Most innovation doesn’t fail during the experiment. It fails after it succeeds. Once an experiment shows promising results, leadership gets excited, and scaling begins. However, six months later adoption is uneven, reporting skyrocket, and managers hesitate. What happened?

Experiments survive as they operate in protected space, when scaling is done it removes exactly the conditions that made experimenting successful.

Innovation rarely dies in the experimental phase. It dies in the institutionalization phase.

Constraints on NGO experimentation are not primarily technical or cultural. They are institutional. And until those institutions are redesigned to accommodate uncertainty, experimentation will continue to be talked about more than it is practiced.

Can the public sector really innovate, or is it designed not to?

Based on my research on public sector innovation and experimental spaces, including immigrant integration experiments in Finland, I’ve come to see experimentation not as a luxury but as a necessity for modern government. In this blog post in the first comment, I reflect on lessons from Finland and similar approaches in Denmark, the UK, India, China, and South America, and why experimental spaces are becoming essential for governing complexity.

You've done it. You've run your first experimental cycle,16 weeks of genuine learning that actually taught you something. Your team is energized. The insights are real. And now comes the question that will determine whether this was a flash in the pan or the beginning of something transformative:

How do you spread this capability without killing it through bureaucracy?

You've read about experimental spaces. You understand why they matter. You've probably even convinced leadership that innovation is critical. Now comes the hard part: actually, making it happen.

In today's fast-paced business environment, large organizations face a fundamental tension: they need the stability to scale operations reliably, while maintaining the agility to innovate and adapt.