The study of retrial queuing systems presents great analytical difficulties. Detailed results are available for some models, whereas for other models the obtained results revealed poor information and are cumbersome (they contain Laplace transforms, integral expressions, etc.). Therefore, in practice, they present limited performance. Often, to overcome this difficulty, we use an approach based on the stochastic decomposition property that can be possessed by the model. It offers the advantages of simplification of solving complex models. This paper deals with the stochastic decomposition property of an M$^{X}$/G/1 retrial queue with impatient customers and exponential retrial times and of an M/G/1 retrial system with feedback and general retrial times.