I'm working on a project where we have the opportunity to test an upgraded version of a system with 4 users. The new version isn't radically different from the current version, but there are some changed task flows and added functionality users have been asking for. We want to test to find any usability problems with a few critical and frequent tasks we've identified - tasks that all users in this user group perform.
Our user base is broad and consists of users who will be completely new to the system (have not used a previous version), all the way out to highly-skilled users who have a lot of experience on the current system version.
I'm wondering what experience level of users we ought to target for our test: Do we test with 4 users whose current system version experience/expertise level is "just below average", or perhaps one from each quartile of the experience/expertise spectrum?
And a related question: if the ax comes down on the project and we have the opportunity to test with only one user, what experience/expertise level should we target?