We all have hunches: that a new market is about to explode, that there's a problem in finance, that consumer behavior is changing. Most people hold very strong opinions about their hunches. They're either bound to be right (because they're instinctive), or they're bound to be wrong (because they aren't rational). So what should you do with a hunch? Trust it or forget it?

In the early 1980s, Herb Meyer had a hunch that the Soviet Union wasn't as stable as official intelligence suggested. At the time, he was serving as a special assistant to the Director of Central Intelligence and Vice Chair of the CIA's National Intelligence Council. In this capacity, he produced the U.S. National Intelligence Estimates. So he had status and he had a hunch, but what he didn't have was any proof that his hunch was any better than a fantasy.

"Everyone was telling us the Soviet Union was fine," Meyer told me. "Our spies would bring information about how many bombs, tanks—all the data they'd always brought. Ask the same people the same questions, of course you're going to get the same answers."

Meyer decided to do something different. He asked himself a great question: If I were right, what would I expect to see? If the Soviet economy were imploding, what would that look like?

"The critical part," Meyer recalled, "was that I made sure all the information went to one guy. Just one. Then, if nothing came in, I had my answer. And if something does come in, it doesn't get lost in the noise."

One of the first pieces of information that came in was news that a weekly meat train had been hijacked and all the meat stolen. At first the army was called out, but then the Politburo told the army to back off and tell no one. Let the people have the meat.

"Well that's not what happens when everything in the economy's just fine, is it? So that started to tell us something. And then there was more like that."

What's so useful about Meyer's process is that it was low risk. Nothing had to change; no big investment in new people or procedures was required. But he had ensured that weak signals, if there were a lot of them, would add up to something meaningful. 

"If I were right, what would I expect to see?" is one of the great questions not least because one of the characteristics of great leadership is pattern recognition and the ability to spot trends or see change before others do. But it's hard to persuade anyone you're right—or your hunch matters—without the data. If you wait until the data's obvious, you've lost your advantage. But if you test your hunches, you could discover you're smarter than you thought.