From Experiment Theatre to Learning Discipline

The moment it disappeared

I watched three years of work evaporate in six months.

Alongside the founding team we built an experimentation capability from nothing. Established cadence. Created processes that gave stakeholders input without killing the learning. Unlocked data that had been hoarded across departments. Started shipping things with actual validated evidence behind them.

Then our executive sponsor left. New leadership arrived. Restructures followed. Within two quarters, the teams got absorbed back into business-as-usual. The experimentation muscle we'd built? Gone.

This is how we built a genuine learning discipline inside one of Australia's largest insurers, and why it couldn't survive. Not because the work failed, but because we'd anchored it at the wrong altitude.

If you're running innovation inside an enterprise, this might save you the same mistake.

Starting from zero

When I joined the internal design lab, we had no true experiments running.

Not struggling experiments. Not poorly-designed experiments. Just some ideas that needed turning into experiments. The team had a mandate to improve customer experience across the insurance product set. We had budget. We had executive sponsorship. What we didn't have was any way to validate ideas before they hit development backlogs, where they'd sit for months, get descoped or quietly die.

The structural problems were familiar to anyone who's worked in a large enterprise:

Silos protecting fiefdoms. Every department guarded its territory. Data sat locked in systems you couldn't access without navigating three levels of approval. Design had no direct line to engineering, everything went through intermediaries with their own priorities.

Political approval from people you didn't report to. Getting anything live meant sign-off from managers whose teams might be affected, even tangentially. Each one wanted input. Each one had concerns. Each one could slow you down indefinitely.

Resources tied up in negotiation. We spent more time on internal politics than customer problems.

The cost wasn't just slow delivery. It was that initiatives got stuck, diverted, or diluted until they solved nothing. We were shipping, but we weren't learning.

The cultural barrier nobody warned us about

What we discovered: the organisation couldn't accept that experiments can fail.

Sounds obvious. Of course experiments can fail, that's the point. You're testing a hypothesis. Learning requires both success and failure as valid outcomes. But in practice? Failure was seen as negative. A black mark. Something to avoid, explain away or blame on someone. We had strong executive sponsorship. Real air cover from the top. Budget and mandate to innovate. Should have been enough. It wasn't.

Middle management was the blocker. Every manager whose team might be touched by an experiment wanted input. Not unreasonable, but their input consistently pushed one direction: de-risk it. Reduce the scope. Limit customer exposure to statistically meaningless numbers. Add controls that eliminated the very uncertainty we needed to test. The requests were always framed as responsible governance. But the cumulative effect was experiments so constrained they couldn't generate learning. We'd run them, get inconclusive results and everyone would shrug. See? Experimentation doesn't work here.

The irony was brutal. The guardrails designed to reduce risk were guaranteeing failure.

The deliberate compromise

We made a decision that felt wrong at the time: run imperfect experiments anyway.

Not because we thought they'd generate breakthrough insights. They wouldn't, not with the constraints we were operating under. We ran them because we needed to show people what experimentation looked like, not just talk about it.

Call it: Experimentation Theatre. We did.

The goal wasn't learning from individual experiments. It was building cultural comfort with the idea of experimentation. Getting people used to these conversations. Normalising the vocabulary. Making "we're running a test" a sentence that didn't trigger defensive reactions.

This only works if you're clear-eyed about what you're doing. Experimentation theatre is a transitional tactic, not a destination. If you're still doing theatre after 12 months, you've failed. The point is to create conditions for real experiments, then graduate to them.

Building the discipline

Cycle by cycle, we moved from theatre to substance.

We established cadence. Multiple experiments running, each with defined timelines. Stakeholders knew when results would land. The rhythm itself built credibility, this wasn't a one-off innovation project, it was how we worked.

We separated involvement from interference. Stakeholders got input on what we tested, which problems, which customer segments and helped shape hypotheses. They didn't get to dictate how we tested. This was critical. It gave people ownership without giving them veto power over methodology.

We created a feedback loop that told a better story. Run experiment → de-risk initiative → ship initiative → measure value. Each cycle showed concrete outcomes. Not "we learned something interesting" but "we validated this, shipped it, here's the impact."

That feedback loop changed everything. Suddenly we weren't asking for permission to experiment, we were showing results that made the next experiment easier to approve.

The unlock was middle management buy-in. Once managers saw experiments reducing their risk, killing bad ideas before they consumed resources, resistance dropped. And with their buy-in came something we'd never had: data access. Systems opened up. Cross-functional collaboration became possible. We could finally measure what mattered.

We'd gone from zero experiments to a functioning learning discipline. It took nearly three years.

The collapse

Then our executive sponsor moved on.

New leadership arrived with different priorities. Restructures followed, the design lab teams got moved, split, recombined. The new leaders weren't hostile to experimentation; they just weren't invested in it. They had their own agendas, their own fires to fight.

Without active protection, the capability eroded fast. Experimentation cadence slipped. The processes we'd built got deprioritised. People who understood the methodology moved to other roles. Within six months, the teams had been absorbed back into standard enterprise product work.

Business as usual won.

What we got wrong

Looking back, the failure wasn't in how we built the capability. It was in where we anchored it.

We had VP-level sponsorship. Strong, genuine, protective sponsorship. But when that VP left, the sponsorship left with them.

What would have made this survive?

CEO or Board-level commitment. Not just awareness. Active commitment. A mandate that existed above any individual executive, embedded in how the organisation defined its strategic priorities. Something that couldn't be quietly deprioritised when leadership changed.

A protected space with institutional backing. Innovation labs get disbanded all the time. The ones that survive have formal protection: dedicated budget lines, board-level reporting, explicit multi-year commitments that make dissolution a visible strategic decision rather than an operational convenience.

Realistic timelines. Culture change takes years. Not quarters. The expectation that experimentation capability can be "stood up" in a few months is fantasy. We were just hitting our stride at year three. Cutting it off at that point meant losing the investment right before it would have compounded.

What this means for you

If you're leading innovation in an enterprise, here's what I'd tell you:

Experimentation theatre is sometimes necessary. Don't feel guilty about running constrained experiments to build cultural comfort. It's a legitimate transitional tactic. But be honest with yourself about what you're doing, and have a clear strategy for moving beyond it.

Middle management is your unlock, or your ceiling. Executive sponsorship gets you started. Middle management buy-in gets you data, resources, and sustainability. Without them, you'll be fighting the same battles forever. Invest in showing them how experimentation reduces their risk.

The feedback loop sells the work. Abstract innovation metrics don't land. "We ran experiments and learned things" doesn't land. What lands is: we tested this, shipped it, measured the impact, here's the value. Build your processes to generate that story.

Anchor at the right altitude. This is the one we got wrong. VP-level sponsorship feels like enough. It isn't. If your innovation capability can't survive a single leadership change, it's not a capability, it's a project. And projects end.

Push for CEO or Board-level commitment. Push for multi-year timelines with explicit protection. Push for institutional anchoring that makes dismantling the capability a strategic decision, not an organisational convenience.

Because here's the truth: building a genuine learning discipline in an enterprise is hard. It takes years. It requires navigating politics, building trust, converting sceptics and demonstrating value over and over again.

But losing it? That's easy. That happens in a single reorg.