Qhubio
    Capability Interpretation

    Process Capability Index Explained

    3 min read Last updated

    A process capability index is a single number that compares the natural variation of a process to the engineering specification. The four most-used indices are Cp, Cpk, Pp, and Ppk.

    The engineering question this page answers

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

    What it means

    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.

    Why it matters

    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.

    Decision logic

    Is the customer requirement explicit about the index? → Report exactly that ↓ Is the study a machine capability (single setup, short run)? → Cmk ↓ Is the study an initial / PPAP study? → Cp and Cpk (within-subgroup σ) ↓ Is the study ongoing production? → Pp and Ppk (overall σ) ↓ Does deviation from target matter (not just spec limits)? → add Cpm ↓ Is the distribution non-normal? → percentile-based (Ppk-nn) with method disclosed

    Examples

    • 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).

    Engineering procedure

    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.

    Typical failure modes

    • 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.

    Engineering insight

    • 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.

    When NOT to use this metric

    • 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.

    Relationship to other capability metrics

    • 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.

    Engineering notes

    • 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.

    Continue the investigation

    For the difference between short-term and long-term indices, read Cpk vs Ppk and Cp vs Pp. For the physical study procedure, see Process Capability Analysis.

    Verification checklist

    • 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

    Assumptions and applicability

    • 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.

    Sources and engineering references

    External engineering references used for this page. Qhubio applies these references to the practical guidance above.

    Frequently asked questions