Link Details

We use the same concept in unit, qa and release testing, just with different maximum values.
Unit has a max of 5, QA at 10, Release at 25.

Anything over a 3 point issue doesn't move on from unit testing at all, and a sum over fifth of maximum means the entire project doesn't move on for QA and release, leaves us with estetical issues instead of functional ones.

We have no public betas, although we do have some clients who are willing to do private betas for lower licensing, they in effect becomg paying QA staff.

Since unit has such a low allowance most issues are caught by the developers, as they should be, and you don't do signoffs yourself so nothing moves to QA until a peer test occurs, each developer does a complete checkout/build and runs through everything, entering bugs and points, your final bug value is the average of the points for each bug, tends to help offset some of the overly anal and overly complacent devs.

Build day is a slow day, every Wednesday, weekly dev meeting, and a bunch of group testing, but since its in the middle of the week it means it happens almost every week, where a monday or friday it wouldn't.

We get dinged for any release where a bug shows up over a 7 pointer on your scale, after public release.
I've been working on a draft for bonuses where a public release happens and has no bugs over 1 pointers for a set period of time, but upper management likes whips more than candy apparently.