That doesn't mean every observation of every experimental replication will be identical, particularly where stochastic events are an integral part of the experiment. In such cases, the replication may be an attempted confirmation of the principle(s) underlying the experiment. It really all depends on the hypothesis being tested.
For example, if you ran multiple replications of an experiment that has determined that a particular bacterium can evolve resistance to a particular antibiotic, you would not expect them all to take exactly the same time or number of generations or produce identical mutations. In some cases, a colony might not evolve resistance at all. Variability in the results is expected because of the stochastic nature of mutations - so various statistical methods are used for analysing results and comparing results across replications.
This is also why abiogenesis research is not specifically aiming to reproduce the origin of life on Earth, but hoping to find one or more ways in which life could have arisen - there may be more than one way to skin that cat and we don't know the original method - that's why empirical research is continuing.