I log them in a small script that appends each error to a text file. Then I parse it to see trends. I don’t bother with fancy dashboards; just enough data to see if a change caused more failures. It keeps the focus on fixing, not on bragging about metrics.