Which VDA 6.3 edition applies in 2026?

The 4th edition, published January 2023. The VDA QMC publications listing as of 22 September 2026 still shows it as current. Results under the 2023 edition are not directly comparable to results under the 2016 edition, so any year-over-year supplier benchmark that crosses the change needs a note on it.

How many questions and star questions are there?

59 questions in the questionnaire, of which 17 are questions of special significance, plus 35 questions for the potential analysis. The star questions were partly reallocated in the 2023 revision but the count did not change. The questions themselves are copyrighted and are not reproduced here or anywhere else legitimate.

What is the A/B/C threshold?

We are not publishing one. The sources we could open disagree: some give B as 80 to 89% and C below 80%, another gives B as 75 to 89% and C below 75%. The 80% boundary is more commonly cited but that does not settle it. Take the threshold from your licensed copy of the standard or from the customer requirement that is driving the audit.

Does VDA 6.3 require an MES or real-time data capture?

No standard we examined mandates a particular system. What P6 asks for is evidence: equipment capability and maintenance records, targets with collected data and deviation analysis, and planned against produced quantities. A plant can produce all of that on paper if the paper is complete, timely and traceable. The reason plants move to automatic capture is that the traceability and the timeliness get hard to sustain by hand, not that a standard demands it.

Our records are complete but stored in three systems. Is that a problem?

It becomes one when the systems use different identifiers for the same characteristic or operation. The recurring points-loss reported by practitioners is production data that cannot be traced back to the relevant characteristic, and the usual mechanism is exactly this: a control plan and a measurement system that name the same thing differently, with no link between them. Fixing the identifiers is usually cheaper than replacing the systems.

Home

Compliance · VDA 6.3 process audit · status 22 September 2026

VDA 6.3 and production data: what P6 examines, and where the points go

For quality managers, process engineers and plant directors preparing a VDA 6.3 process audit. What the 4th edition covers in P6, the recurring points-loss around untraceable production data, and what an auditable data trail actually looks like.

In shortVDA 6.3 in its 4th edition of January 2023 has 59 questions, 17 of special significance, plus 35 for the potential analysis. The production-data pressure sits in P6: equipment capability and maintenance in P6.4, targets, data collection and deviation analysis in P6.5, planned against produced quantities in P6.6. The recurring loss is production data that cannot be traced to the characteristic.

What this is based on and what it is notVDA 6.3 is copyrighted and paywalled. No question, criterion or catalogue text is reproduced here, and none should be. What follows is a paraphrase of the audit's structure and of what practitioners and certification bodies report about it, written from sources that were opened and read. The standard prevails over this page in every case. Buy the current volume from VDA QMC before you build anything on it.

Which edition applies, and why last year's score is not comparable

The current edition is the 4th, published January 2023, superseding the 2016 third edition. As of 22 September 2026 the VDA QMC publications listing still shows the 4th edition of January 2023, so there is no newer version to move to.

59 questions in the questionnaire, of which 17 are questions of special significance, commonly written as star questions. A further 35 questions serve the potential analysis. The star questions were partly reallocated in 2023, but the count stayed the same.
What changed in 2023: software topics folded into the questionnaire; harmonisation with Automotive SPICE and the VDA Maturity Level Assurance method; purchasing requirements added into P3 and P4; guidance for remote audits incorporated; the chapters on the audit process and audit services removed.
Scores are not comparable across editions. DQS states plainly that results under the 2023 edition are not directly comparable to results under 2016. If you benchmark supplier scores year over year, that break has to be in your data.

Questions are scored on a discrete scale of 10, 8, 6, 4 or 0. There are no intermediate values, which is why a single weak answer moves a sub-element noticeably.

Where production data actually gets examined

The audit runs P1 to P7: potential analysis, project management, planning of product and process development, realisation of product and process development, supplier management, process analysis and series production, and customer care. P6 is the core element for anyone running a line, and three of its six sub-elements are about production data.

