The first of these steps is to verify that the simulation is running in the same way that the real world would work in the same circumstances. In theory there is a need to do this every time you make a change to the simulation, but clearly this would be impractical. If the simulation is of an existing factory (or another system that actually exists) then this verification stage is just a matter of ensuring the simulation behaves as the real factory currently does. So you can put some actually customer orders (for example) into the simulation and see if they are handled in the same way and in the same time as they are in the real factory.
The only way to verify a simulation of a new factory (that does not yet exist) is to carefully experiment with the simulation, watching (both the animated screen and the performance measures) to see if the changes you see are reasonable in the light of changes you make to the input to the simulation (such as demand, numbers and speeds of machines etc.). There is a methodological problem here because clearly if these changes were known, or expected, there would be no need to build and use the simulation in the first place. Nevertheless, careful examination of the model's behavior will allow you to see and remove most mistakes or misunderstandings in the simulation building.
It is very easy to omit this stage in the excitement and urgency of trying to help your client find a solution. Omitting it is also one of the easiest ways to loose a client's faith in your work when they see the simulation show some result that is not only intuitively wrong but proves to be caused by a mistake!