If You Don’t Get Praised for Failing, You’re in the Wrong Organization

In knowledge work, your failure rate isn’t a liability — it’s the leading indicator of how much value you’re capable of creating

Here’s an uncomfortable proposition: if your organization isn’t celebrating your failures, it has structurally locked itself out of creating large value.

Not all failures. Not careless mistakes or sloppy execution. But the deliberate, hypothesis-driven experiments that didn’t work out — the prototype that flopped, the strategy that misfired, the product bet that didn’t land. If those failures are met with blame, silence, or career damage instead of curiosity, analysis, and genuine praise, you are working inside an organization that has chosen — whether it knows it or not — to compete with one hand tied behind its back.

The argument I want to make here isn’t about employee wellbeing (though it helps that, too). It’s about economics. In knowledge work, the asymmetry between safe execution and bold experimentation is so extreme that failure tolerance isn’t a cultural nicety — it’s a strategic necessity. And organizations that understand this are beginning to treat their failure rate not as something to minimize, but as something to measure, celebrate, and optimize. Failure, done right, is the KPI.

The Asymmetry That Changes Everything

Most organizations operate as if value creation is linear: work harder, produce more, succeed more. This model made sense in industrial economies. A factory worker who makes 100 widgets is worth roughly twice as much as one who makes 50. Risk-taking adds little. Deviation from the established process is waste.

Knowledge work operates on completely different mathematics.

In knowledge economies, the distribution of value is wildly skewed. A single breakthrough — a new product insight, a novel business model, a strategic reframe — can be worth more than years of competent, incremental execution. Google’s ad model. Pfizer’s mRNA pivot. Amazon’s AWS. None of these emerged from safe, predictable processes. They all required someone to propose something uncertain, pursue it despite the possibility of failure, and be supported by an organization that could tolerate the ambiguity.

A meta-analysis of 85 studies found that psychological safety — the condition under which people feel free to take risks — increases idea generation by roughly 25% and team performance by 30%. But those are averages. The real story is in the tails. The ideas that matter most aren’t the ones that move the average slightly; they’re the ones that emerge from the 95th percentile of creative risk-taking, the proposals that most organizations would never hear because the person who had the idea calculated that proposing it wasn’t worth the personal risk.

McKinsey’s research reinforces this: in times of uncertainty, investing only in incremental changes — business as usual — is actually riskier than bold innovation. The organizations that pull ahead during volatile periods aren’t the ones that hunker down. They’re the ones that increase their experimentation rate precisely when conditions are most ambiguous.

This means that a knowledge organization’s most valuable output isn’t the work that succeeds on the first try. It’s the rate at which it generates and processes intelligent failures — the experiments that produce learning, even when they don’t produce the hoped-for result. If your failure rate is zero, you’re not excellent. You’re timid. And in a competitive landscape shaped by compounding innovation, timidity is the most expensive strategy of all.

Edmondson’s Taxonomy: Not All Failure Is Created Equal

The reason most organizations get this wrong is that they treat all failure as the same thing. A preventable mistake and a bold experiment that didn’t work are filed under the same category: bad.

Amy Edmondson, the Novartis Professor of Leadership and Management at Harvard Business School and arguably the world’s foremost authority on this topic (ranked #2 on the Thinkers50 global list in 2025, #1 in both 2021 and 2023), has spent decades trying to fix this. In her 2023 book Right Kind of Wrong: The Science of Failing Well — winner of the Financial Times and Schroders Business Book of the Year — she introduces a taxonomy that should be printed on the wall of every knowledge organization:

Basic failures are preventable mistakes in known territory. A surgeon operating on the wrong limb. A developer pushing broken code because they skipped the test suite. These failures happen when people deviate from established knowledge and process. They should be minimized through training, checklists, and systems design. Nobody should celebrate them.

Complex failures are system-level breakdowns where multiple factors converge. No single cause. No single person to blame. A hospital-acquired infection. A financial crisis. These require post-mortems and systems thinking, not blame.

Intelligent failures are the engine of all progress in new territory. They occur when you’re pursuing a meaningful goal, operating in territory where the answer genuinely can’t be looked up in advance, acting on a well-formed hypothesis informed by existing knowledge, and keeping the experiment as small as possible to minimize waste while still generating learning.

This is the critical distinction. Intelligent failures aren’t errors. They aren’t mistakes. They’re the only mechanism through which new knowledge is produced when you’re working at the frontier. Edmondson puts it plainly: trial and error is a common term for this kind of experimentation, but “error” is a misnomer. Error implies there was a right way to do it in the first place. In genuinely new territory, there is no right way yet. The failure is the process of finding it.

