Quality and Six Sigma Proposed system

Standards set the rules; Six Sigma statistics show whether a process keeps them

Which quality rulebooks would apply to a botanical ingredient plant, what each covers, and the statistics that turn specifications into measured control: SPC, capability, gauge R&R and designed experiments.

2 of 14quality rulebooks on the map are law: 21 CFR 111 and 117

39%of 1,938 Part 111 citations in fiscal 2025 fell in Subpart E, 14% of its sections

1statistical tool named in Part 111: a sound statistical sampling plan

19%of study variation from measurement in a published gauge R&R study

Status. EUTECNIQ has no laboratory, data, certification or accreditation. Charts use published data or labeled simulations.

Standards map

Two FDA rules are law; every other quality standard on this map is voluntary

Quality rulebooks for a botanical ingredient, by kind of authorityVerified editions, Oct 24, 2026

Click a segment to zoom, the centre to zoom out. Segment size shows structure only (one unit per listed element), not importance.Sources: eCFR 21 CFR 111 and 117; ISO catalogue pages for ISO 9001, 22000, 17025, 13053-1/-2 and 7870-2; FSSC; SQFI; GFSI; ICH; AIAG (checked Oct 24, 2026).

Data table: every element on the map

Kind of authority	Rulebook (edition)	Element
U.S. law (FDA)	21 CFR 111, dietary supplement CGMP	Specifications (Subpart E)
U.S. law (FDA)	21 CFR 111, dietary supplement CGMP	Quality control (Subpart F)
U.S. law (FDA)	21 CFR 111, dietary supplement CGMP	Master and batch records (Subparts H, I)
U.S. law (FDA)	21 CFR 111, dietary supplement CGMP	Laboratory operations (Subpart J)
U.S. law (FDA)	21 CFR 111, dietary supplement CGMP	Complaints and returns (Subparts N, O)
U.S. law (FDA)	21 CFR 117, human food CGMP and preventive controls	CGMP (Subpart B)
U.S. law (FDA)	21 CFR 117, human food CGMP and preventive controls	Hazard analysis and risk-based preventive controls (Subpart C)
U.S. law (FDA)	21 CFR 117, human food CGMP and preventive controls	Supply-chain program (Subpart G)
U.S. law (FDA)	21 CFR 117, human food CGMP and preventive controls	Records (Subpart F)
Certification	ISO 9001:2026, quality management systems (edition 6)	Quality management requirements for any organization
Certification	ISO 9001:2026, quality management systems (edition 6)	Internal audit and management review
Certification	ISO 22000:2018, food safety management systems	HACCP principles (Codex)
Certification	ISO 22000:2018, food safety management systems	Prerequisite programmes
Certification	ISO 22000:2018, food safety management systems	Interactive communication
Certification	FSSC 22000 (version 6; version 7 published May 7, 2026)	ISO 22000
Certification	FSSC 22000 (version 6; version 7 published May 7, 2026)	Sector prerequisite programmes (ISO/TS 22002-x; ISO 22002-x:2025 in v7)
Certification	FSSC 22000 (version 6; version 7 published May 7, 2026)	FSSC additional requirements
Certification	SQF Food Safety Code: Dietary Supplement Manufacturing, Edition 10	SQF System Elements (Module 2)
Certification	SQF Food Safety Code: Dietary Supplement Manufacturing, Edition 10	Food safety plan under Codex HACCP (element 2.4.3)
Certification	SQF Food Safety Code: Dietary Supplement Manufacturing, Edition 10	Good industry practice module
Accreditation	ISO/IEC 17025:2017, testing and calibration laboratories	Impartiality and confidentiality (clause 4)
Accreditation	ISO/IEC 17025:2017, testing and calibration laboratories	Personnel, facilities, equipment, traceability (clause 6)
Accreditation	ISO/IEC 17025:2017, testing and calibration laboratories	Methods, measurement uncertainty, validity of results (clause 7)
Accreditation	ISO/IEC 17025:2017, testing and calibration laboratories	Management system (clause 8)
ICH guidance (drug)	ICH Q8(R2), pharmaceutical development (2009)	Target profile, critical quality attributes, design space
ICH guidance (drug)	ICH Q9(R1), quality risk management (2023)	Risk assessment, control, communication and review
ICH guidance (drug)	ICH Q10, pharmaceutical quality system (2008)	Management review and continual improvement
Statistical method standards	ISO 13053-1 and -2:2011, Six Sigma (confirmed 2022)	DMAIC methodology (part 1)
Statistical method standards	ISO 13053-1 and -2:2011, Six Sigma (confirmed 2022)	Tools and techniques (part 2)
Statistical method standards	ISO 7870-2:2023, Shewhart control charts	Control limits and patterns
Statistical method standards	AIAG Measurement Systems Analysis, 4th edition (2010)	%GRR and number-of-distinct-categories guidelines
Statistical method standards	AIAG and VDA FMEA Handbook, 1st edition (2019)	Seven steps and action priority

