2. APQP phases and deliverables

APQP is not a single review meeting or a checklist that gets signed once. It is a phased product-development discipline in which each phase has a defined purpose, a set of required outputs, and an exit condition that must be satisfied before the next phase begins. The AIAG APQP framework defines five phases, and the logic behind them is simple: reduce risk early, prove capability before commitment, and only release to production when the process has demonstrated that it can consistently meet requirements. This chapter walks through each phase, the deliverables that anchor it, and how the phases connect through the gate review process.

Phase 1 — Plan and Define Program

Phase 1 converts a customer need or business opportunity into a defined program with measurable requirements. The inputs are the voice of the customer (VOC), business case, market data, warranty history, and any regulatory requirements. The outputs are the design goals, reliability targets, preliminary bill of material, preliminary process flow, and the product assurance plan.

The most consequential deliverable here is the design goals and reliability targets. These are not vague aspirations; they are quantified. For example, a seat recliner mechanism might carry targets such as 30,000 cycle durability with no functional loss, operating torque within 2.5–4.0 N·m, and a warranty claim rate below 0.3% at 36 months. These numbers become the acceptance criteria that later phases must prove against. If Phase 1 produces only qualitative goals, every downstream gate becomes subjective.

Phase 1 also produces the preliminary process flow and the preliminary BOM. These are deliberately rough — they exist to expose assumptions about manufacturing early, not to freeze the design. A common failure mode is treating Phase 1 as a formality and rushing to Phase 2. The cost of that shortcut appears later as late design changes, which are the most expensive changes in the program.

Phase 2 — Product Design and Development

Phase 2 is where the design is created and verified analytically. The core deliverables are the Design Failure Mode and Effects Analysis (DFMEA), Design for Manufacturing and Assembly (DFM/DFA) reviews, design verification plan and report (DVP&R), engineering drawings and specifications, material specifications, and design records.

The DFMEA is the analytical engine of this phase. It is not a document to complete for audit purposes; it is a structured method for identifying failure modes, their effects, their severity, occurrence, and detection ratings, and the actions that reduce risk. A well-run DFMEA changes the design. If the DFMEA produces no design changes, either the design is genuinely mature or the analysis was performed superficially.

The DVP&R is the proof that the design meets the Phase 1 targets. Each design goal should map to at least one verification test with a defined sample size, test condition, acceptance criterion, and result. For the recliner example, the DVP&R would include a 30,000-cycle durability test at specified temperature and load, a torque measurement at defined positions, and a corrosion test per the relevant standard. The DVP&R is a living document; it is updated as tests are completed and as design changes occur.

Phase 2 also produces the prototype build and the prototype control plan. Prototype builds are not production-representative, but they must be controlled enough that test results are meaningful. The prototype control plan defines what is checked, how, and with what frequency during the build.

Phase 3 — Process Design and Development

Phase 3 shifts from product to process. The deliverables are the Process Flow Diagram, Process FMEA (PFMEA), pre-launch control plan, work instructions, packaging specifications, and process capability studies.

The Process Flow Diagram is the backbone of Phase 3. It maps every step from incoming material to shipped product, including inspection, storage, and handling. Each step in the flow diagram must appear in the PFMEA and in the control plan. This one-to-one correspondence is a common audit check: if a step exists in the flow but not in the PFMEA, the risk analysis is incomplete.

The PFMEA evaluates how the process can fail to produce the design intent. For a welding operation, failure modes might include insufficient weld penetration, porosity, or incorrect weld location. Each failure mode is rated for severity, occurrence, and detection, and actions are assigned to reduce risk. The output of the PFMEA directly shapes the control plan: high-risk failure modes require more frequent or more capable detection methods.

The pre-launch control plan is more stringent than the production control plan. It typically includes increased inspection frequency, additional check points, and tighter acceptance criteria. The purpose is to contain risk during the period when the process is not yet statistically stable. The production control plan, which replaces it after capability is demonstrated, reflects the reduced risk.

Process capability studies are the quantitative proof that the process can meet tolerances. For a critical characteristic with a tolerance of ±0.15 mm, a capability study might require a minimum Cpk of 1.33. If the initial study yields Cpk = 0.95, the process is not capable, and the gate cannot be passed without corrective action — typically a combination of process parameter adjustment, tooling improvement, or tolerance review.

Phase 4 — Product and Process Validation

Phase 4 is the validation phase. The deliverables are the significant production run, measurement system analysis (MSA), production capability studies, production control plan, Production Part Approval Process (PPAP) submission, and final process sign-off.

The significant production run is the pivotal event. It must be run on production tooling, production equipment, production operators, and production environment. The output is a defined quantity of parts — often 300 consecutive parts or a defined production run — from which capability and measurement data are drawn. If the run uses prototype tooling or non-production operators, the results do not validate the production process.

