Query optimization is relatively easy when you look at a plan and find table scans, hash match joins, and missing indexes. But how do you find opportunities when all the low hanging fruit has been picked? This session is a case study on improving a procedure that runs 350 million times per day, so any improvement is greatly magnified. We will discuss where we found opportunities to improve a procedure running in 3.1 milliseconds; how effective our attempts were, what didn't work, and the results achieved.