QualityOS
Free tool · no signup required

MSA / Gauge R&R calculator

Paste 60 measurements (3 operators × 10 parts × 2 trials) — instantly see %GR&R, EV, AV, NDC per AIAG MSA 4th edition. AIAG verdict (acceptable / marginal / unacceptable). No Excel, no Minitab.

Your data never leaves the browser — all computation is client-side.

Click into a cell and type a value. Or paste 60 numbers from Excel at once using the "Paste from Excel" button below. Pasting directly into a cell auto-distributes values to the following cells.

Filled: 0 / 60
PartOperator AOperator BOperator C
Trial 1Trial 2Trial 1Trial 2Trial 1Trial 2
1
2
3
4
5
6
7
8
9
10

For %P/T (Precision-to-Tolerance) — optional.

Optional, needed for %P/T.

MVP scope: 3 operators × 10 parts × 2 trials (the AIAG default). X̄-R method, sigma estimator R̄/d2*. For other configurations (5 trials, 2 operators, Type 1 gauge capability, ANOVA method) use the AIAG MSA 4th ed. handbook or APIS IQ-MSA / Minitab.

How Gauge R&R works (AIAG X̄-R method)

Short reference. Full derivation in the AIAG MSA 4th ed. handbook (chapters 2 and 3).

Input

3 operators (A, B, C) each measure 10 different parts, each part twice (trial 1 and trial 2). Total 60 measurements. The parts must cover the real process variability range.

Formulas

R̄    = mean of 30 cell-ranges (max−min across the 2 trials,
        per operator-part cell)
EV   = R̄ / d2*(trials)         repeatability
AV   = √(max(0, (X̄diff/d2*(ops))² − EV²/(p·r)))   reproducibility
R&R  = √(EV² + AV²)             combined gauge variation
PV   = R(part avg) / d2*(parts) part variation
TV   = √(R&R² + PV²)            total variation

%GR&R = 100 × R&R / TV
NDC   = floor(1.41 × PV / R&R)
%P/T  = 100 × 6 × R&R / (USL − LSL)    (if tolerance given)

d2*(n=2)     = 1.128
d2*(ops=3)   = 1.91
d2*(parts=10) = 3.18

AIAG verdict

  • %GR&R < 10%: measurement system ACCEPTABLE for all characteristic types including safety/critical.
  • 10% ≤ %GR&R < 30%: MARGINAL — acceptable for non-critical characteristics depending on the cost of gauge repair/replacement. Improve before using on PPAP safety/critical characteristics.
  • %GR&R ≥ 30%: UNACCEPTABLE — gauge variability absorbs more than 30% of process variability. No Cpk study makes sense until you fix this.

NDC — Number of Distinct Categories

NDC = floor(1.41 × PV / R&R) — how many categories of part variability the gauge can distinguish. AIAG recommends a minimum of 5. NDC < 5 means the gauge is too coarse to resolve real differences between parts.

Frequently asked questions

What is the X̄-R method and how is it different from ANOVA?
X̄-R (Average and Range) is a fast manual method based on Range statistics — it uses d2* factors to estimate sigma from ranges. ANOVA (Analysis of Variance) is more precise but requires software like Minitab. AIAG MSA 4th edition accepts both; for a standard 3×10×2 Gauge R&R study X̄-R is sufficient and what most automotive QEs use.
Why exactly 3 operators × 10 parts × 2 trials?
This is the AIAG default and the most common configuration in automotive. 3 operators = covers shift variability. 10 parts = covers process variability (must be representative across the tolerance range). 2 trials = the minimum to estimate repeatability. For other shapes (5 trials, 2 operators, or a Type 1 single-operator study) you need different d2* constants — for those use the AIAG handbook.
What Gauge R&R do VW, Stellantis, BMW, Ford require?
All automotive OEMs accept the AIAG MSA 4th edition criteria: %GR&R < 10% acceptable, 10-30% marginal (subject to the cost of fixing the gauge), ≥ 30% unacceptable. For safety / critical characteristics (CC, SC) < 10% is typically required without exception. For PPAP submission MSA is one of the 18 elements — you submit raw data + report. Verify specifics in your customer Quality Agreement.
How do I pick 10 parts for an MSA study?
AIAG MSA recommends 10 parts covering the entire expected process range (as wide as possible). Do NOT pick 10 parts clustered in the middle of the tolerance — that would underestimate NDC. Ideally: a couple close to LSL, a couple in the middle, a couple close to USL. Label parts 1–10 and keep the labels hidden from operators (they MUST NOT know which part they are measuring — blind study, prevents bias).
What do I do when %GR&R > 30%?
Decide where it comes from: if EV % > AV % → gauge problem (repeatability) — calibration, wear, unclear handling. If AV % > EV % → operator problem (reproducibility) — training, clearer work instructions, jig/fixture. If NDC < 5 → gauge is too coarse — replace. Repeat the MSA study after the fix. Before fixing, NEVER feed the data into a Cpk study — Cpk would be biased by gauge variability.

Built by a QE who actually runs MSA studies

For PPAP Level 3 you don't pass without an MSA study.

12 years in automotive quality. Production Quality Leader at HELLA-Forvia Kočovce — 24-person QA team across 2 factories. MSA certification + VDA 6.3. MSA studies are PPAP element 9 — no Tier 1 supplier passes PPAP submission without one. This calculator is the version I had open during every MSA review.

Read the full story →

Related tools

Quality is a system, not a single index. These calculators work together — that’s why they’re all free.

Cpk Calculator

Process capability — whether your variation gives a measured part the chance to stay in spec (IATF baseline 1.33 / safety 1.67).

Open

RPN + Action Priority

AIAG-VDA 2019 FMEA: Severity × Occurrence × Detection. Tells you where containment investment is worth it.

Open

AI Control Plan

From measurement into the Control Plan — characteristic, spec, method, sample size, frequency, reaction plan. Linked to PFMEA.

Open
QualityOS — quality assistant

Gauge failed MSA? The AI helps with root cause and customer report.

When %GR&R > 30%, the QualityOS AI Assistant helps you decide whether it's a repeatability or reproducibility issue, proposes corrective actions, and — if needed — drafts the customer-facing MSA report for your OEM customer.

Start free for 30 daysSee how QualityOS works

No credit card · cancel anytime · data hosted in the EU (Frankfurt)