Cpk of 1.67 on a PPAP submission. The part goes into production. Six months later, the customer has a field return that traces back to the same characteristic the Cpk was supposed to validate. The PPAP was technically correct. The risk it was supposed to mitigate materialized anyway.
This sequence is more common than it should be — and it consistently traces back to a fundamental misunderstanding of what Cpk actually measures, and what it does not.
1. The Distinction That Most Supplier Quality Programmes Get Wrong
Cpk — Process Capability Index — measures what a process can achieve under controlled, short-term conditions. It is calculated from a sample taken during a specific window, typically under stable setup conditions with an experienced operator, using material from a single lot, on equipment that was recently calibrated and serviced. Cpk answers the question: what is this process capable of when everything is aligned correctly?
Ppk — Process Performance Index — measures what the process actually delivered over an extended period, including all sources of variation that were present during normal production: shift changes, tool wear across multiple setups, material lot variation between deliveries, environmental changes across weeks or months, operator variability across the full range of personnel who run the process. Ppk answers the question: what did this process actually produce when it was operating as it normally operates?
Datalyzer’s published analysis of process capability indices is precise on this point: „The difference between Cpk and Ppk indicates if the process is stable or in other words if there are special causes of variation which are influencing the average of the process even if control limits are not properly set. The requirement within most industries is that the Ppk value should exceed 1.67.“ The AIAG Capability reference manual confirms: „Ppk is preferable to Cpk since it captures all the process variation.“
Process capability thresholds in practice: the AIAG requires Cpk 1.33 for initial production and Cpk 1.67 for critical safety characteristics. Aerospace may require Cpk 2.0 for flight-critical dimensions. But these thresholds apply to the capability study — the short-term measurement. A supplier can submit a Cpk of 1.67 from a 30-piece initial capability study. If the same characteristic measured over six months of serial production shows a Ppk of 0.98, the PPAP was technically correct — and the risk management value of the capability data was near zero.
2. Where the Gap Between Cpk and Ppk Is Largest
The gap between initial capability (Cpk) and sustained performance (Ppk) is predictably largest in three situations:
Processes with significant material lot variation. Machining operations on aluminum alloys, composite layup processes using prepreg from multiple lots, and assembly operations using adhesives with variable viscosity across production batches all exhibit variation that is not captured in a single-lot initial capability study. The PPAP sample was drawn from one material lot under one setup. Production draws from dozens of lots across the production run. As 1factory’s quality academy notes, the goal of a capability study is to determine how often a process meets its desired outcomes — but that answer is only meaningful if the study conditions represent actual production conditions.
Processes that require frequent setup and adjustment. A machine set up correctly at the beginning of a shift and drifting across the shift produces an initial capability study that reflects the beginning-of-shift condition. The Cpk is the best-case estimate for that process. The Ppk — calculated across multiple shifts, multiple setups, and the full range of operator adjustment decisions — is the realistic estimate. AMREPinspect’s December 2025 analysis illustrated this with a specific example: Cpk of 1.67 under controlled conditions versus Ppk of 1.11 long-term, describing the divergence as „possible process drift or instability.“
Processes where operator competency is a significant source of variation. Any process where the outcome depends materially on operator skill — composite repair, manual welding, hand layup, precision assembly — has an inherent gap between what an experienced operator produces in a controlled capability study and what the full range of qualified operators produces in normal production. The PPAP was conducted by the most experienced technician available. Production is executed by the full operator pool, with varying experience levels and varying adherence to work instruction details.
3. What This Means for Supplier Qualification and PPAP Review
The AIAG PPAP Reference Manual specifies minimum sample sizes and study durations for initial process capability studies. AS9100 and EN 9100 reference similar principles. These requirements define the floor for what constitutes an acceptable initial capability submission — they do not define what constitutes adequate ongoing process performance assurance.