P6 sub-elements and the production data behind them
Sub-element	Subject	Data it rests on
P6.1	Process input: project handover, material availability, storage, identification and records	Material lot identity carried through to the part
P6.2	Process management: control-plan requirements in practice, repeat release after interruptions, special characteristics, handling of defective parts, material flow	Interruption and restart events, and the release decision attached to each one
P6.3	Personnel resources: operator capability, responsibilities, availability	Who was logged on, and against which qualification
P6.4	Material resources: equipment and machine capability, maintenance, measuring equipment, workstations	Per-asset availability, stoppage events with causes, completed maintenance, calibration status
P6.5	Effectiveness and efficiency: targets and key figures, data collection, analysis of deviations, regular audits	Target against actual per line and shift, each deviation with a reason and a linked action
P6.6	Process output: quantities and batch sizes, storage and transport, records, customer requirements at delivery	Planned against produced quantity per order and batch, split into good, scrap and rework

Sub-element scope as described in openly published training material, August 2024. Paraphrase only; the standard prevails.

P6.4 and P6.5 are the adjacent territory to IATF 16949 clause 8.5.1.5 on total productive maintenance, and the same records serve both. If you have built maintenance objectives for the IATF clause, see our page on IATF 16949 §8.5.1.5 and reuse them here rather than building a second set. P6.6 is the one people underestimate: it is about output rate and batch reality, not only conformity.

The recurring points-loss: data that cannot be traced to the characteristic

This is the finding that appears again and again in practitioner accounts of VDA 6.3 audits, and it is worth spelling out because the mechanism is mundane.

The worked exampleThe control plan calls a bore diameter a critical characteristic. The measurement system stores the same value under a different name, with different limits. Nothing links the two. Proving that the characteristic was checked correctly turns into manual data matching in front of the auditor, and what the auditor sees is a system that cannot answer its own question.

Auditors review SPC charts, capability values and inspection records, and expect a reaction plan when capability falls below requirement. All of that depends on being able to resolve a control-plan line to its readings. Where that resolution requires a person with local knowledge and half an hour, the evidence is weak even when the underlying work was done properly.

One widely circulated practitioner article reports that roughly 75% of critical findings fall in P5 and P6, attributed to weak detection of and reaction to process deviations in real time and to records that exist on paper but cannot be backed by a concrete, verifiable case when the auditor asks. That source carries no author and no date, so treat the percentage as an impression rather than a statistic. The underlying observation matches what certification bodies report about IATF audits generally: the frequent findings are not "we did not do it", they are "we cannot show that we did it".

Free tool

OEM Capacity & Audit Evidence Pack

Capacity declaration with demonstrated rate, a run-conditions log for capability studies, an audit evidence index, and an EnPI and baseline sheet. Excel.

Get the evidence pack →

What an auditable data trail looks like

The test is whether a single question can be answered from the record without human reconstruction. Pick a part made three months ago and follow it.

Auditor's question to the record that answers it without reconstruction
Question	What must exist	Sub-element
Which machine made this part, when, and who was on it?	Part or batch identity resolving to machine, operation, time window and operator logon	P6.1, P6.3
Show me this characteristic being checked at the planned frequency last quarter	Measurement records keyed to the control-plan characteristic ID, with achieved frequency against planned	P6.2
What happened after the line was interrupted on that shift?	Interruption event, the repeat-release decision, and the disposition of parts made since the previous good check	P6.2
Was this asset maintained as planned?	Preventive maintenance planned against done, on schedule, per asset, computed rather than transcribed	P6.4
Show me a target that was missed and what you did	Target against actual per line and shift, each deviation carrying a recorded reason and a linked action with an owner and a date	P6.5
Planned quantity was X. What did you actually make?	Planned against produced per order and batch, split good, scrap and rework, reconcilable to the delivery note	P6.6

Three properties make the difference between a record that survives this and one that does not. Timestamps written at the event, not entered at the end of the shift. Identifiers shared between systems, so the control plan, the measurement record and the production record use the same name for the same characteristic. A retention window that covers the audit period and is queryable during the audit rather than restorable afterwards.

