It's not that big a deal but that doesn't mean it's easy. Because from what I've seen most organizations cannot figure this one out.
- Single-loop learning is when you improve in the EXISTING system you've set up. If you do dumb, unnecessary, duplicative, inefficient things -- you do them faster.
- Double-loop learning is when you ask WHY am I doing these things in the first place. What are my assumptions, that lead me to act like this? Because if they make no sense, I should stop doing them.
The single vs. double-loop learning distinction was introduced by Chris Argyris in the '70s. See article in the Harvard Business Review, or simply Google "double-loop learning."
Unfortunately it usually takes a crisis to make an organization confront its own faulty assumptions. But it does not have to be that way. All you need to do is read, noting the gap between external coverage of the organization, and internal talk.
This is actually part of a communicator's job, although we don't often hear about it -- not only representing the organization to the outside world, but representing the outside viewpoint to internal constituents.