There are two types of collusion which can harm Clients. This paper proposes two methods of detecting collusion’s and estimation a clients’ potential losses. The first is based on officially recommended factors for collusion detecting. The second one utilizes MLP artificial neural networks. Results are compared and discussed.