Unfortunately what is really needed here is a suggestion of something else to measure that is better. I fight this a lot with bad metrics (code test coverage), nobody wants to give them up unless I can suggest a replacement to measure.