Before you can determine if a process is capable of meeting specifications, you must first prove it's stable. Think of it like trying to measure how fast a car can go. If the car is sputtering, stalling, and lurching unpredictably, any speed reading you take is meaningless. You first need a smooth, stable engine. In statistical process control, a stable process is one that is predictable, exhibiting only common cause variation—the natural, inherent noise in the system.

Process capability analysis, which gives us metrics like Cp and Cpk, assumes this stability. If your process has special cause variation—unpredictable events like a tool breaking, a bad batch of material, or an operator error—any capability index you calculate will be misleading. It’s a snapshot of a chaotic system, not a reliable measure of its potential. Our first job, therefore, is to hunt for and eliminate these special causes using control charts.

Minitab offers a suite of control charts under Stat > Control Charts, but for continuous data, our choice often boils down to two main types, based on how we collect our data.

When your data consists of single, individual observations taken over time, the Individuals and Moving Range (I-MR) chart is your tool. This is common in situations where measurements are expensive, destructive, or have a long cycle time. For example, you might measure the viscosity of a chemical batch once every hour. Each measurement is a standalone data point.

The I-MR chart is actually two charts in one:

1. Individuals (I) Chart: This plots each individual data point. It helps you see shifts and trends in the process average.
2. Moving Range (MR) Chart: This plots the range (the difference) between consecutive data points. It monitors the variation or consistency between measurements.

When you can collect data in small, logical groups, or subgroups, the Xbar-R chart is more powerful. For instance, you might pull 5 consecutive parts off a production line every 30 minutes. That group of 5 is a subgroup.

This chart also has two components:

1. Xbar ($\bar{x}$) Chart: This plots the average of each subgroup. It monitors the variation between the subgroup averages.
2. Range (R) Chart: This plots the range (maximum - minimum) within each subgroup. It monitors the variation within the subgroups.

The logic here is that items produced close together in time should be very similar. The variation within a subgroup represents the natural, short-term process variation. The variation between the subgroup averages shows how the process is changing over time. Always examine the R chart first. If the within-subgroup variation is unstable, the control limits on the Xbar chart will be unreliable.

Once you've plotted your data, Minitab automatically applies a series of tests to detect special cause variation patterns. These are the signals that your process is out of statistical control. While you can configure which tests to run (Options > Tests), Minitab's default set is a great starting point. Let's look at the most common signals.

If any of these signals appear, your process is not stable. The work isn't to just note the signal, but to investigate and eliminate the root cause. Was there a power surge? Did a new operator start their shift? Did a tool wear out? Only after these special causes are removed and the process runs predictably can you proceed to a meaningful capability analysis.

Checking for stability isn't just a preliminary step; it is the foundation of effective process control and improvement. An unstable process will always yield unreliable capability metrics, sending you on a wild goose chase. First, achieve stability. Then, and only then, assess capability.