I've found "brown M&M" tests are good for telling me how much I need to investigate things. For example, let's say your job includes verifying data submitted b

I've found "brown M&M" tests are good for telling me how much I need to investigate things. For example, let's say your job includes verifying data submitted by hundreds of hospitals. You don't have the time to do deep investigations for all of them, so you use a script to look for red flags (e.g., discharge date preceding admission date). If there are no red flags, you still quickly look over a small sample. If there are red flags, now you're going to spend a lot of time looking for any errors and their causes. And it's not just to save time. You're only human, so trying to be thorough with every submission will likely result in you getting sloppy because all those reviews of good submissions gives you a false sense of security.