The disciplined practice of testing ideas, approaches, and assumptions through small, safe-to-fail experiments that generate evidence and insight. Experimenting is not about proving what you already believe, but about exploring possibilities, reducing uncertainty, and allowing successful patterns to emerge. Leaders who experiment build adaptability, innovation, and resilience in complex environments by learning their way forward.

“In complex systems, you can’t predict the future. You run a series of small experiments, see what works, and let direction emerge.” Dave Snowden

Why experimenting matters

Experimenting matters because in complex, fast-changing environments, no one can reliably plan their way to the right answer in advance. The only way to reduce real uncertainty is to test assumptions against reality in small, controlled ways and let the evidence guide the next step. Leaders who experiment build organisations that adapt as conditions shift, rather than committing early to a single untested plan and defending it long after the evidence has turned against it. Small, safe-to-fail experiments generate the kind of concrete, situation-specific knowledge that no amount of analysis or forecasting can substitute for.

Without a culture of experimentation, organisations default to over-investing in big, visible initiatives chosen more for confidence than evidence, and treat any sign of failure as something to hide rather than learn from. This drives risk-taking underground: people stop proposing untested ideas, problems get addressed with the same familiar tools regardless of whether they’re working, and genuine innovation stalls. Leaders who normalise testing, reviewing, and adjusting build teams that surface useful failures early and cheaply, long before those failures would otherwise show up as expensive, public ones.

“Our success is a function of how many experiments we do per year, per month, per week, per day.” Jeff Bezos

What good and bad experimenting looks like

What bad looks like What good looks like
Insists on having a definitive answer before acting, treating admission of uncertainty as a weakness to hide. Openly names what is uncertain and proposes a small test to learn more, rather than pretending to already know.
Lets pilots or projects grow so large, costly, or visible that they can no longer be stopped without serious consequences. Keeps experiments small, bounded, and genuinely safe-to-fail, so they can be shut down quickly if they aren’t working.
Runs tests without a clear hypothesis, so even a completed experiment produces no clear answer about what was actually learned. Frames each experiment around a specific question or assumption, so results generate a clear, actionable insight.
Treats a failed experiment as a personal or team failure, punishing the outcome rather than examining the learning. Treats a failed experiment as valuable data, extracting the insight and moving on without assigning blame.
Has no pre-agreed signal for when to scale up or stop an experiment, so decisions get made on gut feel or attachment to the idea. Defines in advance what result would justify scaling up and what would justify stopping, then follows that signal even when it’s inconvenient.
Launches experiments without a way to measure what happened, relying on impressions or convenient anecdotes afterward. Builds in clear measures and observation from the start, combining data and stories to judge the outcome honestly.
Creates an environment where people fear ridicule or blame for proposing an unconventional idea, so risk-taking stays hidden. Protects people who propose and run experiments from blame, making it safe to suggest something that might not work.
Runs experiments for the appearance of innovation, then moves on without ever reviewing what was learned. Builds in a structured review after every experiment, so insights are captured and shared rather than lost.

“If you aren’t experiencing failure, then you are making a far worse mistake: you are being driven by the desire to avoid it.” Ed Catmull

Barriers to experimenting

Pressure for certainty: Leaders are often expected to provide definitive answers. This discourages experimentation, which begins with admitting uncertainty and exploring options instead of promising solutions, and it pushes leaders toward confident-sounding plans over honest, testable ones.

Fear of failure: In cultures where mistakes are punished, experiments become too risky. Leaders and teams avoid experimentation if failure is seen as weakness rather than learning, and over time the safest-looking option always wins, even when it isn’t the best one.

Over-investment: When experiments are too large, costly, or visible, they lose their “safe-to-fail” nature. Leaders sometimes label pilots or projects as experiments, but if they cannot be stopped without consequences, they are too risky, and the label becomes a way of avoiding scrutiny rather than genuine testing.