The pharmaceutical case is illustrative. When Eli Lilly spent over a decade developing the drug Alimta to treat lung cancer and the medication came up short in expanded trials, that failure was worthy of praise. The scientists had no way to advance the project other than to experiment and observe the results. The failure met all four criteria — new territory, meaningful goal, hypothesis-driven, appropriately scoped. And the learning from that failure ultimately led to a reformulated version that succeeded. Without permission to fail, the breakthrough never arrives.

The problem is that most organizations can’t make this distinction operationally. They say they value innovation, but their incentive systems, performance reviews, promotion criteria, and cultural norms all punish the same thing: being associated with something that didn’t work. The result is a workforce that has rationally concluded that the safest career strategy is to never propose anything uncertain. And an organization that has, without quite realizing it, optimized for mediocrity.

Failure Rate as a Leading Indicator

Here’s where the argument gets practical. If intelligent failure is the mechanism through which knowledge organizations create their most valuable outputs, then the rate of intelligent failure isn’t a problem to manage — it’s a metric to track.

Only 22% of companies currently have metrics in place to track innovation performance effectively, according to McKinsey. And of those that do, 60% focus on final outcomes — revenue from new products, market share gains — while only 35% track early-stage pipeline metrics. Almost none track failure rate as a positive indicator.

This is a massive blind spot. As one innovation expert at the Wharton School’s Mack Institute put it, measuring failure rate is essential because if you don’t have any failures, that probably means you’re not trying enough things. Failure is toxic when it’s hidden, but it’s generative when it’s recorded, analyzed, and learned from. Another participant in the same research forum noted that companies need to embrace failures and learn well from them to be better next time — and that this requires making failure a measurable, visible part of the innovation process.

Think about what an intelligent failure rate as a KPI actually measures. It measures the organization’s willingness to operate at the frontier. It measures the volume of experiments being run — the raw input of the innovation engine. It measures whether people feel safe enough to propose uncertain ideas and report honestly when those ideas don’t pan out. And it measures the organization’s capacity for learning, because a failure that generates no insight is just waste, while a failure that surfaces a new understanding is an investment.

Innovation-forward organizations are starting to build this into their dashboards. They track “number of innovation experiments initiated” as an input metric alongside traditional output metrics. They track “kill rate” — how many innovation projects are deliberately abandoned after generating learning — as a health indicator rather than a sign of waste. They conduct post-mortems focused on what was learned rather than who was responsible.

But the truly differentiated organizations go further. They don’t just tolerate failure reporting — they celebrate it. They create forums where teams present what went wrong and what they learned. They give awards for the best-executed failed experiment. They make the learning from failure as visible and valued as the results from success. This isn’t touchy-feely management theater. It’s a deliberate signal to the entire organization: the thing we value most is the rate at which you generate new knowledge, and that requires failing at a rate that would terrify a traditional organization.

The Neuroscience Problem: Why Fear Kills the Innovation Engine

If failure tolerance is so obviously valuable, why don’t more organizations practice it? Because fear is neurologically powerful, culturally entrenched, and organizationally rational in the short term even though it’s catastrophic in the long term.

Research in cognitive neuroscience shows that social pain — the fear of looking incompetent, being blamed, losing status — activates the same neural circuits as physical pain. When people operate under the threat of being punished for failure, their brains divert physiological resources from the regions that manage working memory, process new information, and generate creative insights. Fear doesn’t just discourage risk-taking at the motivational level. It physically degrades the brain’s capacity for the kind of thinking that knowledge work demands.

Edmondson captures this precisely: motivation by fear is indeed highly effective — but effective at creating the illusion that goals are being achieved. It is not effective at ensuring that people bring the creativity, good process, and passion needed to accomplish challenging goals in knowledge-intensive workplaces.

This illusion is dangerous because it’s self-reinforcing. Leaders set ambitious targets. People, afraid to report honestly, present optimistic progress. Leaders, seeing apparent progress, push harder. People, under increasing pressure, cut corners or hide problems. The cycle continues until reality breaks through — usually catastrophically.

The 2026 research from MHFA England quantifies the damage: nearly half of employees feel unsafe speaking up at work, and one in seven report making more mistakes specifically due to stress and fear. A systematic review published in Health Psychology in 2025, synthesizing 84 studies and nearly 35,000 employees, found a strong statistical association between employee silence and burnout. A longitudinal study in the Journal of Business Ethics (2025) found that involuntary silence predicts declining organizational identification, which in turn predicts rising workplace incivility. Silence corrodes everything it touches.

