I have set and climbed at gyms a-plenty where they've used the V-grading system, the category style (easy/medium/hard or R/I/A/O), and a few even leaves their problems ungraded. I'm not going to go into what works better because there are obviously pros and cons for all. What it really comes down to is subjectivity and ego-stroking, and realize that you can't make everyone happy. With that said, this is what I experimented with and worked best.
After problems are set, the boulder problems are tagged in categories of R, I, A, O (Recreational, Intermediate, Advanced, Open). On the card also in the upper right-hand corner is a section for "suggested grade", the keyword there be "suggested". Once the boulder problems are tagged and categorized, I leave the problems ungraded for a few days to get a better consensus from locals. After that, I then grade it to as close to an accurate grade as possible.
This works well because by categorizing the problems first without grading it, it gives your climbers a general idea in terms of difficulty, and motivation to actually try out the problem without having to worry about and fight with numbers (you'll be surprised how many people are shyed away from even trying out new problems by simply looking at the V-grade). And by putting a grade in the "suggested grade" box a few days later, it solidifies the problem so to speak, giving those that want a more definite number on the grade an answer. Also, by putting it as a "suggested grade", it still leaves room for subjectivity (and put off any grade-chasing douchebags) - hardly anyone is going to argue over a "suggested grade".