The % model would indeed work better with a low volume sales. So, if I wanted to distribute a file just once or twice, then this would work, since I end up paying just 50c (if the price of the digital goods were $10) instead of a fixed price for the subscription.

But if someone intends to sell several hundreds of copies, or the price is bigger (say, 5% of $10,000), then it is not a good deal.

So, I really only see the target audience as someone that must have DropBox integration (though other services also have integration with your webserver, and mounting a DropBox folder and exposing that through the webserver seems trivial), or maybe people with such a dispersion of files that the other services won't work (they allow you a fixed number of products).

while true, it seems if you are selling 0 for too long, you just take it down. i'd rather plan for success than failure, and $5 is pretty trivially worth it to test a product out.