MSA is required before capability data can be trusted. A gauge that is not capable will produce capability indices that are meaningless. The standard approach is a Gage R&R study, which separates variation into repeatability, reproducibility, and part-to-part variation. A common acceptance criterion is that the gauge R&R contribution to total variation should be below 10%, with 10–30% acceptable depending on the characteristic's criticality.

PPAP is the formal submission that packages the evidence. It includes the design records, authorized engineering change documents, customer engineering approval, DFMEA, process flow diagram, PFMEA, control plan, MSA results, dimensional results, material and performance test results, initial process capability studies, qualified laboratory documentation, appearance approval report (if applicable), sample production parts, master sample, checking aids, and the part submission warrant (PSW). The PSW is the formal statement that the requirements have been met.

Phase 5 — Feedback, Assessment, and Corrective Action

Phase 5 begins after launch and continues through the product lifecycle. The deliverables are reduced variation, increased customer satisfaction, delivery and service performance, and lessons learned. This phase is often neglected because the program has "shipped," but it is where the organization captures the knowledge that improves the next program.

Effective Phase 5 practice includes monitoring warranty data, tracking customer complaints, and feeding that information back into the DFMEA and PFMEA for the next program. It also includes a formal lessons-learned review that documents what worked, what failed, and what should change in the APQP process itself.

How the Phases Connect Through Gates

The phases are connected by gate reviews. A gate is a decision point where the cross-functional team reviews the deliverables, confirms that the exit criteria are met, and decides whether to proceed, proceed with conditions, or hold. The gate is not a presentation; it is a decision. If the deliverables are incomplete or the data does not support the exit criteria, the gate should not be passed.

The following diagram shows the five phases, their primary deliverables, and the gate decision points that connect them.

Practical Gate Review Discipline

The gate review is where the APQP process either works or becomes theater. A gate that passes without evidence is worse than no gate at all, because it creates false confidence. The following practices distinguish effective gates from ceremonial ones.

Evidence over presentation. Each deliverable should be traceable to objective data. A DFMEA that lists failure modes but shows no risk reduction actions is incomplete. A capability study that reports Cpk = 1.45 but uses data from a non-production run is invalid. The gate reviewer's job is to ask "show me the data" for every claim.

Conditions are tracked, not forgotten. When a gate is passed with conditions, those conditions must be entered into an open-items list with owners and due dates. A common failure is that conditions are verbally agreed and never closed. The next gate should begin by reviewing the previous gate's open conditions.

The gate is cross-functional. Engineering, manufacturing, quality, purchasing, and program management all have a stake in the decision. A gate run by engineering alone will miss manufacturing feasibility issues. A gate run by quality alone will miss design intent issues. The cross-functional team is not a formality; it is the mechanism that catches problems before they become expensive.

Common Failure Modes Across Phases

Certain failures recur across programs and phases. Recognizing them early is a form of risk mitigation.

• Phase 1 goals are qualitative. "Improve reliability" is not a target. "Reduce warranty claims from 0.8% to 0.3% at 36 months" is a target. Without quantified goals, Phase 4 has no acceptance criteria.
• DFMEA is completed after the design is frozen. The DFMEA should influence the design, not document it. If the DFMEA is written after drawings are released, its risk-reduction actions become change requests, which are costly and often rejected.
• PFMEA and control plan are not linked. Every failure mode in the PFMEA should have a corresponding control in the control plan. If a high-severity failure mode has no detection control, the control plan is incomplete.
• MSA is skipped or rushed. Capability data from an unvalidated measurement system is unreliable. Gage R&R should be completed before capability studies, not after.
• PPAP is treated as a paperwork exercise. PPAP is a summary of evidence, not a substitute for it. A PPAP submission with missing or stale data will be rejected by the customer, and the rework delays launch.
• Phase 5 is neglected. Without a feedback loop, the organization repeats the same mistakes on the next program. Lessons learned should be captured while the program is fresh, not months later.

Deliverable Traceability

The table below summarizes the primary deliverables by phase and the gate at which they are typically reviewed. This is a reference, not a rigid template; the specific deliverables and gate assignments should be tailored to the customer's requirements and the program's risk profile.

The phases are not independent. A weakness in Phase 1 propagates through every subsequent phase. A DFMEA that misses a critical failure mode becomes a PFMEA that misses the corresponding process control, which becomes a control plan that fails to detect the defect, which becomes a warranty claim after launch. The gate review exists to break that chain before it reaches the customer.

In practice, the most effective APQP programs treat the gates as genuine decision points, maintain traceability from requirements to evidence, and use the lessons-learned loop to improve the next program. The deliverables are the artifacts; the discipline is the process. Both are necessary.