Bias toward big initiatives: Many leaders favour large-scale, well-structured projects over small steps. This bias creates rigidity and reduces the organisation’s ability to adapt when conditions change, since a single large bet is much harder to reverse than several small ones.

Short-term urgency: The pressure for immediate results often overrides curiosity. Leaders may default to action plans instead of exploring uncertainties first, missing opportunities for deeper learning that would have made the eventual action plan more likely to succeed.

Attachment to favourites: Leaders sometimes fall in love with their own ideas, amplifying them regardless of feedback. This undermines the objectivity needed to test and learn, and it quietly discourages a team from delivering results the leader doesn’t want to hear.

Confusing pilots with experiments: Pilots are designed to validate a chosen solution. Experiments are designed to test assumptions. Treating one as the other risks narrowing learning and missing unexpected insights that a genuinely open test would have surfaced.

Low psychological safety: Without trust, people hesitate to suggest unconventional or risky ideas. Teams avoid proposing experiments if they fear embarrassment, ridicule, or blame for failure, so the ideas most worth testing are often the ones that never get raised.

Neglecting feedback: Experiments that lack clear measures or observation mechanisms yield little value. Leaders sometimes run tests without monitoring signals, resulting in wasted effort and anecdotal conclusions that tell a comfortable story rather than an accurate one.

Activity without review (innovation theatre): In some organisations, experiments are launched for the sake of looking innovative, but results are never systematically reviewed. Without analysis and reflection, experimentation becomes busywork instead of a learning tool, and the same untested assumptions get carried forward indefinitely.

“I have not failed. I’ve just found 10,000 ways that won’t work.” Thomas Edison

Enablers of experimenting

Authorise safe-to-fail: Encourage experiments that are small, bounded, and reversible. Leaders who make it clear that failure is acceptable signal that learning is valued more than the appearance of success, which invites people to propose ideas they’d otherwise keep to themselves.

Frame clear hypotheses: Each experiment should test a specific question or assumption. This keeps experimentation disciplined and ensures that outcomes generate actionable insights, not just activity, and it makes it obvious afterward whether anything was actually learned.

Design a portfolio of probes: Running several small experiments in parallel provides resilience. Diversity across approaches, focus areas, or teams increases the chance of discovering valuable patterns and reduces reliance on a single initiative succeeding.

Set boundaries: Keep experiments time-boxed, limited in cost, and ring-fenced in scope. This protects safety and reassures stakeholders that risks are being managed responsibly, which makes it easier to get support for the next experiment too.

Define amplify and dampen triggers: Establish clear signals for when to scale up an experiment or when to stop it. This prevents bias, ensures timely action, and reinforces discipline, especially in the moments when attachment to an idea would otherwise cloud the decision.

Balance data and stories: Collect both quantitative evidence (metrics, uptake, performance indicators) and qualitative evidence (anecdotes, team narratives, customer stories). This gives a richer picture of impact and helps detect early weak signals that pure numbers would miss.

Value learning as much as outcomes: Celebrate insights gained even from failed experiments. Leaders who shift the focus from success/failure to learning foster a culture of exploration and resilience, and they get more, not fewer, honest attempts as a result.

Model curiosity: Leaders who say, “Let’s try and see what happens,” normalise uncertainty. By showing openness and curiosity, they encourage others to take thoughtful risks and explore new ideas without waiting for permission.

Create review loops: Build structured reflection into team routines so that every experiment is analysed, insights are captured, and learning is shared. This prevents experiments from being forgotten and strengthens organisational memory over time.

Protect cultural and psychological safety: Leaders must shield experimenters from blame or reputational damage. In risk-averse or hierarchical cultures, this protection is even more critical. When people know they are safe, they are more willing to try bold experiments rather than only safe ones.

“In organisations, real change begins with small disturbances.” Margaret Wheatley

Self-reflection questions for experimenting

Comfort with uncertainty: How willing am I to admit I don’t know and explore options rather than provide answers? When was the last time I told my team, “I don’t know, let’s test it”? Do I see not knowing as a weakness or as a chance to learn?

