Removing measurement system error from total observed error can be what I call a “quick win”. Who couldn’t use a quick win these days? If you look at a process distribution, we ask ourselves, what percent of this variation is assigned to measurement system variation?