I've often used the "Hundred Dollar Exercise" with groups of folks at SDR's to get a rough sense of how to prioritize features.

This is where you give people a list of possible things you might build and 100 virtual dollars to spend across all of those. The more important you think a feature is, the more money you spend on it. The results aren't really scientific but it's a quick way to get a general sense of how important various things are to customers.

In smaller groups, it's interesting to run the exercise multiple times.

Between each iteration, you can have a discussion about why a person chose to invest more in Feature X vs Feature Y and the rationale behind their decisions. Everyone in the room listens to the arguments, and then you run the exercise again until the everybody's "budgets" agree within some reasonably small epsilon.

This is same consensus-building approach that Wideband Delphi uses for task breakouts and work item estimation, but applied to the domain of relative priority instead of individual task duration. 

The hunch is that the same crowd-based wisdom that helps Wideband Delphi produce accurate estimates will also apply well to other nebulous problem domains like strategy and opportunity analysis.