Many software applications continue to grow in terms of the number of features they offer. Reduced-functionality interfaces have been proposed as a solution by several researchers, but evaluations have been limited in number and scope. We argue that traditional performance measures are not sufficient for these interfaces, so we introduce and distinguish feature findability and feature awareness measures. We have conducted a controlled study that demonstrates the tradeoff between these two measures: findability in a minimal layered approach was better than in the full interface alone, but subjects were more aware of advanced features if they used the full interface from the outset. A marked layered approach was also evaluated, but provided little benefit over the other interfaces. Ours is also the first experiment comparing more than one multi-layer approach to a control interface.