Most marketing teams say they value experimentation. Few actually practise it. The gap between intention and reality usually comes down to culture — the unwritten rules, incentives, and behaviours that determine whether your team runs fast, fearless experiments or defaults to safe, predictable work.
Building a culture of experimentation is not about buying a testing tool or running a few A/B tests. It is about creating an environment where testing ideas is the default way of working, where failure is treated as learning, and where data consistently beats opinion.
Table of contents
Open Table of contents
What is a culture of experimentation?
In a culture of experimentation, the team tests ideas before committing resources, measures results rigorously, and uses those learnings to inform future decisions. It is the opposite of the HiPPO effect — where the Highest Paid Person’s Opinion determines what gets done.
In an experimentation culture:
- Ideas are treated as hypotheses, not mandates
- Running experiments is expected, not exceptional
- Failure is a source of learning, not blame
- Data informs decisions at every level
- Anyone on the team can propose and run tests
Research from Stefan Thomke, professor at Harvard Business School, found that the biggest barrier to scaling experimentation is not technology — it is culture. In his study of Booking.com, Microsoft, and Amazon, the organisations that experimented most successfully had made experimentation a core part of how they operated.
“To build a culture of experimentation, you need an environment where curiosity is nurtured, data trumps opinion, anyone can run a test, and managers embrace a new model of leadership.”
Stefan Thomke, Harvard Business School
Why experimentation culture matters
The case for experimentation is straightforward: most marketing ideas do not work. Research from Ronny Kohavi, former VP of experimentation at Microsoft, found that only about one-third of experiments at companies like Microsoft, Amazon, and Booking.com produce positive results. Another third show no significant difference, and the final third actually make things worse.
Without experimentation, your team is likely spending two-thirds of its effort on things that either do not help or actively hurt results. A culture of experimentation helps you find the one-third that works — and stop wasting resources on the rest.
Companies that embed experimentation into their culture also tend to:
- Move faster. Testing small ideas quickly beats debating big ideas slowly.
- Make better decisions. Data from experiments replaces gut feelings and internal politics.
- Compound learnings. Each experiment teaches you something, even when it fails. Over time, these learnings accumulate into a real competitive advantage.
- Retain talent. Marketers want autonomy and the ability to test their ideas. A culture of experimentation gives them that.
How to build a culture of experimentation
1. Leadership has to go first
Experimentation culture starts at the top. If leadership rewards only successful outcomes and punishes failure, the team will avoid risk. If leadership celebrates the process of testing and learning — regardless of individual results — the team will experiment more.
Steven Bartlett, founder of Social Chain, makes this point directly:
“Your team either behave in that way or they don’t, and that’s usually, honestly, down to the leadership. Whether the leadership are risk-taking, whether they’re fast moving, whether they trust their team members, whether they empower people, whether they’re punishing people for doing the wrong things. All of these things matter.”
Steven Bartlett, founder of Social Chain
Leaders need to model the behaviour they want to see. Share your own failed experiments openly. Ask “what did we learn?” instead of “why didn’t it work?” Make it clear that not experimenting is riskier than running a test that fails.
2. Make experimentation a habit, not an event
The best experimentation cultures treat testing as a continuous process, not a quarterly initiative. Build experimentation into the team’s regular workflow.
Practical steps:
-
Set an experiment cadence. Commit to launching a minimum number of experiments per week or sprint. Even one experiment per week adds up to 50 per year — far more than most marketing teams run.
-
Build experiment reporting into your regular meetings. Steven Bartlett requires his team to report on experiments weekly:
“Even in our team now, every Monday, even my social team, they report up to me a list of experiments that they did in the last 7 days.”
Steven Bartlett, founder of Social Chain
-
Use a prioritisation framework like ICE or RICE to decide which experiments to run first. This removes opinion from the process and gives everyone a shared language for evaluating ideas.
-
Track experiment velocity. Measure how many experiments your team runs per month. This is one of the strongest leading indicators of growth.
3. Reward the process, not just the outcomes
This is the hardest cultural shift. Most organisations reward success and punish failure. In an experimentation culture, you reward the act of experimenting — the hypothesis, the test design, the rigour of measurement — regardless of whether the experiment “won.”
Practical approaches:
- Celebrate experiments that produced clear learnings, even if the result was negative
- Share failed experiments in team meetings with the same visibility as successes
- Recognise team members who run the most experiments, not just those who run the most successful ones
- Avoid tying individual performance reviews to experiment win rates
4. Lower the cost of running experiments
The easier it is to run an experiment, the more experiments your team will run. Remove friction wherever you can:
-
Reduce approval layers. If every test needs sign-off from three stakeholders, your team will test less. Give marketers the autonomy to run low-risk experiments without approval.
-
Invest in tools. Use platforms that make it easy to set up, run, and measure experiments without engineering support.
-
Start small. Not every experiment needs to be an elaborate A/B test. A simple before-and-after comparison, a landing page test, or a new email subject line all count.
-
Timebox experiments. Set clear start and end dates. Experiments that drag on indefinitely lose momentum and block future tests.
5. Build a system for capturing and sharing learnings
Without a system for capturing what you have learned, each experiment exists in isolation and the team keeps re-learning the same lessons.
- Maintain an experiment log that records the hypothesis, design, results, and key learnings for every test
- Make the log searchable and accessible to the whole team
- Review past experiments before designing new ones — someone may have already tested a similar idea
- Share a monthly digest of the most interesting learnings across the organisation
Common mistakes
Treating experimentation as optional. If experiments only happen when there is “spare time,” they will never happen consistently. Experimentation needs dedicated time and resources, like any other core marketing activity.
Optimising too early. Teams often jump straight to optimising existing channels (A/B testing button colours) before testing bigger strategic bets. Balance quick-win optimisation with higher-risk, higher-reward experiments.
Confusing activity with experimentation. Launching a new campaign is not an experiment unless you have a hypothesis, a control, and a way to measure the result. Be rigorous about what counts as an experiment.
Giving up after early failures. The first few experiments often fail. This is normal — remember, two-thirds of experiments at the best companies do not produce positive results. The value comes from the cumulative learning, not any single test.
Measuring experimentation culture
Track these metrics to gauge whether your experimentation culture is working:
-
Experiment velocity — How many experiments does your team run per month? Increasing velocity is the clearest signal that experimentation is becoming a habit.
-
Idea-to-experiment ratio — How many ideas does your team generate versus how many actually get tested? A low ratio points to bottlenecks in execution.
-
Learning documentation rate — What percentage of experiments have documented learnings? If learnings are not being captured, the compound value of experimentation is lost.
-
Decision quality — Are strategic decisions increasingly backed by experiment data rather than opinion? This is harder to measure but worth tracking qualitatively.
Booking.com, one of the most cited examples of experimentation culture, runs over 25,000 experiments per year. You do not need to match that number, but the principle holds: the more you experiment, the faster you learn and grow.
“Experiment patiently, accept failures, plant seeds, protect saplings, and double down when you see customer delight.”
Jeff Bezos, Amazon, 2016 Letter to Shareholders
Getting started
You do not need to transform your culture overnight. Start with one change: commit to running one experiment per week for the next month. Track the results. Share the learnings. Build from there.
Growth Method is the work management platform built for growth teams, combining ideation, experimentation, and analytics in one place. Book a call to learn more.