A lot. I work with large amounts of data so relatively minor changes in algorithmic complexity can determine if a method is practical or not. Recently I found

A lot. I work with large amounts of data so relatively minor changes in algorithmic complexity can determine if a method is practical or not. Recently I found a great new dataset but it didn't work well with our database - the naive method of doing lookups was at least 10x too slow to be practical. Studying the underlying algorithms of our database helped me find a solution that took up 10x as much space but was 100x faster.