A/B/C thresholds: sources disagree, so check the standard

The overall result is classified A, B or C. We are not publishing a threshold, because the secondary sources we could open do not agree and the standard itself is paywalled.

Two practitioner sources give A at 90% and above, B from 80 to 89%, C below 80%.
Another gives A from 90 to 100%, B from 75 to 89%, C below 75%.
The 80% boundary is the more commonly cited of the two. That is not the same as being correct. Take the threshold from your licensed copy of VDA 6.3 or from your customer's own requirement, and do not let a supplier development plan hang on a number from a blog.

Downgrading rules also exist and can pull a result below what the arithmetic average alone would give: the 2016 rules were carried into the 2023 edition. Published descriptions of the exact mechanics vary and come from vendor commentary, so we are not restating them either. What is safe to plan around is the direction: a zero, and in some accounts a low score on a star question, hurts disproportionately, so a single unanswerable question in P6 is not averaged away.

Which customers run VDA 6.3, and which run something else

Volkswagen conducts process audits per VDA 6.3 using elements P5 to P7, and product audits per VDA 6.5 at least annually, according to the openly published Formel Q Capability, 9th edition, December 2022. Escalation runs through the Critical Suppliers programme.
Renault does not carry a VDA 6.3 clause in its customer-specific requirements. It requires a minimum ASES rank instead, which makes ASES the operative process assessment. Published descriptions of the ASES structure and grading come from a 2018 practitioner blog and the scheme may have changed, so confirm the current method with Renault rather than with us.
Tier 1 cascades matter. Ficosa escalates a supplier to its Level 1 on four or more non-A months out of six, two consecutive C ratings, or a C rating in a VDA 6.3 audit. A process-audit grade can therefore trigger a commercial escalation directly.

Questions

Which VDA 6.3 edition applies in 2026?: The 4th edition, published January 2023. The VDA QMC publications listing as of 22 September 2026 still shows it as current. Results under the 2023 edition are not directly comparable to results under the 2016 edition, so any year-over-year supplier benchmark that crosses the change needs a note on it.
How many questions and star questions are there?: 59 questions in the questionnaire, of which 17 are questions of special significance, plus 35 questions for the potential analysis. The star questions were partly reallocated in the 2023 revision but the count did not change. The questions themselves are copyrighted and are not reproduced here or anywhere else legitimate.
What is the A/B/C threshold?: We are not publishing one. The sources we could open disagree: some give B as 80 to 89% and C below 80%, another gives B as 75 to 89% and C below 75%. The 80% boundary is more commonly cited but that does not settle it. Take the threshold from your licensed copy of the standard or from the customer requirement that is driving the audit.
Does VDA 6.3 require an MES or real-time data capture?: No standard we examined mandates a particular system. What P6 asks for is evidence: equipment capability and maintenance records, targets with collected data and deviation analysis, and planned against produced quantities. A plant can produce all of that on paper if the paper is complete, timely and traceable. The reason plants move to automatic capture is that the traceability and the timeliness get hard to sustain by hand, not that a standard demands it.
Our records are complete but stored in three systems. Is that a problem?: It becomes one when the systems use different identifiers for the same characteristic or operation. The recurring points-loss reported by practitioners is production data that cannot be traced back to the relevant characteristic, and the usual mechanism is exactly this: a control plan and a measurement system that name the same thing differently, with no link between them. Fixing the identifiers is usually cheaper than replacing the systems.

Sources

Published by TEEPTRAK SAS, which makes production-monitoring and OEE software. Every figure is sourced on the page. Funding rules, standards and reporting duties change: check the official documents before you budget or commit.