Exploratory usability test [on hold]

I have a nearly finished prototype app, but would like to see if there are still things that have been missed. I remember reading somewhere that an exploratory method found more critical errors, because the user has a license to play with the product rather than a rigid task-based scenario.

My question is how do I conduct an exploratory usability test, and more importantly which metrics should I be measuring?

Should this be used at the latter stage of a development process?