Automated configuration of neighbor cell lists, the so-called Automatic Neighbor Relation (ANR) function, is one of the first SON features being deployed in commercial networks. From the operators' point of view, it is beneficial to know how many ANR enabled UEs should be activated to help the full ANR list configuration. In other words, it is important to predict, given a fixed percentage of ANR enabled UEs, how much time is needed to finish the establishment of neighbor relation list. In this work, we defined an ANR penetration probability prediction method and use this method to, calculate the probability of UE detecting a neighbor relationship based on the number of ANR capable UE number and other related network parameters which can be obtained from the network operators. Two prediction cases using this method are discussed and we use simulation to validate the prediction results.