FDA's Parts 111 and 117 bind; certificates, accreditation, ICH guidance and the statistical standards are chosen by the firm or required by customers.

Scope

No single certificate covers law, food safety, laboratory competence and management review

What each rulebook is, and what it is notRegulation and standards

21 CFR 111.1, 117.5(e), 117.130 (eCFR, Oct 22, 2026); ISO 9001, 22000, 17025 pages; FSSC 22000 and UKAS bulletin (v7 published May 7, 2026); SQFI Dietary Supplement Manufacturing code, Edition 10; GFSI recognised programmes; ICH Q10 §4.1.

A plant making both bulk extracts and finished supplements would answer to Part 117 and Part 111 at once; certificates add evidence, not legal compliance.

Explanation, details and sources

Rulebook	Kind	Checked by	Edition in force, Oct 24, 2026	Does not cover
21 CFR Part 111	Regulation (final)	FDA investigators	eCFR current to Oct 22, 2026; final rule 72 FR 34752 (2007)	Dietary-ingredient-only operations whose customers further process the ingredient (FDA SECG); drug manufacturing
21 CFR Part 117	Regulation (final)	FDA investigators	eCFR current to Oct 22, 2026; final rule 80 FR 55908 (2015)	Subparts C and G do not apply to supplement operations that comply with Part 111 (§ 117.5(e))
ISO 9001:2026	Voluntary standard	Accredited certification bodies (ISO does not certify)	Edition 6, published Sep 16, 2026; replaced ISO 9001:2015	Food-safety hazard analysis; laboratory competence
ISO 22000:2018	Voluntary standard	Certification bodies; certification is optional	Edition 2; confirmed 2023; a revision (ISO/DIS 22000) is under development	GFSI recognition on its own; laboratory competence
FSSC 22000	Certification scheme	Licensed certification bodies	Version 6 (mandatory since Apr 1, 2024); Version 7 published May 7, 2026, all audits to v7 from May 1, 2027 (UKAS)	U.S. legal compliance
SQF	Certification scheme	SQF certification bodies	Edition 10 released March 2026; audits anticipated from Jan 1, 2027 (SGS); code "Food Safety Code: Dietary Supplement Manufacturing" (FSC 31)	U.S. legal compliance
ISO/IEC 17025:2017	Voluntary standard	Accreditation bodies (ILAC MRA signatories accredit testing labs to it)	Edition 3; confirmed 2023	Manufacturing GMP; product release
ICH Q8(R2), Q9(R1), Q10	Drug guidance	Drug regulators; no authority over supplements	Q8(R2) Aug 2009; Q9(R1) Jan 18, 2023; Q10 Jun 4, 2008	Dietary supplements as a legal requirement

Parts 111 and 117 require specifications, verification and records, not a particular statistical method. The only statistical tool Part 111 names is a "sound statistical sampling plan" for choosing which finished batches to test (§§ 111.75(c), 111.80(c)); Part 117 names none. The statistics below are one disciplined way to meet these rules.

Inspection data

Specifications and production records drew 57% of FDA's 1,938 Part 111 citations in fiscal 2025

Each Part 111 subpart's share of the rule against its share of the citationsGovernment data

FDA, Inspection observations FY2025 (Foods sheet), downloaded Oct 24, 2026; eCFR 21 CFR Part 111 structure, Oct 22, 2026. Grouping by subpart computed for this page (src/modules/qual_six/compute/p111.py).

