This paper describes principles of a novel two-level multi-output Boolean minimizer FC-Min, namely its Find Coverage phase. The problem of Boolean minimization is approached in a reverse way than common minimizers do. First, the cover of the on-set is found, and after that the appropriate implicants are being constructed to satisfy this cover. Thus, only the necessary group implicants are being generated, which makes FC-Min an extremely fast and efficient minimizer for functions with many output variables. An essential phase of the algorithm is the Find Coverage procedure. This phase determines the number of terms in the final solution, which has to be reduced to minimum. It solves an NP-hard problem, thus some heuristic has to be applied. We propose our heuristic method to solve this problem and study the influence of parameters on the final solution quality and runtime.