If in a graph of the type time vs. target function you see a simultaneous sudden drop of both the target function and its free-value counterpart, you may be in business. Wait until the minimisation finishes, and examine the corresponding packing PDB file produced by QS. If this also looks good, go ahead and try some rigid-body refinement with your favorite program. For examples of what "a simultaneous sudden drop" is, please refer to the QS-related papers (they are full of such graphs).