A compensation is definitely advisable. But I'm not sure, if it's the best idea, to use antibodies with alternative fluorochromes for it. My suggestion is the use of a spectrum viewer tool (available on web pages of numerous companies like Life Technologies or BD) to calculate the theoretical spill over between the different channels and use this values for compensation.
Of course, you should deal with the compensation. You should use three control to make sure the right compensation. One is cells without any dye, one is cells stained with AnnexinV-FITC only, and one is stained with PI only.