There is much discussion about continuous improvement, kaizen and operational excellence pursuits in operations management. Many times, the topic of process improvement neglects the more important topic of how to change a process. While there is a desire to avoid stifling creativity among the best and brightest within an organization, a process change without process discipline can become the opening of Pandora’s Box and lead to degraded performance. Following are eight steps that, if followed, allow us to properly change a process with discipline while still embracing the insights and excitement of our rank-and-file operators.

1. Know the Current Process

One of the most important questions to ask at this stage is “Why is the process in its current state?” All too often, we see individuals try to drive change without understanding what caused the process to become the way it is. It is a sign of lazy analytics if the change proposer assumes the process is flawed because individuals who designed it were lacking in process design knowledge. Perhaps the changes we seek are only possible because of new technology or change in supplier. Regardless of the reasons, we need to understand the lineage of the process to avoid past mistakes. Additionally, we need a readily communicable model of the current process. If the organization is ISO-certified and fully compliant, a good documentation of the process should be readily available.

2. Know Why We Wish to Change the Process

The typical reason for process change is either cost reduction or variation reduction. For cost-reduction changes, a good cost deployment is essential. For variation reduction, the change agent should know whether random variation or special-cause variation (or both) is to be eliminated. We can typically erase special-cause variation with rifle-shot solutions (turn off the special cause and we turn off the variation). For random variation, typically increased capital expenditure (except in the cases of very sloppy processes) is needed to decrease variation.

3. Clearly Identify the Change to be Made

Once the current process is found to be deficient in some way, the new process should be articulated clearly and concisely. This may be in the form of a text-based document, a flow chart or other organization-appropriate form of documentation. Sample testing of the documentation to ensure clarity is important. Can operators look at the document and explain back to you how the new process will work? Do several operators understand the documentation the same way, or do they interpret it differently? Documentation should lead to clear understanding with little to no operator-to-operator interpretation variation.

4. Obtain Feedback and Buy-in from All Affected Stakeholders

In a forum of your choice (town hall meeting, posting with opportunity for anonymous feedback, etc.), the new process should be expressed to all individuals who are affected by the change. This includes all individuals who provide an input to the process as well as those who receive an output from the process. Will they be affected in a meaningful way? Will the change cause unanticipated results in their portions of the process? Will this process change add value (or eliminate value-destroyers) for the customer?

5. Revalidate Process Discipline, Data and Measuring Systems for the Change

At all times, we should be certain that processes are followed, data are reliable and our measuring systems are capable of providing data we can use to make good managerial decisions. That said, once we “shine the management light” on the process we are working to improve, change can occur that makes it necessary to re-validate our behaviors, methods and measurements.

Before driving the change in our process, we must ensure that the process that’s in place is adhered to properly. This alone can drive significant reduction in random variation. Once we’re certain of strong process discipline, we should measure the process once again and ensure that we are obtaining reliable data from our measuring systems. If necessary, a good gage R&R should be conducted to ensure quality and reliability of data.