You get it. Sometimes the problems are cool - like some of the financial recon projects I worked on required neat algorithms (not complex or exotic ones, but "fun" ones) to solve something in Python in 1 minute that a R programmer's (they weren't great) script took 10 hours to solve - but most projects are boring implementation wise.
Actually, most of my really, really good and valuable work doesn't amount to any more than comparing Excel files, and giving you a report on that data. Tons of companies need that.