-Steve Blank claims you should close sales with wireframes and presentations before they are developed (particularly in a B2B context) in his customer development methodology.
-Dropbox produced a video that went viral explaining their product. They used this to produce a large list of emails of potential users.
-Paul Howe, founder of blue spark, used a $40 greasemonkey script to simulate his product (it put fake posts in user's facebook stream) then recruited people via craigslist for a social media focus group. His conclusion was the people hated it, and he changed course.
-Food on the Table did the tasks for their early customers by hand. Rather than creating web software, they implemented the service in person.
-Aardvark, the question site, used what people call the wizard of Oz approach. They build software to collect the questions, but provided results manually.
Fake it before you make it ideas don't have to be for the initial product. Numerous people suggest that once you have some traction you can introduce features as links to see if people even click on them. If the users do click on them present them with a form saying the feature is in development and they can choose to be notified when it's completed. If few people click it, don't implement it. Also, you don't have to show the feature to all your customers.
The underlying principle is that the biggest risk is making things that people don't want. So the measure of efficiency is how quickly you can determine if people are interested in your product or feature.