Process Capability Index Explained

Which capability index should I actually report for this characteristic — and why?

Capability indices normalize process performance to the tolerance width. For a stable, centered, normal process, a value of 1.00 means the ±3σ natural spread reaches the specification limits and predicts about 2,700 PPM outside both limits. A value near 1.33 places the limits at roughly ±4σ when the process is centered.

Indices let engineers, customers, and management compare very different processes on the same scale. They are widely used in automotive and other manufacturing systems, but the required index, study design, and acceptance criterion must come from the applicable customer or internal requirement.

• Cp — short-term potential (within-subgroup sigma).
• Cpk — short-term actual (includes centering).
• Pp — long-term potential (overall sigma).
• Ppk — long-term actual (overall sigma + centering).

1. Read the Control Plan and CSR to confirm the required index and study window.
2. Confirm characteristic type: two-sided, one-sided, or target-based.
3. Confirm stability with an appropriate control chart before computing any index.
4. Select σ estimator: within-subgroup (Cp/Cpk) vs overall (Pp/Ppk).
5. Report the index paired with sample size, sampling plan, study window, and MSA status.
6. State the distribution assumption and the normality evidence.
7. Cross-check Cpk vs Ppk; a large gap signals long-term drift or between-subgroup variation.

• Reporting Cpk on a Ppk sample plan (or vice versa) without disclosure.
• Pooling data from multiple streams to inflate sample size.
• Applying two-sided PPM formulas to one-sided characteristics.
• Using a 1.5σ shift convention without stating it explicitly.

• The index is metadata about the study, not a property of the process alone. Reporting Cpk without σ method, subgrouping, and window is not auditable.
• Cp and Pp compare potential; Cpk and Ppk compare delivered performance. Mixing them across reports is the most common PPAP finding.
• An index computed from 30 samples over one hour cannot be compared to one computed from 300 samples over two weeks — even if the number is identical.

• Attribute data, destructive tests, unstable processes, or when MSA is unacceptable.
• Characteristics without a valid engineering specification (e.g., aesthetic-only).
• Very small samples (n < 30) — report with confidence intervals or defer to a full study.

• Cp/Cpk vs Pp/Ppk: short-term potential vs long-term actual.
• Cpk vs Sigma level: Sigma ≈ 3 × Cpk when using nearest-limit convention.
• Cpk vs PPM: PPM(nearest tail) = 1 − Φ(3·Cpk); doubles for two-sided centered.
• Cpm vs Cpk: Cpm penalizes off-target operation; Cpk only penalizes proximity to a limit.

Use the free Process Capability Calculator to compute Cp, Cpk, Pp, Ppk, sigma level, and predicted PPM from your measurement data — no sign-up required.

• Never report an index without its σ estimator and study window in the same table.
• Confidence intervals on Cpk are wide for small n — a point estimate is not a decision.

• ✓Required index taken from the Control Plan or CSR, not assumed
• ✓σ estimator disclosed (R̄/d₂, s̄/c₄, pooled, or overall)
• ✓Sample size, sampling plan and study window stated
• ✓MSA result attached to the report
• ✓Distribution assumption explicit
• ✓One-sided vs two-sided handled correctly

• Process condition: statistical stability is required.
• Distribution assumption: use a distribution model justified for the data.
• Confirm process stability and measurement-system adequacy before interpreting a capability index.
• Use a justified distribution model or non-normal method when the normal model is unsuitable.