Launching a new product is the moment of greatest quality risk in an organization's entire lifetime. It is when the design is not yet frozen, the suppliers have not yet demonstrated capability, and the production line has not reached its stable operating regime. The discipline that brings order to this chaos is called APQP (Advanced Product Quality Planning), a methodology born in the automotive industry and captured in the reference manual of the AIAG, which today is applied in sectors as diverse as medical devices, consumer electronics, and food. Its premise is simple: a product's quality is decided in the design phase, not in final inspection. Inspecting at the end is expensive, late, and does not recover the cost of the defect.
In this article we walk through the safe development of new product lines, from the voice of the customer to the approval of the production part, with an emphasis on the tools that prevent a launch from turning into a crisis of recalls, complaints, and scrap.
The economic framework that justifies all this effort is the rule of ten: the cost of correcting a defect multiplies by roughly ten at each stage it advances without being detected. A design error costs a few euros if corrected on the drawing; hundreds if discovered in the prototype; thousands in production; and it can reach catastrophic figures, including product recall and reputational damage, if it reaches the end customer. APQP is, at its core, a machine for pushing defect detection toward the earliest and cheapest phases of development.
The five phases of APQP and why they matter
APQP structures development into five sequential phases, but with deliberate overlaps. Phase 1, planning and definition, translates the voice of the customer (VOC) into measurable requirements through quality function deployment (QFD) and sets reliability and cost targets. Phase 2, product design and development, produces the design FMEA (DFMEA), the drawings with special characteristics, and the test equipment requirements. Phase 3, process design and development, generates the flow diagram, the process FMEA (PFMEA), and the control plan. Phase 4, product and process validation, executes the significant production run, the capability study, and the PPAP. Phase 5, feedback and corrective action, measures reduced variation, customer satisfaction, and closes the learning loop into the next program.
The most frequent mistake is compressing phases 2 and 3 under schedule pressure: the design is frozen without having analyzed the process failure modes, and the result is a line that produces, but that produces defects. The phases exist precisely to force decisions in the correct order.
FMEA: anticipating failure before it happens
Failure Mode and Effects Analysis (FMEA) is the technical heart of a safe launch. Its current version follows the harmonized AIAG-VDA methodology published in 2019, which replaced the classic Risk Priority Number (RPN = Severity × Occurrence × Detection) with a three-level Action Priority (AP) matrix: high, medium, and low. The change responds to a well-founded criticism: two combinations could yield the same RPN (for example, 8×3×4 = 96 and 4×4×6 = 96) when their real severity was very different. AP gives priority to high severity even when occurrence is low, which is exactly what product safety demands.
A well-executed FMEA identifies the special characteristics (critical and significant) that will be marked on the drawing and monitored in the control plan. A poorly executed FMEA is a spreadsheet filled in the day before the audit so there is paperwork to show. The difference between the two is measured in field complaints.
PPAP: the proof that the process can deliver
The PPAP (Production Part Approval Process) is the dossier that demonstrates to the customer that the supplier understands the requirements and that its process is capable of meeting them repeatedly. Level 3, the most common for new parts, requires up to 18 elements: design records, design and process FMEAs, flow diagram, control plan, measurement system analysis (MSA), process capability studies, dimensional results, material and performance records, and the signed sample part. The document that crowns it is the PSW (Part Submission Warrant).
The capability study deserves special attention. A Ppk ≥ 1.67 is normally required in the initial run (preliminary performance capability, on a process not yet stabilized) and a Cpk ≥ 1.33 in stable production. Confusing Pp/Ppk with Cp/Cpk is a technical error that invalidates the interpretation: the former uses the overall standard deviation, the latter the within-subgroup variation. Reporting Cpk when there is not yet stable statistical control is deceiving your own system.
Control plan and MSA: making sure what is measured can be trusted
The control plan is the document that connects the risk analysis with the reality of the plant floor. It inherits the special characteristics from the FMEA and specifies, for each one, the measurement method, the sample size and frequency, the specification limits, and the reaction plan when a deviation occurs. It exists at three levels, prototype, pre-launch, and production, that are tightened as the process matures. A control plan that does not define the reaction plan is only half a tool: it says what to measure but not what to do when the measurement goes out of limits, which is precisely the critical moment.
Before trusting any data from the control plan, you must validate the system that generates it through Measurement System Analysis (MSA). The key study is the repeatability and reproducibility study (Gage R&R), which decomposes the observed variation into product variation versus variation of the measurement system itself (the same operator measuring several times, and different operators measuring the same part). The usual criterion: a %GRR below 10% is acceptable, between 10% and 30% is conditional, and above 30% the measurement system is useless. A process with an excellent Cpk measured with an unvalidated instrument is a statistical illusion.
Comparison table: reactive approach versus APQP
| Dimension | Reactive launch (final inspection) | Launch with APQP |
|---|---|---|
| When the defect is detected | Customer or final audit | Design phase (DFMEA) |
| Relative cost to correct | Up to 1,000x (rule of ten) | Cost of revising a drawing |
| Evidence for the customer | One-off certificate of conformity | Complete PPAP dossier |
| Process capability known | No, discovered over time | Yes, Ppk verified before SOP |
| Traceability of decisions | Scattered | Living FMEA + control plan |
Common mistakes in new product development
- Freezing the design without a PFMEA. The product is validated but not the process that makes it; the line starts up with unanalyzed failure modes.
- Control plan disconnected from the FMEA. The control plan must inherit the special characteristics from the FMEA; if it lives apart, it monitors what it should not.
- Skipping the MSA. Without measurement system analysis (an R&R study below 10%), you do not know whether the measured variation comes from the product or from the instrument.
- PPAP as a formality. Gathering 18 documents without having executed a significant run turns the dossier into fiction.
- Ignoring phase 5. The launch does not end at SOP (Start of Production): variation reduction and lessons learned feed the next program.
Frequently asked questions
Is APQP only for automotive?
No. It was born in IATF 16949, but its logic of planning quality by phases applies to any physical product. The medical sector uses equivalents (Design Controls under the ISO 13485 standard and FDA guidance), and consumer electronics replicates the same flow under a different name.
What is the difference between Cpk and Ppk?
Cpk measures the capability of a stable process using within-subgroup variation; Ppk measures overall performance including all observed variation. In a launch, Ppk is reported because the process has not yet demonstrated statistical stability.
How many elements does a PPAP require?
Up to 18 at the standard level 3, but the customer defines the submission level (1 to 5). What is never missing is the PSW, the signed warrant of part submission.
Why did AIAG-VDA replace the RPN with Action Priority?
Because the RPN, by multiplying three factors, allowed risks of very different severity to obtain the same score. The AP matrix always prioritizes severity, aligning the analysis with the real safety of the product.
Conclusion
A new product fails or succeeds long before the first unit is sold: it is decided on the design table, in the honesty of the FMEA, and in the rigor of the capability study. APQP is not bureaucracy; it is the operational translation of an uncomfortable but true idea: the most expensive defect is the one designed in without anyone noticing. Organizations that treat the PPAP as a living dossier, and not as last-minute paperwork, reach the start of production with variation under control, the measurement system validated, and a customer who trusts what they receive. At Summum Calidad we support complete launches, from the voice of the customer to the signed PSW, so that innovation reaches the market without paying the toll of recalls.