A usage-based trial is really just the same thing as a time-based trial. There is a certain pace at which users are going to use up their usage quota, and that translates to a time period. Generally, though, you want the time period to end after the user has seen the benefits and before they have exhausted their initial enthusiasm or encountered any bugs. Giving a user four months to make their choice probably puts you on the wrong end of all those considerations.

After the time-based trial, there are going to be different packages based on different usage quotas anyway. KISSmetrics is probably catering to very different and more established customers than the author, and their pricing model is structured accordingly.