Off the top of my head:
1) Target signatures will be different. How different depends on how good the threat intelligence has been, which is anyone's guess, really. Might be better, might be worse.
2) Countermeasures (if used) will be different. Again, how different depends on how good the intell is.
3) The overall EM environment will be more dense and complex. You don't see a lot of ABM tests where other AEGIS ships around the SM-3 shooter are simultaneously trying to engage complex air raids with heavy jamming, for example.
4) It's likely that there will be more targets to handle (even the best tests I've seen are pairs, not four or five inbounds at once).
5) Warning times will likely be shorter (in testing, you know the general window, at least; an operational attack might occur with little warning at all).
6) Related to that, the system will have been running at sea for days or weeks without constant tweaking by a swarm of Tech Reps. It may not be working at the peak of efficiency.
7) Also, stress levels among operators will be radically higher, which can lead to operator error. Again, the fact that other things will likely be happening at the same time can be a huge factor -- if the CIC has been busy wrangling an ASW threat (for example), it take time and effort to shift gears.
All of these factors make a live engagement different from a test. They don't mean the system won't or can't work, but to pretend that there's no difference just flies in the face of decades of experience.