Well I think it would vary tank to tank, depending upon the amount of media you're using, the flow rate, and the amount of phosphate present.
The optimum point would be to change the media as soon as it becomes exhausted, and no longer can remove PO4. The goal would be to replace it as soon as it exhausts, so you would see a steady decline in the PO4 values. Rather than the media exhausting, which would then allow PO4 to begin to accumulate again prior to changing the media... then once you change it, you would need to remove the excess that had accumulated once the media exhausted before you could begin hitting the bottom line again... sort of the 2 step forward 1 step back deal.
I believe you can test for this by testing the detectable PO4 value in the reactor's output... the thought is that if you are running the flow through the reactor properly the effluent should not have any PO4 in it. So once you detect it in a test, the media has become saturated, and needs to be changed. However with the inability of most of the available test kits to test accurately down to the lower levels (.03 and below) hitting the sweet spot is a little harder, but it may get you in the ballpark... and probably a heck of a lot closer than setting random frame of time for a changeout.
I'll try and locate a thread I read a LONG time ago, where a guy was testing the hourly impact GFO had on his tank, and how quickly the media exhausted.
Also, I believe you should expect to replace the media faster in the first few weeks/months that you begin running GFO.