Risk mindset: How do I personally respond when an experiment does not deliver the expected result? Do I treat it as failure, or as a source of insight? What signals do I send to others about my tolerance for setbacks?

Experiment size: Are the experiments I run truly small and safe-to-fail, or do I over-commit resources? Would I feel comfortable shutting one down quickly if it was not working? How often do my pilots become too big to abandon?

Curiosity in action: Do I actively seek out alternative ideas and test them, or default to proven paths? How often do I try something new when the outcome is uncertain? Do I encourage my team to suggest experiments, even unconventional ones?

Feedback discipline: Do I design experiments with clear ways of knowing if they are working? What data or stories do I rely on to assess outcomes? How do I avoid only seeing what I want to see?

Amplify and dampen: Do I pre-define conditions for scaling up or stopping an experiment? When have I let bias override signals to stop? What signals would tell me that something deserves more investment?

Time for reflection: Do I create space to review experiments and embed learning? How do I ensure that insights are shared across the team? What practices help me avoid letting experiments fade without follow-up?

Psychological safety: How safe do people feel proposing an experiment that might not work? What have I done, intentionally or not, that might discourage someone from suggesting something unconventional? How would I know if fear of blame was quietly suppressing good ideas in my team?

Portfolio thinking: Do I tend to bet everything on one big initiative, or do I run several smaller probes in parallel? What would it look like to spread risk across a portfolio of small experiments rather than one large one? How comfortable am I with some of those probes simply not working out?

Hypothesis discipline: Before I start an experiment, do I know exactly what question or assumption it is meant to test? How often have I run something that, in hindsight, didn’t actually tell me anything useful? What would change if I insisted on a clear hypothesis before every test?

“If you want to test a new idea, the cost of not testing it is often higher than the cost of failure.” Chris Argyris

Micro practices for experimenting

1. Name the assumption before you test it: Before launching any experiment, write down in one sentence the specific assumption being tested and what result would prove it wrong. Share this with your team before starting, not after. This small discipline prevents experiments from quietly becoming pilots that only ever confirm what you already believed.

2. Set the stop rule in advance: For every experiment, decide up front what result, or how much time or budget, will trigger stopping it. Write the rule down before you start, when you are least emotionally attached to the outcome. This makes it far easier to end something that isn’t working without it feeling like defeat.

3. Run two, not one: Where possible, test two different approaches to the same problem in parallel rather than committing fully to your first idea. Keep both small enough to compare fairly. This habit builds the instinct for portfolio thinking and reduces the temptation to fall in love with a single favourite.

4. Debrief every experiment, especially the failed ones: After each experiment finishes, hold a short, structured conversation about what was learned, regardless of whether it succeeded. Ask explicitly, “What surprised us?” and “What would we do differently?” This keeps learning visible and stops good insights from disappearing along with a failed test.

5. Thank the person who proposed the risky idea: When someone suggests an experiment that might not work, acknowledge the suggestion itself as valuable, separate from whatever the outcome turns out to be. Do this visibly, in front of others where appropriate. This is one of the fastest ways to build the psychological safety that sustained experimentation depends on.

“If you are not embarrassed by the first version of your product, you’ve launched too late.” Reid Hoffman

Explore related leadership resources

To further develop this capability, examine how it intersects with other core leadership dimensions across the libraries:

Leadership library:

  • Questions (Asking good): Use powerful, open-ended inquiries to challenge assumptions and uncover the underlying variables worth testing.
  • Creativity: Generate a diverse range of novel ideas and unconventional solutions that serve as the raw material for your experiments.
  • Adaptability: Pivot your strategy based on real-world feedback, ensuring you don’t stay committed to a path that the evidence no longer supports.
  • Perseverance: Maintain the stamina to continue testing and iterating even when initial experiments fail to yield the desired results.

Supporting libraries

Continue exploring: Return to the Leadership Library to view the full directory of competencies and resources.