The first step is creating a clear, well-understood process. Each of the steps and the order of the steps are clear and consistent. Where it makes sense, each of the steps is measurable and metrics are built into the step.
The second step is to execute the business process and collect the measurements. In order to execute the business process, the inputs to the business process need to be found or created. Running a process in a hypothetical way is always different than running a process in the real world.
Based on the measurements and metrics, the process is refined and the whole process is repeated.
One of the keys is that even though the initial process isn’t the most efficient or elegant process, it is a 100% complete process. If the process is not complete, you are continually running workarounds and fixes. These workarounds and fixes then invalidate the measurements and there for the refinements.
It is the same for a business. You cannot create half a business and hope to iterate yourself to success. You need to create a complete business.
For those who would argue against that, saying that you can never be perfect, I agree. But that imperfection needs to be built into the process, not ignored and dealt with as the problems come up. You can have steps in the process the encapsulates the uncertainty and resolves it. By having a business process it doesn’t mean all the steps have to be elegant. It is fine to have a process that takes in certain inputs and then within that step does whatever is required to produce the outputs of that step. When done this way you now know what went into the step, notes on what happened within the step, and how long the step is taking. This is plenty of solid information to refine the step in the context of the whole process.