And perhaps most critically for the argument about failure as KPI: a 10-week experience sampling study published in the Journal of Occupational and Organizational Psychology in 2025 found that voice behavior and psychological climate exist in a reciprocal cycle. When people speak up and are heard, the climate improves, which encourages more voice. When voice is suppressed, the climate deteriorates, suppressing future voice further. The system is self-reinforcing in either direction. Once an organization enters the silence spiral, reversing it requires far more energy than maintaining a culture of openness would have cost in the first place.

Case Study: Nokia — What Happens When Failure Becomes Unspeakable

Nokia’s collapse from controlling roughly 30% of the global mobile phone market to selling its phone business to Microsoft in 2013 is not primarily a technology story. It’s a story about what happens when an organization makes failure unspeakable.

A groundbreaking study by Tim Vuori (Aalto University) and Quy Huy (INSEAD), based on interviews with 76 Nokia managers, engineers, and consultants, revealed an organization consumed by fear. Top management had a history of aggressive behavior — shouting, banging tables, threatening removal. Middle managers, terrified of delivering bad news, began falsifying their reports. They moved data points on charts. They reported smartphone development progress that wasn’t real.

Nokia’s engineers knew their Symbian operating system was far behind Android and iOS. They knew the company was losing the innovation race. But speaking that truth felt futile and dangerous. The fear culture had created what the researchers called “temporal myopia” — an organization so consumed by short-term political survival that it couldn’t process long-term strategic threats.

The result: Nokia’s top management, receiving falsely optimistic reports, pushed even harder for delivery. Products were rushed to market with declining quality. The N97, Nokia’s iPhone competitor, shipped with weaker connections than previous Nokia models and a nearly unusable touchscreen. Meanwhile, the fundamental problem — that Symbian couldn’t compete as a platform — went unaddressed because no one could safely say so.

When Vuori published the research, he was approached by people from about 10 other Finnish organizations sharing nearly identical stories from their own companies. As one INSEAD commentator observed: fear-based management is perceived as an easier way to control employees, but when it becomes the dominant force, it grows exponentially and leads inevitably to worse results.

Nokia didn’t fail because it lacked smart engineers. It failed because those engineers couldn’t fail safely. Every intelligent failure that should have happened — every experiment with alternative operating systems, every honest assessment of competitive positioning, every prototype that would have revealed Symbian’s limitations years earlier — was prevented by a culture that made the cost of being associated with failure higher than the cost of collective self-deception.

Case Study: Volkswagen — Stretch Goals Without Permission to Miss

Volkswagen’s Dieselgate scandal follows the same pattern with even starker consequences.

Edmondson identifies VW as a textbook case of what she calls the recipe for organizational failure: stretch goals combined with closed ears. Ambitious targets are valuable — they drive performance. But ambitious targets without the permission to report that those targets can’t be met through legitimate means is a formula for fraud.

Under Ferdinand Piëch and Martin Winterkorn, VW cultivated a culture of extreme performance pressure with zero tolerance for falling short. A competitor described it as a reign of terror driven by fear and intimidation. Former employees reported that presenting bad news could become unpleasant, loud, and demeaning. One management trainee compared the culture to “North Korea without labor camps.”

When VW’s engineers couldn’t meet aggressive diesel emissions targets through real engineering, they had two choices: report the failure honestly and face personal consequences, or cheat. They installed defeat devices — software that detected when a vehicle was being tested and altered its performance. The seeds of a $9.5 billion scandal grew over years while senior management remained blind — not because the information didn’t exist, but because the culture had made it unspeakable.

VW’s engineers were, in effect, prevented from failing intelligently. The honest outcome — “we’ve tested this approach and it can’t meet the emissions targets with current technology; here’s what we learned and here’s what we propose instead” — was structurally impossible. Failure wasn’t a KPI. It wasn’t even tolerable. And so when failure inevitably occurred, it took the form of corporate fraud rather than productive learning.

Edmondson frames this with a single devastating insight: VW had stretch goals but closed ears. Half of that equation drives excellence. The combination drives catastrophe.

The 27,000-Person Study: Psychological Safety as Organizational Infrastructure

In November 2025, Harvard Business Review published findings from Edmondson, Michaela Kerrissey (Harvard T.H. Chan School of Public Health), and HBS doctoral student Hassina Bahadurzada. Their study examined survey data from over 27,000 U.S. healthcare workers collected before and during the COVID-19 pandemic.

