Rushing to fix problems without good data leads to costly mistakes. Take a scientist's approach: "I don't know—let's find out!"