The difficulty with getting freemium right is in having a free product that's sufficiently useful to have people take the time to use it and a paid version that is worth paying for. Convincing someone to switch from paying $0 and $1 is hard, much harder than $1 to $10. If there's not enough value in the paid version, why switch? Finding that killer feature that will cause people to pay without chasing off the free us…
For example, I run a data visualisation SaaS app. For a long time, I tried all sorts of combinations of free and premium, but no one was upgrading. When I spoke to my users, the most serious ones kept saying they would be happy pay for the free version itself because it solves a big problem for them.
I ignored them because how could I have any pricing model other than freemium, right? But, I kept getting the same feedback consistently.
And then it hit me - the most successful freemium companies give away the most valuable core part of their product for free.
Once they've got people hooked, they can sell premium services so that their users can get even more of the core value.
For example, the value of Dropbox is not lots of disk space. It's the hassle-free file sync, backups and easy file sharing with friends and family. 2GB is more than enough for most. But you can pay a bit to buy more space to get even more of those core benefits.
You pay for something mundane to get more of the priceless.
But, like I said in the other comment, the catch is that you have to have deep pockets to survive long enough to get to critical mass. Some people may get lucky and bootstrap a freemium business quickly, but the odds are low.
And that's why I settled on doing away with freemium and just straight selling a product with obvious value to a good number of people.