Unless something would be prone to sliding off of the slab etc, I would normally look to slab cracking to be more indicative of ULS failure than deflection. Normally, you'd have a hard time convincing me that a cracked / lightly cracked slab is overstressed solely because it has deflected.
Can you tell if the deflection is a product of curvature along the cantilever length versus backspan induced rotation at the support? Those are different problems.
How sure are you that the deflection has taken place in service rather than at or near the time of construction?
We're speaking of downwards deflection of the cantilever tip, correct?
Is there a concentrated load applied to the end of the cantilever? A supported wall etc?