What Is Process Capability? Definition and Uses

Process capability is the comparison between what a process naturally produces and what the specification demands — the voice of the process set against the voice of the customer. A capable process fits its output comfortably inside the specification limits; an incapable one ships defects even on its best-behaved days.

How it works

Capability analysis has a prerequisite: stability. A process must first be predictable — shown by a control chart to vary within consistent limits — before its capability means anything, because capability is a prediction about the future, not a summary of the past. Once stability is established, capability indices compress the comparison into single numbers: Cp measures whether the process spread fits the specification width, and Cpk adjusts for whether the process is centered. Long-term counterparts, Pp and Ppk, tell the same story using overall variation.

The output of a capability study is a decision: run the process as is, re-center it, reduce its variation, or renegotiate the specification.

Capability also gives specification conversations teeth. When a process is stable, well centered, and still incapable, the study documents that the specification and the process are simply mismatched — and that delivering conforming output will require investment, not exhortation. Many long-running arguments between operations and engineering end the day someone finally runs the study.

A worked mini-example

An illustrative machine shop turns shafts to a diameter specification. A control chart kept over several weeks shows the process is stable. The capability study then compares the process spread against the tolerance and finds that the spread fits, but the process average sits close to the upper limit — capable in width, poorly centered. The remedy is not new equipment; it is an adjustment that moves the average toward the middle of the tolerance. Without the study, the shop would have kept scrapping parts and blaming the machine.

• Establish stability first; capability figures from an unstable process are fiction
• Validate the measurement system before the study — gauge error masquerades as process spread
• Check the normality assumption behind the standard indices, or use methods that fit the data
• Read Cp and Cpk together; the gap between them is the centering problem
• Recompute after any process change — capability is not a permanent property

Capability is a prediction, and only stable processes permit predictions.

Capability studies with Cp and Cpk are Green Belt material in our curriculum, taught inside the Measure and Control phases of the simulated project. Black Belt extends the topic to non-normal data and the advanced statistics that rigorous capability work demands.