The results reframe psychological safety not as a cultural preference but as organizational infrastructure — as fundamental to performance as technology stacks or supply chains.

Increasing psychological safety by one standard deviation decreased burnout by 0.72 points and increased willingness to stay by 0.63 points. Workers with higher psychological safety coped better with severe resource constraints — not because they had more material resources, but because the social environment itself functioned as a resource for resilience.

The effects persisted across time. Psychological safety measured in 2019 predicted employee retention during the 2021 crisis. Organizations that had built this infrastructure before the pandemic reaped its benefits during it.

And the effects were strongest for the groups whose voices are most often suppressed: physicians, women, and people of color experienced disproportionate protection from burnout when psychological safety was high. This matters enormously for the failure-as-KPI argument, because the people whose intelligent failures are most likely to be punished — those with less institutional power — are also the people with the most untapped potential to generate diverse, novel insights.

Kerrissey framed the implication clearly: by thinking of psychological safety as a “social resource,” organizations unlock a new dimension of performance that can’t be purchased with material investment alone. Many organizations have crisis management plans but say almost nothing about establishing the social infrastructure that makes honest communication possible during difficult times.

Google’s Proof of Concept: Project Aristotle

Google’s Project Aristotle, studying over 180 internal teams across two years, independently confirmed the same principle. After analyzing hundreds of variables, the research team found that individual talent, team composition, seniority, co-location, and team size did not predict team effectiveness. Psychological safety did — and it was, in Google’s framing, the foundation on which every other dynamic rested.

Teams where members could admit mistakes, propose uncertain ideas, and challenge assumptions without fear of embarrassment were the highest-performing teams in the company. They brought in more revenue. They were rated as effective twice as often by executives. They were less likely to lose members.

The 2019 DORA State of DevOps report extended these findings beyond Google, demonstrating that psychological safety is predictive of software delivery performance, organizational performance, and productivity across the technology industry.

What’s particularly relevant for the failure-as-KPI argument is what Project Aristotle found about conversational dynamics. High-performing teams exhibited equality in conversational turn-taking — everyone spoke in roughly equal proportions. They didn’t have the smartest people. They had the safest environment. And in safe environments, more ideas surface, more experiments get proposed, more failures get reported honestly, and more learning accumulates. The team with the highest failure throughput — intelligently managed — is the team that learns fastest and eventually wins.

Building the Failure KPI: What It Looks Like in Practice

So what does it mean operationally to treat failure as a KPI?

First, you separate the signal from the noise. Not all failure is signal. Basic failures — mistakes from carelessness or ignorance in known territory — should be minimized, not celebrated. The KPI isn’t “total failure rate.” It’s intelligent failure rate: the number of hypothesis-driven experiments in new territory that generated actionable learning, regardless of whether they produced the desired outcome.

Second, you make failure visible. This means creating institutional rituals — post-mortems, learning reviews, “failure forums” — where teams present what they tried, what happened, and what they learned. The format matters: these can’t be blame sessions disguised as learning exercises. They must be genuinely curious, focused on knowledge extraction rather than responsibility assignment. The audience should be asking “what did we learn?” not “whose fault was this?”

Third, you reward the practice, not just the outcome. If your promotion criteria, performance reviews, and bonus structures only reward successful outcomes, you’ve built an incentive system that penalizes experimentation regardless of what your stated values say. The most innovative organizations reward people for the quality of their experimental design — were they in new territory? Was there a real hypothesis? Was the experiment appropriately scoped? — not just for whether the experiment happened to work.

Fourth, you measure experimentation velocity. How many experiments is each team running per quarter? How quickly do experiments move from proposal to execution to learning extraction? If a team has run zero experiments in six months, that’s not a sign of competence — it’s a sign of fear. The absence of failure is diagnostic. As Edmondson notes: if you are not regularly experiencing intelligent failures, you are probably not operating at your full potential.

Fifth, leadership models the behavior. This is non-negotiable. If leaders don’t publicly share their own failures, don’t visibly celebrate others’ intelligent failures, and don’t respond to bad news with curiosity rather than anger, no amount of process design will overcome the cultural signal. Edmondson suggests three concrete practices: frame the work as a learning problem rather than an execution problem, acknowledge your own fallibility openly, and model curiosity by asking more questions than you answer.

