One of my professors in college was telling stories about finding out the reason yield dropped after lunchtime was someone eating salty potato chips and then going back to work.

The motivation seems easy though, wafers cost a lot of money! No one wants to see low yields. They are constantly testing and measuring everything to try and find root causes. Look at some of these diagrams if something can be isolated they try to do it.

You run split experiments over subsections of the line to find the source. It's essentially like binary search. You can do this pretty quickly with a skilled team.