A mature supplier quality programme treats PPAP approval as the beginning of process performance monitoring, not the end of it. The iSixSigma practitioner community has documented this distinction extensively: „Cp and Cpk are for computing the index concerning the subgrouping of your data (different shifts, machines, operators), while Pp and Ppk are for the whole process (no subgrouping). It is more realistic to use Pp and Ppk than Cp or Cpk as the process variation cannot be tempered by inappropriate subgrouping.“ The kiuey.com analysis of process capability added the key practical insight: „If the process is stable enough, Cpk and Ppk will be uniform — but if unstable, Cpk will be greater than Ppk, implying shifts or drifts over time.“ That divergence is your signal.
The distinction between Cpk and Ppk should be explicitly addressed in supplier quality agreements for safety-critical and quality-critical characteristics. Suppliers should be required to report Ppk — not Cpk — for ongoing production, calculated from at least three production lots or 90 days of serial production. This is a straightforward addition to a supplier quality agreement that changes the information available for risk-based supplier management from „what can this process do?“ to „what is this process doing?“
4. The Audit Application
When reviewing supplier capability data during a second party audit, the question is not only what the Cpk value is — it is whether the study conditions that produced that number are representative of normal production. Useful audit questions: How many material lots were included in the capability study? How many operators ran the process during the study period? Was the equipment in its normal production condition or specifically serviced for the study? How does the Cpk from the capability study compare to the Ppk calculated from the last three months of production SPC data?
If the supplier does not have Ppk data from serial production for their quality-critical characteristics — because they do not have ongoing SPC — that is an audit finding. Not because SPC is required for its own sake, but because the absence of ongoing process performance data means the supplier is operating without objective evidence that the process that was qualified is still performing within the qualified parameters.
3 Concrete Actions
1. Review your PPAP approval criteria and add an explicit requirement that Ppk — not only Cpk — be reported for safety-critical and quality-critical characteristics once production has been running for at least 90 days. Set a minimum Ppk acceptance threshold consistent with your industry requirements (Ppk 1.33 as minimum, 1.67 for safety-critical characteristics per AIAG guidance) and make Ppk below threshold a trigger for supplier process review, not just corrective action request.
2. For characteristics where your supplier has submitted Cpk data but does not have ongoing SPC: include a review of actual production measurement data in your next supplier audit. Ask the supplier to pull the last 50-100 measurements for the characteristic in question from their production records and calculate the Ppk from that data in real time during the audit. The result will tell you more about actual process risk than the PPAP file.
3. When onboarding new suppliers in regulated industries: structure the production qualification in two stages. Stage one: initial capability study (Cpk) to confirm process can perform to specification. Stage two: 90-day production data review (Ppk) to confirm process is performing to specification under normal production conditions. Full qualification approval — and removal from elevated monitoring status — only after Stage two is complete with Ppk at or above your defined threshold.
👉 Request a Technical Consultation → www.skqc.org
Sources
- AIAG — Production Part Approval Process (PPAP) Reference Manual, 4th Edition. Automotive Industry Action Group.
- AIAG — Statistical Process Control (SPC) Reference Manual, 2nd Edition. Automotive Industry Action Group.
- Gunter, B.H. — Understanding Process Capability Indices (University of Waterloo reference), sas.uwaterloo.ca
- SixSigma.us — A Practitioner’s Guide to Choosing Between Cpk vs Ppk (January 13, 2025), 6sigma.us
- iSixSigma — Process Capability Cp, Cpk and Process Performance Pp, Ppk — What is the Difference? (October 2024), isixsigma.com
- AMREPinspect — Ppk vs Cpk: Understand Process Capability with Clear Formulas and Examples (December 27, 2025), amrepinspect.com
- Datalyzer — Process Capabilities Explained: Cp, Cpk, Pp and Ppk (August 20, 2025), datalyzer.com
- 1factory — A Guide to Process Capability (Cp, Cpk) and Process Performance (Pp, Ppk), 1factory.com
- Kiuey — Understanding Process Capability Analysis (Cp, Cpk, Pp and Ppk) (June 18, 2025), kiuey.com
- TheStatSigma — Process Capability and Performance Analysis Tool: Industry Variation by Sector, thestatsigma.com
#ProcessCapability #SPC #PPAP #SupplierQuality #QualityManagement #IATF16949 #EN9100 #AIAG #SKQC