The Industrial Hangover: Why Most Organizations Still Get This Wrong

Most organizations still operate on what Edmondson calls a vestige of the industrial era. In the age of mass production, reliable execution was the path to competitiveness. Once a process and a market were established, you could succeed for decades without experimentation or change. Deviation from the standard was waste. Failure was incompetence. The entire management apparatus — performance reviews, promotion criteria, hierarchical reporting, command-and-control leadership — was built to minimize variance.

This mental model persists even though the work has fundamentally changed. Only 35% of organizations track early-stage innovation pipeline metrics. Only 22% have any formal innovation measurement at all. And 41% of employees report suffering from “KPI overload” that specifically inhibits their creativity and experimentation. The measurement systems in most organizations are not just failing to track the right things — they’re actively suppressing the behaviors that generate the most value.

The default management mindset — failure is not an option, only perfection is acceptable — is worse than wrong in knowledge work. It’s actively destructive. It ensures that the only failures that occur are the catastrophic, unmanaged kind — the ones that happen because honest signals were suppressed for so long that reality eventually forces a reckoning. Nokia. VW. Wells Fargo. Enron. In every case, the organization had the talent, the resources, and the market position. What it didn’t have was the ability to fail safely, learn quickly, and correct course before the accumulating deception became unsustainable.

The Decision

So here’s the test. Look at your organization and ask a single question: What happened the last time someone failed at something genuinely new?

If the answer is that the person was blamed, sidelined, or learned never to try again — you’re in an organization that has structurally capped its own potential. It may be profitable. It may be stable. But it is operating well below its innovation frontier, and over time, that gap will widen until a competitor that does tolerate intelligent failure eats its market from underneath.

If no one has failed at anything genuinely new in recent memory — that’s even worse. It means the experiments aren’t happening at all. The organization has achieved the most dangerous state: a zero failure rate, which means a zero learning rate, which means a zero adaptation rate. In a competitive landscape shaped by exponential technological change, that’s not stability. It’s a countdown.

The organizations that will dominate the next decade are the ones that understand failure not as the opposite of success but as the input to success. They’re building cultures where intelligent failure is praised, measured, and optimized — where the question isn’t “why did this fail?” but “what did this failure teach us, and how quickly can we run the next experiment?”

If your organization isn’t one of them — if your failures are punished instead of studied, hidden instead of shared, blamed instead of celebrated — the most strategic thing you can do for your career is find one that operates differently. Not because it feels better (though it does). But because your ability to generate the bold, uncertain, frontier-pushing work that creates outsized value can only exist in an environment where failure is treated as the work itself.

Find the organization that will praise you for your best failures. That’s where the biggest value gets created — and where your best work will happen.

Sources and further reading:

Edmondson, A.C., Right Kind of Wrong: The Science of Failing Well (Atria Books, 2023) — Winner, FT/Schroders Business Book of the Year
Edmondson, A.C., The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth (Wiley, 2018)
Edmondson, A.C., Kerrissey, M., & Bahadurzada, H., “Psychological Safety as an Enduring Resource Amid Constraints,” International Journal of Public Health (2024), featured in Harvard Business Review (November 2025)
Vuori, T.O. & Huy, Q., “Distributed Attention and Shared Emotions in the Innovation Process: How Nokia Lost the Smartphone Battle,” Administrative Science Quarterly (2016)
INSEAD Knowledge, “Who Killed Nokia? Nokia Did” (2022)
Google re:Work, “Guide: Understand Team Effectiveness” (Project Aristotle)
DORA State of DevOps Report (2019)
MHFA England, “The Cost of Workplace Silence” (February 2026)
Systematic review: “Associations between burnout, employee silence and voice,” Health Psychology, 84 studies, N=34,975 (2025)
Journal of Business Ethics, “Still Waters Run Deep: How Employee Silence Affects Instigated Workplace Incivility Over Time” (2025)
Parker et al., “A 10-week longitudinal study of voice and silence,” Journal of Occupational and Organizational Psychology (2025)
Zhu, Lv & Feng, meta-analysis of 85 studies on psychological safety and team performance (2022)
McKinsey & Company, innovation measurement and technology trends research (2025)
Wharton Mack Institute, “Innovation Metrics for Success in the Digital Era” — on failure rate as measurement
Foundernest, “The New Metrics of Innovation: How KPI Adoption Is Transforming in 2025”
Global Peter Drucker Forum, “The Next Knowledge Work: How Can Innovation and Value Creation Be Managed?”