mediocrity gets a motivational poster.

Why post-mortems often don’t change anything

Even when intentions are good, learning fails for predictable reasons.

First, failure threatens identity. Leaders and teams protect competence, reputation and relationships. That produces defensiveness, selective memory, and careful storytelling.

Second, organisations crave a single cause. One cause feels controllable. It produces a neat corrective action that offers the comforting illusion that ‘we’ve fixed it’. Sadly, complex systems don’t cooperate with this type of storytelling.

Third, the organisation learns the wrong lesson. Sociologist Diane Vaughan’s concept⁴ of the “normalisation of deviance” explains how repeated anomalies, if they don’t immediately trigger catastrophe, gradually become accepted as normal. This isn’t stupidity. It’s drift, under pressure, in a system that rewards delivery.

Lastly, insight isn’t embedded. A finding is not a risk control, and a recommendation is not a new behavioural habit. Unless they are decision rights, meeting rhythms, process checklists, thresholds, workplace training, leaders’ behaviour, or escalation pathways don’t change.

Finding your options

So what do you do instead of repeating the “fail fast” mantra?

You want to learn small and learn often because it’s not about speed, it’s about learning density: how much useful information you get per unit of time, cost, and risk.

Research on debriefs is clear; reflection works when it’s done well. A meta-analysis⁵ found that properly conducted debriefs can improve performance by roughly 20–25%. Another meta-analysis⁶ on after-action reviews also found meaningful improvements, with outcomes depending on how the review is designed and executed.

So the mechanism you select and how you apply it matter. Here are five mechanisms you can apply.

Mechanism One – Make the bet small
When the stakes are high, failure does not just trigger problem-solving. It triggers self-protection. People manage optics, minimise exposure, and reach for certainty. That’s when post-mortems become more about defending a story than improving a system.

The antidote is to shrink the bet. Use small, safe experiments where the downside is bounded and the feedback is quick. When you opt for a pilot with guardrails rather than a big-bang transformation, you will learn faster and with greater accountability.

Mechanism Two – Lower the cost
Small experiments work because they lower the cost of being wrong. But leaders still need to lower the cost of being honest. That means separating learning from blame, rewarding early signals, and running debriefs that focus on what the system made likely, not who is at fault. If people believe speaking up will lead to embarrassment, punishment, or politics, they will stay quiet, and the organisation will “learn” only the safe, sanitised version.

You will want to set clear norms for debriefs. No blame. Curiosity first. Focus on what made the outcome likely.

In practice, it sounds like:

• What did we notice early?
• What felt unclear or risky?
• What did we do that made sense at the time?

When you model candour and transparency, your team will share earlier, and the learning gets richer and more useful.

Mechanism Three – Debrief on purpose (and keep it tight)
Even with the right tone, many debriefs fail because they sprawl. They become therapy, prosecution, or an endless hunt for “the” root cause.

Use a five-to-ten-minute review with ruthless constraints:

• What did we set out to do?
• What actually happened?
• What was the single biggest driver of the gap?
• What will we do differently next time?

Then stop. Capture one behaviour change and one system change, and schedule when you’ll test them.

Mechanism Four – Watch the stories you copy
Survivor tales are seductive. They are also statistically biased.

By that I mean they’re drawn from a biased sample. We hear from the people who made it through a setback and can now tell a clean, uplifting story. We don’t hear from those who failed and exited, who didn’t get another chance, or who were quietly sidelined.

Kellogg’s research (mentioned earlier) is a useful warning here. We tend to overestimate how quickly people naturally bounce back after failure, and then treat a single dramatic turnaround as a reliable blueprint.

So resist the headline anecdote. Look for patterns across many attempts: what consistently predicts improvement, what reliably trips people up, and what conditions make learning more likely.

Even the evidence for “failure fuel” is conditional. Research⁷ on early-career scientists found that near-misses were linked to stronger later performance among those who stayed in the field, but setbacks also increased the likelihood of people dropping out altogether.

Failure doesn’t automatically strengthen people. It strengthens people when there is enough support, opportunity, and runway to keep going.

Mechanism Five – Add scaffolding
Do not assume you or your team will bounce back on grit alone.

If the work is high-pressure, politically loaded, or personally exposing, you need support structures: coaching, peer feedback, time to reset, and psychological safety.

Where to next

Most failures don’t arrive with a bang. They show up as a wobble: a decision that didn’t land, a stakeholder who went cold, a meeting that created more heat than progress, a project that slipped for the third time.

Those moments are easy to wave away. You move on, patch it and promise yourself you’ll deal with it properly later. But later rarely comes, and, over time, small misses become familiar patterns.

So instead of waiting for a crisis to force a big investigation, use the next wobble as a small learning opportunity. Here’s a simple way to do that.

Pick one recent wobble. Not a catastrophe. Just something that fell short, and follow these five steps:

1. Name the moment. What outcome did you want, and what happened instead?
2. Find the signal. What drove the gap most: skill, process, timing, stakeholder, or context?
3. Write a rule. One sentence you will use next time to help guide the work/decision.
4. Shrink the next step. Design the smallest test that would prove the rule helps and schedule it.
5. Add support. Who can give quick feedback or remove friction before you run the test?

The point is not to become obsessed with failure. It’s to become disciplined about learning. Because while investigations matter, the resulting report is only the diagnostic. Leadership is the treatment plan.

You want to identify what type of failure you are dealing with, match the response to the type, and embed the learning into the system, whether it’s the meeting cadence, decision-rights, thresholds, checklists, training, escalation paths or something else.

You will learn more reliably from many small misses than from a few big crashes. The goal isn’t just to bounce back. It’s to build a system that adapts: one that catches problems earlier, responds faster, and repeats fewer of the same mistakes.

For more on embracing the non-linear nature of progress and learning from small wins and setbacks, see Michelle Gibbings’ article Progress Isn’t a One-Way Street

Getting you ready for tomorrow, today®

Award-winning author and global workplace expert, Dr Michelle Gibbings, helps leaders, teams and organisations unlock strategic influence to accelerate progress.

References

1. Warner, C. (2026, February 19). NASA releases report on Starliner crewed flight test investigation. National Aeronautics and Space Administration. https://www.nasa.gov/news-release/nasa-releases-report-on-starliner-crewed-flight-test-investigation/
2. Wu, Y. (2024, September 1). Why we shouldn’t romanticize failure. Kellogg Insight. https://insight.kellogg.northwestern.edu/article/why-we-shouldnt-romanticize-failure
3. Edmondson, A. C. (2011, April). Strategies for learning from failure. Harvard Business Review. https://hbr.org/2011/04/strategies-for-learning-from-failure
4. Vaughan, D. (2016). The Challenger launch decision: Risky technology, culture, and deviance at NASA (Enlarged ed.). University of Chicago Press.
5. Tannenbaum, S. I., & Cerasoli, C. P. (2013). Do team and individual debriefs enhance performance? A meta-analysis. Human Factors, 55(1), 231–245. https://doi.org/10.1177/0018720812448394
6. Keiser, N. L., & Arthur, W., Jr. (2021). A meta-analysis of the effectiveness of the after-action review (or debrief) and factors that influence its effectiveness. Journal of Applied Psychology, 106(7), 1007–1032. https://doi.org/10.1037/apl0000821
7. Wang, Y., Jones, B. F., & Wang, D. (2019). Early-career setback and future career impact. Nature Communications, 10(1), Article 4331. https://doi.org/10.1038/s41467-019-12189-3