Excel is fantastic for exploring small sets of data... "quick and dirty" stuff. When you want rigorous statistics or a more formal analysis of data, R and friends are far superior. And anything even remotely repetitive should be done in something with a better scripting language. But I'd hate to lose Excel just as much as I'd hate to be forced to use MATLAB or Python to plot results from some small screening experiment.
And of course, we are completely deviating from Excel's forte as a financial tool, where it is much stronger.
Sometimes I'll even use it to clean up data for insertion into a database or some other such task. It has some nice built-in "Filter" functions.