Your PPAP submission is due Friday. The control plan calls for a capability study on five critical dimensions. Your SQE sends back the package with one comment: "Element 11 requires Ppk, not Cpk."
You used Cpk. They are close in value. The math looks similar. But they are not the same thing, and submitting the wrong one is a legitimate basis for rejection.
This article explains the difference precisely, which index PPAP Element 11 requires and why, what the acceptance thresholds actually are, and what to do when your process does not hit them.
What Cpk Measures
Cpk is the process capability index for ongoing, stable production. It answers: given that my process is in statistical control, how well does it fit within the specification limits?
The formula:
Cpk = min[ (USL - x̄) / (3σ̂), (x̄ - LSL) / (3σ̂) ]
The critical variable is σ̂ (sigma-hat): the estimated within-subgroup standard deviation. This is derived from your control chart, typically from the average range (R-bar/d2 method) or the pooled standard deviation across subgroups. It measures only the short-term, within-subgroup variation -- the variation that exists when the process is running under stable, consistent conditions.
Because σ̂ excludes between-subgroup variation (shift, drift, day-to-day machine warmup effects, operator changeovers), Cpk reflects the best-case capability of a stable process. If Cpk is 1.45 and your process is in control, you have demonstrated that the process can produce to spec -- assuming it stays that way.
Cpk is the right metric for:
- SPC monitoring after production launch
- Control plan reaction thresholds
- Process improvement studies comparing before and after a change
- Customer reporting on ongoing performance
What Ppk Measures
Ppk is the preliminary process performance index used before and during PPAP. It answers: given the actual variation we observed in this production run, how well do parts fit within the specification limits?
The formula:
Ppk = min[ (USL - x̄) / (3σ), (x̄ - LSL) / (3σ) ]
The critical variable here is σ (overall standard deviation): the standard deviation of all individual measurements in the study, calculated the same way you would calculate it on any dataset. It includes within-subgroup variation AND between-subgroup variation -- shift, drift, lot-to-lot differences, and any other sources of real-world variation that occurred during the data collection window.
Because σ includes all observed variation, Ppk reflects actual production performance, not an idealized model of a stable process. A short-term study might look clean on a control chart but still show meaningful run-to-run shifts when you look at the full data distribution.
Ppk is the right metric for:
- PPAP Element 11 (Initial Process Studies)
- Launch readiness gates
- Pre-production validation studies
- Any capability study where process stability has not been established
The Core Difference: Which Sigma
Both Cpk and Ppk use the same numerator. Both measure how far the process mean is from the nearest specification limit, normalized by three standard deviations. The only difference is which standard deviation goes in the denominator.
| Index | Sigma Used | Variation Included |
|---|---|---|
| Cpk | σ̂ (within-subgroup) | Within-subgroup only |
| Ppk | σ (overall) | Within-subgroup + between-subgroup |
| Cp | σ̂ (within-subgroup) | Within-subgroup only (no centering adjustment) |
| Pp | σ (overall) | Within-subgroup + between-subgroup (no centering adjustment) |
Because overall sigma is always greater than or equal to within-subgroup sigma, Ppk is always less than or equal to Cpk for the same dataset. If the two values are close, your process is stable -- between-subgroup variation is minimal. If Ppk is significantly lower than Cpk, you have meaningful shift, drift, or other sources of variation between subgroups that a control chart may not have flagged yet.
This relationship matters at PPAP time. A process with Cpk = 1.85 and Ppk = 1.12 is technically capable within subgroups but is experiencing enough real-world variation to be borderline unacceptable. Submitting Cpk would mask that problem.
PPAP Element 11: What the Standard Actually Requires
PPAP 4th Edition calls for initial process capability studies at Element 11. The AIAG manual is explicit: the required index for initial studies is Ppk.
The minimum acceptance thresholds:
| Ppk Value | Status |
|---|---|
| ≥ 1.67 | Acceptable -- full PSW approval |
| ≥ 1.33 and < 1.67 | Conditionally acceptable -- customer approval required, corrective action plan expected |
| < 1.33 | Not acceptable -- PSW cannot be approved without customer disposition |
The 1.67 threshold is not arbitrary. It builds in process shift allowance. The traditional assumption is that a process mean can shift ±1.5σ over time. A Ppk of 1.67 with that shift still yields a Cpk of approximately 1.33 during ongoing production -- which is the standard monitoring threshold.
For safety-critical and regulatory characteristics, many automotive OEMs (particularly German Tier 1 suppliers operating under VDA standards, and IATF-certified sites with customer-specific requirements) require Ppk ≥ 2.00. This is common for characteristics tied to airbag components, brake systems, and structural welds. If your customer has a CSRS (Customer Specific Requirements Supplement), read it before assuming 1.67 is sufficient.
