Screen Potential Causes

You have chosen whether the customer issue is cycle time or VTW. Part of this phase is understanding the variation in customer demand. This variation can be handled either by leveling the inventory or production. Causes of large span can be tricky to find if individuals are shuffling priorities in their input queues. If there is a reason why the business process should be stratified for different service lines or customer priorities, make sure that the business process behaves that way. If there is a source of variation and you expect it, then things are working fine: anything else is unexpected and should be investigated. When you have identified areas that require improvement, then the application of the 5Ss will usually result in a quicker process. The 3Ms require a balance between reduction of waste without overburdening the workers or creating unevenness in workload. Focusing on only one of them will create problems with the others.

Discover Variable Relationships

By discovering and documenting the relationships between the important drivers of process variation and the output, you should be able to tabulate equations predicting the consequences of changing your process in particular ways. This prediction can be very complex for typical transactional process and will require simulation of the process to both identify key areas for leverage and assessing the magnitude of the effect. Transactional processes show very non linear behaviour, so consider using DOE to investigate a wide range and combination of strategies.

Choose the Best Improvement Strategy

It is a mistake to focus on only the one metric of either cycle time or VTW. You will have multiple process outputs for your different process stakeholders (CTQs and CTPs). These have been documented during the Measure phase and will be used here to score different Improvement strategies. This phase is where you must ask management and the process owners to make a commitment of time, training, job reallocation, and process changes. These changes are not going to occur just because it will have an impact on a single metric. All the different aspects of the proposed solutions will be considered before making any changes.
If the project is piloted, then make sure the pilot is as close to the real process as is possible. In order to prove the improve consider simultaneously collecting data from both the new process and old process during the pilot phase.