MSA Standards and Requirements
Table of Contents
The document landscape
The most common source of confusion is treating all of these as interchangeable rulebooks. They serve distinct roles.
| Document | What it is | What it addresses |
|---|---|---|
| IATF 16949 | A certifiable requirement standard for automotive quality management systems. | Requires that measurement systems analysis be performed, without prescribing the statistical method in detail. |
| AIAG MSA reference manual | A reference manual describing methods. Not itself a certifiable standard. | The practical how-to for bias, linearity, stability, Gauge R and R and attribute studies. Widely treated as the default method in automotive supply chains. |
| ISO 22514-7 | An international standard for capability of measurement processes. | An uncertainty-based approach to measurement process capability, used more commonly in European and German automotive practice. |
| JCGM 100 (the GUM) | The internationally agreed guide to expressing measurement uncertainty. | How to build and report an uncertainty budget. The underlying basis for uncertainty statements everywhere else. |
| ISO/IEC 17025 | A certifiable standard for the competence of testing and calibration laboratories. | What a calibration laboratory must do, including reporting uncertainty. Relevant to the certificates you receive rather than to studies you run in production. |
The practical relationship is that a requirement standard such as IATF 16949 obliges you to analyse your measurement systems, while a reference manual such as the AIAG MSA manual tells you how it is conventionally done. Auditors generally expect a documented, justified method rather than one specific calculation.
How the AIAG MSA manual is organized
The manual is structured around the measurement system properties rather than around tools, which is why its contents map cleanly onto the studies described elsewhere in this Learning Center.
- General concepts: measurement as a process, the sources of variation, and the distinction between accuracy and precision.
- Location studies: bias, linearity and stability, meaning the properties concerned with where the readings sit.
- Width studies: repeatability and reproducibility, presented with both the Average and Range method and the ANOVA method.
- Attribute studies: agreement analysis for pass and fail inspections, including kappa.
It also introduces the number of distinct categories, ndc, and the percentage acceptance guidance that is widely quoted. Note that the manual presents these as guidelines to be applied with judgement, not as pass marks. The frequently cited 10 percent and 30 percent boundaries for percent GRR are conventions from this guidance, and customer requirements can and do override them.
IATF 16949 and the MSA requirement
For automotive suppliers, IATF 16949 is usually the reason an MSA is being performed at all. The measurement systems analysis requirement sits in the clause covering measurement system analysis within the monitoring and measuring resources section, referenced in the 2016 edition as clause 7.1.5.1.1.
What the requirement asks for, in substance
That statistical studies be conducted to analyse the variation present in the results of each type of inspection, measurement and test equipment identified in the control plan, and that the methods and acceptance criteria used conform to those in the applicable reference manuals, with other methods permitted where approved by the customer.
Two consequences follow. First, the scope is driven by your control plan: the equipment listed there is the equipment expected to have a study. Second, alternative methods are explicitly allowed with customer approval, which is why some suppliers use the ISO 22514-7 uncertainty approach instead.
Always verify the clause number and wording against the edition your certification is issued under. Clause numbering has changed across editions and sanctioned interpretations are issued periodically, so a number quoted in training material can be out of date.
ISO 22514-7 and the uncertainty approach
ISO 22514-7 treats the measurement process as something to be characterized by its uncertainty rather than by a variance decomposition. Instead of splitting observed variation into repeatability and reproducibility, it builds a budget of the contributions affecting the measurement process and compares the resulting uncertainty to the tolerance.
The advantage is that it naturally accommodates contributions a Gauge R and R never sees, such as the calibration uncertainty of the master and thermal effects. The cost is that it requires a properly constructed uncertainty budget, which is more work and demands more metrological knowledge. Where a customer specifies this approach, the uncertainty primer in this Learning Center covers the underlying method.
What GaugeConnection produces
| Report | Property addressed | Typical relevance |
|---|---|---|
| Gauge R and R | Repeatability and reproducibility | The width study most often expected for variable gauges listed in a control plan. |
| Gauge Capability | Repeatability and bias against tolerance | Common where one operator or an automated gauge is involved, and in German automotive practice. |
| Stability Tracker | Stability over time | Evidence that a measurement system continues to behave as it did when it was validated. |
| Uncertainty | Combined and expanded uncertainty | Supports uncertainty-based approaches and decision rules near specification limits. |
| Calibration schedule and certificates | Calibration status and traceability records | Supports the separate requirement that measuring equipment be calibrated and traceable, which is distinct from MSA. |
What GaugeConnection does not cover
Stating the gaps plainly is more useful than implying complete coverage:
- Bias studies. No dedicated tool. Bias appears only indirectly, through Cgk in a capability study and through calibration certificate data.
- Linearity studies. No dedicated tool. Assessing bias across the measuring range must be done outside this software.
- Attribute agreement analysis. No dedicated tool. All current analysis tools are variable-data methods.
- Any determination of conformance. The software calculates and records; it does not decide whether your quality system meets a standard.
If your control plan or your customer requires one of these studies, it must be run against your own documented procedure and retained accordingly.
Preparing for the audit conversation
Auditors rarely challenge the arithmetic. They challenge whether you understand and can justify what you did. Be ready to answer:
- Which method you used and why, whether ANOVA or Average and Range, and whether that choice is documented.
- How the parts were selected, and whether they represent the actual range of production variation rather than a convenient handful.
- Which denominator your percent GRR uses, total variation or tolerance, since the same study yields different percentages under each.
- What your acceptance criteria are, where they come from, and whether a customer requirement overrides the general convention.
- What you did when a study failed. An organization that can show a failed study and the resulting action is in a far stronger position than one whose studies all pass.