Process Capability (Cp, Cpk, Pp, Ppk): Formulas, Interpretation, and When to Use Each

The Question Capability Answers

A control chart tells you whether your process is stable. But stability alone is not enough. A process can be perfectly stable and still produce parts outside the specification limits. You need a way to measure how much room your process has within those limits.

That is what process capability indices do. They compare the width of your specification tolerance to the width of your process variation. The result is a single number that tells you whether the process comfortably fits within the spec, barely fits, or does not fit at all.

Cp: Process Capability (Centered)

Cp measures the potential capability of a process, assuming it is perfectly centered between the specification limits. It compares the total tolerance width to the process spread.

Formula: Cp = (USL – LSL) / (6 x sigma-within)

Where USL is the Upper Specification Limit, LSL is the Lower Specification Limit, and sigma-within is the within-subgroup standard deviation (estimated from the R-bar or S-bar from your control chart).

Interpretation: A Cp of 1.0 means the process spread exactly fills the specification width. There is no room for error. A Cp of 1.33 means the specification is 33% wider than the process, providing a reasonable safety margin. A Cp of 2.0 means the specification is twice the process spread, which is excellent.

Limitation: Cp does not account for where the process is centered. A process with Cp = 2.0 but centered far from the target could still produce defects. That is why Cp alone is never sufficient.

Cpk: Process Capability (Adjusted for Centering)

Cpk measures actual capability by accounting for how far the process mean is from the nearest specification limit. It penalizes the Cp value for off-center processes.

Formula: Cpk = minimum of [(USL – X-bar) / (3 x sigma-within)] or [(X-bar – LSL) / (3 x sigma-within)]

Where X-bar is the process mean.

Interpretation: Cpk takes the worst-case side. If the process is shifted toward the upper spec limit, Cpk reflects the capability against that limit. Cpk can never exceed Cp. When Cpk equals Cp, the process is perfectly centered.

In practice: Cpk is the number most customers and quality standards require. When an automotive OEM asks for “Cpk of 1.33 minimum,” they want proof that the process comfortably meets specifications even accounting for any centering offset.

Pp: Process Performance (Centered)

Pp is the long-term equivalent of Cp. It uses the same formula but with a different standard deviation.

Formula: Pp = (USL – LSL) / (6 x sigma-overall)

Where sigma-overall is the overall standard deviation calculated from all individual data points, not from within-subgroup estimates.

Key difference from Cp: Cp uses within-subgroup variation (sigma-within), which excludes between-subgroup shifts. Pp uses overall variation (sigma-overall), which includes all sources of variation: within-subgroup, between-subgroup, shifts over time, and any other factors present in the data.

Because sigma-overall is always equal to or larger than sigma-within, Pp is always equal to or less than Cp for the same data set.

Ppk: Process Performance (Adjusted for Centering)

Ppk is the long-term equivalent of Cpk. It accounts for both the total variation in the data and the centering of the process.

Formula: Ppk = minimum of [(USL – X-bar) / (3 x sigma-overall)] or [(X-bar – LSL) / (3 x sigma-overall)]

When to use it: Ppk is typically required for initial process studies, such as those submitted in a PPAP package. When you are evaluating a new process or a process change and do not yet have long-term stability data, Ppk calculated from the initial production run is the standard metric.

Cp/Cpk vs. Pp/Ppk: When to Use Each

This is one of the most common questions in SPC. The distinction comes down to what type of variation you are measuring.

Use Cp and Cpk when you have a process that has been running under stable, controlled conditions and you want to measure its inherent short-term capability. The within-subgroup standard deviation captures the process potential when special causes are absent.

Use Pp and Ppk when you want to evaluate overall process performance including all sources of variation. This is the right choice for initial process studies, PPAP submissions, and any situation where you are looking at total observed variation over a production run.

The gap between Cp and Pp tells you something important. If Cp is much higher than Pp, it means there is significant between-subgroup variation (process shifts, tool wear, material changes) that the within-subgroup estimate does not capture. A large gap is a signal that the process has instability that needs to be addressed.

Acceptable Capability Values

Industry standards and customer requirements define minimum acceptable capability indices. Here are the most common thresholds.

Cpk or Ppk of 1.33: The standard minimum for most automotive production processes. This is the baseline requirement in many Customer-Specific Requirements (CSRs) and PPAP submissions.

Cpk or Ppk of 1.67: Required for safety-related or critical characteristics in many automotive programs. Some OEMs apply this threshold to all significant characteristics.

Cpk or Ppk of 2.0: Considered excellent and sometimes required for the most critical applications, such as safety-critical dimensions or processes with very high consequence of failure.

Cpk below 1.0: The process is not capable of meeting specifications. Parts are being produced outside the tolerance. Immediate action is required.

Always check your specific customer requirements. Some OEMs have unique thresholds or require capability on characteristics that other customers do not.

Prerequisites for Meaningful Capability Analysis

Capability indices are only meaningful if certain conditions are met.

The process must be in statistical control. Calculating Cpk on an out-of-control process produces a number, but that number does not represent the process reliably. Establish control first, then assess capability.

The data must be normally distributed (or at least approximately so). The standard Cp/Cpk formulas assume a normal distribution. If your data is significantly non-normal, you may need alternative capability methods.

The measurement system must be adequate. A gauge with excessive variation inflates the observed process variation and deflates your capability index. Complete a Measurement System Analysis (MSA) before relying on capability results.

Sufficient data is required. For initial process studies, a minimum of 25 subgroups (or approximately 100 individual readings) is the standard recommendation. Fewer data points produce capability estimates with wide confidence intervals.

Common Mistakes in Capability Analysis

Calculating capability without verifying stability. This is the most frequent error. Always confirm statistical control with a control chart before calculating Cp, Cpk, Pp, or Ppk.

Confusing Cp with Cpk. Reporting a Cp of 1.5 without Cpk gives an incomplete picture. A centered process with Cp = 1.5 might have a Cpk of only 0.8 if the mean has shifted. Always report Cpk (or Ppk).

Using the wrong sigma. Mixing up sigma-within and sigma-overall, or using the wrong one for the index you are calculating, produces incorrect results. Know which standard deviation each formula requires.

Ignoring the gap between Cp and Pp. If your Cp is 1.8 and your Pp is 1.1, your process has instability. Address the between-subgroup variation before declaring the process capable.

Frequently Asked Questions