Freemium maybe a neologism, but the model has been around since the days of shareware, and the free trial concept is prevalent not just in 'web 2.0,' but in everything from end-user software to enterprise software to dev tools.
After all these years, it's interesting to see how far off some vendors end up when they try to decide what exactly they want to give away and why.
Let's look at "why" first: there lots of reasons to offer the free SKU, among them
- The obvious: get users hooked somehow and get them to buy the paid SKU
- Build a really big user base with an appealing free product, never mind how many actually upgrade. Variant (a): your ability to give stuff away is limited by specific costs (e.g., bandwidth, burn). Variant (b): you are intentionally burning investor cash to give away more than you can "afford" to in order to build your user base.
- Build mindshare by giving away something (else) that has exchange value (i.e., is worth real money). This is the most unusual, but examples include sites giving away iTunes song tokens, or software companies giving away a license to a tool or server product if you join one of their sites and do some activities.
Once you know why you're giving something away, you can use that knowledge to inform what and how.
Online, there is a big difference between giving away space (e.g, remote storage of documents) and time (a trial period that ends). If you offer a time-limited trial of an online service, you're discouraging your users from making a commitment of their time or energy, or from storing their valuable data. Why? Because their initial assumption is that if they don't pay, then at the end of the trial period they will not be using your product anymore. Paradoxically, this initial "pessimistic assumption" makes them unlikely to commit enough that they really get hooked and decide the service is worth paying for.
Free space is a much better (and more popular) model: it tells users to relax, get comfortable ... as long as they don't exceed a certain allotment, they can use your service for free forever. Something about free forever makes folks more comfortable. Eventually they really love -- nay, need -- the service, but their free space allotment is full. It's a smaller cognitive leap to toss in a few bucks for some more space, bandwidth, or logins at that point.
There's a big difference between a long trial period, a short trial period, and a "number of uses" period in an offline product. A short trial period makes it difficult for people to settle in before having to take the leap. If they project-switch a lot, they could end up spending a big part of the trial period working on something unrelated and never getting to really eval the product.
A long trial will pick up some casual users who treat it as their own licensed product. Especially if, a couple of months after the trial expires, they can expect to get a beta trial of your next version. A number-of-uses period fixes some of the "short period" problems, but can discourage users from getting comfortable by launching your app just to play around with it.
The idea that the trial period should be purely trial, and not enough for the user to accomplish a real task, is a poor idea. The trial should allow a well-intentioned user to get something done so that he or she can ascertain that the product actually solves a real problem.
There's also a big difference in how you "degrade" functionality for a free version. Online, an ad-encrusted free version is nice because it is very clear what the core product does, even if it may encourage more parasitism than you ultimately want. Removing features, on the other hand, can make the core product look weak and unappealing.
In the offline software world, degradation may take some thought. One time, I was testing video conversion products for Windows. The DirectShow architecture notwithstanding, some software is robust at processing video, even when there are issues in the video or the DShow filters, and some software explodes violently whenever anything unexpected happens. So it's necessary to thoroughly "trial" any app that claims to convert a big range of formats.
Some trial packages added a watermark or logo to the output. Other trials didn't add a watermark, but limited the length of the video clip you could process. Still others had calendar-time-limit features.
Of these approaches, the "limited-length clip" approach was a total loser: I could see the beautiful output quality produced in a few cases, but by making it impossible to convert a lot videos for real viewing in the trial, the authors made sure I couldn't learn whether the app works in the general case.
The watermark apps were somewhat better -- I could use these for a lot of "fooling around" projects and find out if they worked or not. And the calendar-limit apps were the best: within the limit, I used them as though they were my solution ... until one or another demonstrated through failure that it was not.
In the end, the obvious tip is to think hard about your trials. But more to the point: if you're doing free trial or freemium software or services, the free/trial experience should really be designed into the customer lifecycle experience from day one.