The Pigeon Test is a companion to the book Why Change Fails by Mark Eddleston. It helps you tell apart three kinds of situation: simple, where cause and effect hold still and the rules settle it; complicated, where they still hold but it takes someone qualified to trace them; and complex, where they only line up looking back, and no plan survives contact.
It is a crude tool and it can be wrong. Treat the verdict as a way to look more closely, not a ruling.
Taken as a whole, an organisation is always complex, because it is made of people interacting constantly. The three states are states a complex system can be held in, not three kinds of organisation. A bureaucracy is not a simple organisation, it is a complex one held still. So the tool does not read your organisation, it reads one situation inside it, and that situation can be any of the three. Some come apart into pieces that are not the same. A plannable part inside a complex one is normal rather than exotic: the payroll run inside the restructure, the data migration inside the culture change.
This tool gives you a single read. Where a situation does come apart, that gets flattened into one answer, which is one of the ways a crude tool is crude. If part of what you described does not fit the verdict, that part deserves more of your attention, not less. Be wary of pulling a situation apart too eagerly, though. What makes something complex usually lives between the pieces rather than inside them, so a part only separates cleanly if it would behave the same way lifted out.
The skill isn't having better answers. It's knowing which kind of problem you're facing before you reach for any answer at all. Samir Rihani, whose work the book builds on, put it plainly: before we rush in with solutions, we have to determine first the nature of the system in question. This tool exists to help with that first step.
Nothing you type here leaves your device. The scenario, the sliders, the verdict: all of it stays in your browser. Two things are kept, both on your own device: a count of how many times you have used the tool, to adjust how much guidance is shown, and an anonymous copy of each reading.
A later version may, with your clear consent, collect anonymised readings to improve the tool. This version does not.