On building experimentation culture

I get asked about "experimentation culture" a lot. Usually the question comes from someone who's bought a testing tool, run some experiments, and is frustrated that it hasn't spread beyond their team.

The uncomfortable truth is that culture isn't something you install. It's not about mandates from leadership or better tooling or more training sessions. It's about what questions people ask when no one is watching.

What culture actually looks like

In the best experimentation cultures I've worked with, I've noticed a few patterns:

People default to "how would we test this?" rather than "let's ship this and see." Not because they're told to, but because they've learned that testing leads to better outcomes.

Failures are genuinely interesting. When an experiment doesn't work, the conversation is about what was learned, not who was wrong. This sounds obvious, but it's rare.

There's institutional memory. Learnings from experiments actually inform future decisions. Someone can ask "didn't we test something like this before?" and get an actual answer.

The problem with mandates

I've seen companies try to force experimentation culture through process: mandatory test coverage, experimentation gates in the product development lifecycle, OKRs tied to test volume.

These rarely work. At best, you get compliance without conviction—people run tests because they have to, not because they believe in the approach. At worst, you get gaming: tests designed to pass, not to learn.

What I think works better

Instead of mandating behaviour, I've had more success with three approaches:

Start with visible wins. Find a sceptic, work with them on one well-designed experiment, and let the results do the talking. Belief spreads through demonstration, not instruction.

Make learning visible. Share experiment results broadly, including failures. Create a rhythm—weekly insights, monthly deep-dives—that keeps experimentation in the organisational conversation.

Build capability, not dependency. The goal isn't to centralise experimentation in one team. It's to give people the skills and frameworks to design their own experiments. This is slower, but more durable.

A note on patience

Culture change is slow. In my experience, it takes 12-18 months of consistent work before experimentation becomes genuinely embedded in how an organisation thinks. Most companies give up after six months.

If you're in that period where it feels like pushing uphill, that's normal. The question isn't whether it's hard—it is—but whether you believe in where it leads.