Continuing from our series last month, I discussed last month how defining the problem is a critical first step when seeking improvement on an internal business problem. I outlined the Define phase and how understanding your customer, your organization’s support team and its process, available resources, and your ultimate goal are all critical in the beginning before peeling the layers and fixing the problem or issue at hand.
Once these areas are defined, you are ready to go to the next step in fixing your problem!
We discussed the DMAIC methodology and its various steps in a business setting:
The second aspect of this phase, Measure, outlines gathering our data and establishes what needs to be measured in the project we’re tackling.
So What Are We Measuring?
There are several areas to understand when we gather our data.
- Units — The smallest point of reference
- Defect — The problem with the product or service that stems from an issue that occurred in the process
- Opportunity — Potential areas in the process where the possibility for a defect presents itself
How does this apply to your business? In the Define phase, you would have established the situation where you can gather this data from. So start by creating a data collection plan.
How Do We Measure Up?
The gathering of data during the Measure phase of the DMAIC process seems straight-forward enough. However, the proper collection of accurate and seamless data will be crucial to the future success of your project. From the data collected, a baseline will be established to measure the performance of your initiative. Cluttered or inaccurate data will have long term damaging effects. As the old saying goes, “Measure twice and cut once.” This should be your mantra moving forward.
Get things started with your teams establishing the current state or baseline. These become the standard of measurement again, which all future performance is measured. Data collected during the Measure phase will be compared to data collected in the Improve phase to confirm results. One of the challenges with data collection becomes translating outcomes into numerical values. In a manufacturing environment, things are pretty clear and easy to understand. Other environments are more complicated and muddled. Keep in mind, in every situation that historical data can be skewed, based on how the data was collected. The historical data will almost surely not have been compiled with the same structures and methodologies as you are creating. Be very cautious with historical data. It is crucial at this point to have a well thought out data collection plan.
Measure Twice, Cut Once
In any valid piece of research, it is crucial to test the methodology or system of measurement before things move forward. Here are four things to consider when testing the methodology of data gathering;
- Repeatability – Can we accurately get the same outcome multiple times, and can we see a comfortable pattern of repeatability?
- Reproducibility – Can it become reproducible over multiple operations?
- Accuracy – Can it be said that there is a noticeable difference between observed average and a known standard value?
- Stability – Can we get the same outcomes from measuring the process every time? Are there external factors that impact reproducibility?
The best way to validate your measurement system is to apply a Gage Repeatability and Reproducibility Study (GR&R). Once you have validated your system, you can proceed to collect your data. The key to success is to stick with your plan and follow the established research practices and methods which were validated.
See discussions on LinkedIn