Unit testing is typically performed by the developer who writes(or modifies)a subroutine,procedure,module,class,or other building blocks of an application.The purpose of this testing is to be sure the unit correctlyperforms its specified functions.
This 90% pass rate thing kinda bothers me. Do you mean to tell me that 10% of your applications are KNOWN not to work?
I can see beginning testing with some unit tests failing but I can't imagine going to production with them failing. Either the tests are bad and need to be rewritten or removed or the non-test code has to be made to comply.
Many applications do go to production with known errors, these, at least in my company, are deemed as acceptable and are identified as "features" (until there are too many Help Desk complaints) or as future fixes.
I have not failed. I've just found 10,000 ways that won't work" --Thomas Edison