Also a big fan of the "desktop version" test, because it is triggered by the action of a user trying to get the desktop version as opposed to thinking they are

Also a big fan of the "desktop version" test, because it is triggered by the action of a user trying to get the desktop version as opposed to thinking they are about to answer a poll. To that end, I wouldn't trust the data you would get if you asked people to state what they would be willing to pay. Self-reported data out of context is notoriously unreliable. Instead, really put it to the test. Tell people that you'll develop a desktop version if enough pledges are received, and ask people to make a binding pledge in advance. Give them a discount for jumping in early, of course, but use the average pledge as a guideline for final pricing. If the median pledge amount is, say, $10, then you could probably assume that people would be willing to pay $15 once the product was actually available for download. (Those are made-up numbers, but you get the idea.)