IATF 16949:2016 Clause 9.1.1.3 requires the organization to use statistical methods for product/process analysis and to determine which characteristics require capability studies -- but it defers to customer requirements and the control plan for specific thresholds. The PPAP standard sets those thresholds for the automotive supply chain.
The Most Common Mistake
Suppliers running process capability studies in Excel often calculate the wrong thing without realizing it. Standard Excel STDEV() computes overall sigma (the right input for Ppk). But teams familiar with SPC software sometimes pull the within-subgroup sigma from their SPC chart and drop it into the formula, producing a Cpk value while labeling it Ppk on the submission.
The math looks nearly identical. The customer reviewer may not catch it on first review. But during a formal PPAP audit or customer source approval visit, this discrepancy will surface, and it will delay the program.
The check: Ppk and Cpk should be labeled distinctly on Element 11 submissions. If the study was collected over 25 or more consecutive pieces (the AIAG minimum), Ppk should always appear in the Element 11 section. Cpk belongs in the SPC plan on the control plan.
What to Do When You Cannot Hit 1.67
If your preliminary study returns Ppk = 1.18, you have three options:
Option 1: Negotiate a conditional approval. If Ppk is between 1.33 and 1.67, most customers will accept a PSW with a corrective action plan attached. This requires a documented timeline to improve the process to 1.67+, typically 60 to 90 days. This is legitimate -- use it when you know the fix but need time to validate it.
Option 2: Tighten the process, re-study. If the capability shortfall comes from excessive variation rather than poor centering, tools like DOE, measurement system analysis, or fixture improvements can reduce sigma without touching the specification. Before re-running the study, verify the study itself is not inflating variation (inadequate measurement system, inconsistent sample collection, environmental effects during the study window).
Option 3: Escalate to the customer immediately. Do not wait until submission day to surface a Ppk of 0.95. Customers can arrange engineering deviation approvals, inspection containment agreements, or adjusted acceptance criteria -- but only if they have time to process the request. Surprising a customer at PSW review with a failing capability index is a program risk, not a quality issue.
One thing that does not work: switching to Cpk to get a better number and hoping no one notices. Element 11 requires Ppk for a reason.
After PPAP Approval: Switching to Cpk
Once the PSW is approved and production begins, the focus shifts from Ppk to Cpk. Your control plan defines the monitoring frequency, subgroup size, and reaction plan. SPC tracks Cpk (and X-bar/R or X-bar/S charts) on an ongoing basis.
The ongoing Cpk threshold is 1.33 for most characteristics. If Cpk drops below 1.33 during production, the control plan should specify immediate reaction: contain suspect product, identify the assignable cause, implement corrective action.
The relationship between your PPAP Ppk and your ongoing Cpk is also a diagnostic signal. If production Cpk consistently falls below your initial Ppk from the PPAP study, the process has degraded. If Cpk is stable and higher than your Ppk, the initial study may have captured an atypical variation window. Either way, tracking both over time reveals process behavior that point estimates cannot.
Running These Studies in QualityEngineer.ai
The Analyze module handles both Cpk and Ppk calculations with the correct sigma for each. Paste in your measurement data, select the specification limits, and the module outputs both indices, the Cp and Pp (spread-only versions), process centering, and a histogram overlay showing where the distribution sits relative to the specification.
For PPAP preparation, the Package module structures Element 11 with the correct index labels and generates the submission-ready format. If Ppk is below threshold, the system flags it before you submit, not after the customer reviews it.
Summary
Cpk and Ppk differ by one variable: which sigma goes in the denominator. Cpk uses within-subgroup sigma and reflects stable-process, best-case capability. Ppk uses overall sigma and reflects actual observed performance including all sources of variation.
PPAP Element 11 requires Ppk. The minimum is 1.67 for most characteristics. Submitting Cpk when Ppk is required is a documentation error, not a rounding difference.
If your initial study returns Ppk below 1.33, surface it to the customer immediately and bring a plan. If it is between 1.33 and 1.67, negotiate a conditional approval with a corrective action timeline. Do not substitute Cpk to pass a threshold you did not hit.
After launch, switch to Cpk for ongoing SPC monitoring with a threshold of 1.33. Track both indices over time to detect process drift before it produces non-conforming parts.
Need to run a capability study for your next PPAP? The Analyze module at QualityEngineer.ai computes Cpk, Ppk, Cp, and Pp with the correct methodology, plus SPC charts, Gauge R&R, and ANOVA. Start free, no credit card required.
About the Author
Daniel Crouse is the founder of QualityEngineer.ai and has spent 15+ years in supplier quality, PPAP, and manufacturing systems. He built QualityEngineer.ai because quality engineers deserve better tools than Excel.
