In this paper , the second in a series of two , we justify two important assumptions on which the result is based that in course of a galaxy merger the slingshot ejection of bound stars is sufficiently efficient to allow a supermassive black hole binary to merge .
A steep cusp with a power law index of 2.5 \text { - - } 3 is required which is as massive as the binary and surrounds the BHs when the binary becomes hard .
This cusp is probably formed when both clusters , surrounding each black hole , merge and combine with the matter funneled into the center .
We find this profile to be in agreement with observed post-merger distributions after the cusp has been destroyed .
The time dependency we derive for the merger predicts that stalled black holes , if they exist at all , will preferably be found in less than \sim 0.2 \mathrm { pc } distance .
To test this prediction we compute the current semimajor axis of 12 candidates of ongoing mergers .
We find all binaries unambiguously to be already in the last phase when they decay due to the emission of gravitational waves .
Therefore , in striking contradiction with predictions of a depleted loss-cone , the abscence of even a single source in the slingshot phase strongly supports our previous and current results : Binaries merge due to slingshot ejection of stars which have been funneled into the central regions in course of a galaxy collision .