Cp, Cpk, Pp, and Ppk are the most widely used metrics in Six Sigma and SPC — and the most commonly misunderstood, even by experienced quality engineers. The difference is simple: Cp and Cpk measure process capability using short-term (within-subgroup) variation, while Pp and Ppk measure process performance using long-term (overall) variation. All four compare process variation against customer specification limits, but they answer fundamentally different questions.
This article breaks down exactly what sets them apart — the formulas, the sigma calculations, and how to interpret them together.
The Core Distinction: Capability vs Performance
Both metric pairs evaluate how well a process fits within its specification limits (USL and LSL), but they differ in which variation they measure:
- Cp and Cpk measure process capability using within-subgroup variation (short-term variation).
- Pp and Ppk measure process performance using overall variation (long-term variation).
Put simply:
Cp and Cpk ask: What is this process capable of when operating under stable conditions?
Pp and Ppk ask: How is this process actually performing over time?
Cp Cpk vs Pp Ppk Formulas: Almost Identical
Compared side by side, the Cp/Cpk and Pp/Ppk formulas are nearly the same — and that’s intentional. The only meaningful difference is how the standard deviation (σ) is estimated:
- Cp and Cpk use σ within — estimated from within-subgroup variation.
- Pp and Ppk use σ overall — calculated from all measurements combined.
Everything else in the formulas is identical. This single difference in sigma estimation drives every practical difference between the two metric pairs.
Process Capability Formulas (Cp & Cpk)
Where σ within is estimated from subgroup statistics — either the average range (R̄) divided by the control chart constant d₂, or the average subgroup standard deviation (s̄) divided by c₄:
or
Process Performance Formulas (Pp & Ppk)
The numerators and structure are identical across all four formulas. The only difference is the denominator: σ within for capability, σ overall for performance. Where σ overall is the standard sample standard deviation calculated from all individual measurements:
How Cp and Cpk Are Calculated
Subgroup Collection
A Cp/Cpk study requires measurements collected in rational subgroups — the same sampling structure used for an X-bar R control chart. A common industry approach, aligned with AIAG recommendations, is to collect 25 subgroups of 5 samples each, for a total of 125 measurements. Subgroups are typically defined by time, shift, batch, or another meaningful grouping factor.
Sigma Within: Short-Term Variation Only
When calculating σ within, between-subgroup variation is intentionally excluded. The standard deviation is estimated using only the variation observed within each subgroup — typically via the R̄/d₂ method (average range divided by the appropriate control chart constant).
The long-term sources of variation being filtered out include:
- Shift-to-shift differences
- Operator-to-operator variation
- Raw material lot changes
- Machine setup adjustments
- Environmental fluctuations
By removing these long-term sources, Cp and Cpk reveal the best-case potential of the process — what it can achieve when running consistently under a single, stable set of conditions. This is why, even if a process drifts significantly between subgroups, Cp and Cpk can remain high as long as within-subgroup variation stays tight.
How Pp and Ppk Are Calculated
For Pp and Ppk, subgrouping does not affect the sigma estimate. The standard deviation is calculated directly from all measurements combined, using the standard sample standard deviation formula.
Because nothing is filtered out, σ overall captures every real-world source of process variation: different operators, different shifts, raw material lot-to-lot differences, machine wear, seasonal changes, and anything else that influences the process over the full study period.
As a result, Pp and Ppk represent actual process performance as a customer or downstream process would experience it.
Interpreting the Gap Between Cp/Cpk and Pp/Ppk
In practice, Pp and Ppk are almost always lower than Cp and Cpk. This is expected — σ overall is typically larger than σ within, so the performance indices are more conservative. The size of the gap is itself diagnostic:
- A large gap signals significant long-term variation sources — such as lot-to-lot material shifts or operator differences — that within-subgroup analysis cannot see. A control chart combined with Western Electric Rules can help identify when and why these shifts occur.
- A small gap indicates that within-subgroup and overall variation are similar — the process is stable across subgroups, with no dominant long-term effects.
Summary: Cp/Cpk vs Pp/Ppk at a Glance
| Cp & Cpk | Pp & Ppk | |
|---|---|---|
| Variation measured | Within-subgroup (short-term) | Overall (long-term) |
| Sigma method | R̄/d₂ or s̄/c₄ (σ within) | Sample std dev (σ overall) |
| Subgroups required | Yes | No |
| Question answered | What can the process achieve under stable conditions? | How is the process actually performing over time? |
Process Capability (Cp/Cpk) tells you the potential. Process Performance (Pp/Ppk) tells you the reality. Used together in a full process capability analysis, they give you a complete picture of where your process stands and where improvement effort should focus.
Frequently Asked Questions
What is the difference between Cp Cpk and Pp Ppk?
Cp and Cpk measure process capability using within-subgroup (short-term) variation, while Pp and Ppk measure process performance using overall (long-term) variation. The formulas are otherwise identical — only the sigma estimate differs.
Why is Ppk usually lower than Cpk?
Because σ overall includes long-term variation sources — shift changes, material lots, machine wear — that σ within filters out. Since Ppk divides by a larger sigma, it is almost always the more conservative index. The difference between Cp and Cpk within each pair, by contrast, reflects process centering.
What is a good value for Cpk and Ppk?
A common minimum is 1.33 for existing processes, with 1.67 required for new processes or critical characteristics. See our guide on Cpk values and sigma levels for a full breakdown of thresholds and their defect rates.
Should you report Cpk or Ppk to a customer?
Report Ppk (or Pp) when a customer wants to know what the process actually delivered, since it is calculated from the overall standard deviation and reflects real, observed variation, which is why PPAP submissions typically ask for Ppk. Use Cpk internally to judge the process’s potential once it has been demonstrated to be in statistical control. Many organizations report both, so the customer sees actual performance alongside long-term capability.
What does it mean if Ppk comes out higher than Cpk?
It usually points to subgroups that were not formed rationally. Cpk is based on within-subgroup variation, so if a subgroup accidentally spans a shift change, two machines, or a tool change, the within-subgroup spread is inflated and Cpk is dragged below Ppk. On a small study it can also be ordinary sampling noise, but the first thing to check is always how the subgroups were collected.
Should a new process use Cp/Cpk or Pp/Ppk during initial qualification?
Use Pp/Ppk during initial qualification, such as a PPAP submission, since a new process hasn’t yet demonstrated statistical control and Cp/Cpk assumes a stability that doesn’t exist yet. Once the process has run long enough to confirm control on a chart, Cp/Cpk becomes the meaningful ongoing metric.
What does it mean when Cpk and Ppk are nearly equal?
It means within-subgroup and overall variation came out close, which is the signature of a process that stayed stable across the whole study. Nothing systematic shifted between subgroups, so short-term capability is a fair prediction of long-term performance. That is the result you want, not a coincidence that needs explaining away.