Data table: sections and citations by subpart

Subpart	Title (eCFR)	Sections	Share of 84	Citations	Share of 1,938	Citation types
E	Requirement to Establish a Production and Process Control System	12	14.3%	759	39.2%	33
F	Production and Process Control System: Requirements for Quality Control	11	13.1%	216	11.1%	41
H	Production and Process Control System: Requirements for the Master Manufacturing Record	2	2.4%	188	9.7%	5
I	Production and Process Control System: Requirements for the Batch Production Record	2	2.4%	163	8.4%	5
J	Production and Process Control System: Requirements for Laboratory Operations	5	6.0%	77	4.0%	10
D	Equipment and Utensils	4	4.8%	75	3.9%	13
M	Holding and Distributing	6	7.1%	72	3.7%	7
C	Physical Plant and Grounds	4	4.8%	67	3.5%	19
O	Product Complaints	3	3.6%	66	3.4%	7
N	Returned Dietary Supplements	7	8.3%	65	3.4%	4
B	Personnel	5	6.0%	53	2.7%	8
K	Production and Process Control System: Requirements for Manufacturing Operations	6	7.1%	49	2.5%	9
G	Production and Process Control System: Requirements for Components, Packaging, and Labels and for Product That You Receive for Packaging or Labeling as a Dietary Supplement	6	7.1%	43	2.2%	17
L	Production and Process Control System: Requirements for Packaging and Labeling Operations	6	7.1%	37	1.9%	9
P	Records and Recordkeeping	2	2.4%	8	0.4%	2
A	General Provisions	3	3.6%	0	0.0%	0

Subpart E holds 14% of Part 111's sections and drew 39% of the citations; writing measurable specifications comes before any statistics.

Explanation, details and sources

Five years of the same data, by subpart and section, are on Quality system; a market view is on Competitive landscape. Record, audit-trail and traceability controls, the other half of the records problem, are on Digital lab.

DMAIC

DMAIC orders the work: define what matters, prove the measurement, then improve and hold the gain

Define, Measure, Analyze, Improve, Control: the tool for each questionSchematic

ISO 13053-1:2011 (DMAIC methodology) and ISO 13053-2:2011 (tools and techniques), both confirmed 2022; Montgomery 2019.

SIPOC map of a spray-dried botanical extractIllustrative

SIPOC (suppliers, inputs, process, outputs, customers): ASQ, "SIPOC diagram". Example process is generic.

The operational excellence library works DMAIC end to end on one example; this page supplies the statistics each phase needs.

Define

A critical-to-quality tree turns "what the label says" into ten measurable characteristics

From customer need to the five specification categories of 21 CFR 111.70(e)Illustrative

21 CFR 111.70(e): specifications for identity, purity, strength and composition, and limits on contamination (eCFR, Oct 22, 2026). Test methods: USP, ICH and COA.

Variables earn capability indices; pass-or-fail attributes earn defect rates. Either way, the specification limit comes first.

Measure

A gauge R&R study splits observed variation into the parts and the measurement system

Where the variance goes in a crossed gauge R&R studySchematic with standard equations

Expected mean squares as printed by SAS PROC VARCOMP (METHOD=GRR); Burdick, Borror and Montgomery 2003; ndc formula: Minitab, methods and formulas.

Data

Operators included (published study) Operator 1 Operator 2 Operator 3

Repeatability SD (what-if)0.71 Reproducibility SD (what-if)1.13 Part-to-part SD (what-if)6.95 Tolerance, USL − LSL (°C/W)40

%GRR of study variation19.0%

%Contribution of GRR3.60%

%Tolerance (6σ_GRR)20.1%

Distinct categories (ndc)7

Repeatability share of GRR variance28.3%

Interaction p-value5.1 × 10⁻⁷

AIAG MSA 4th ed. guideline for %GRRmay be acceptable, depending on the application

Measurements by part and operator, and the variance componentsPublished data

Data table: the 90 published measurements (°C/W)

Part	Op 1 t1	Op 1 t2	Op 1 t3	Op 2 t1	Op 2 t2	Op 2 t3	Op 3 t1	Op 3 t2	Op 3 t3
1	37	38	37	41	41	40	41	42	41
2	42	41	43	42	42	42	43	42	43
3	30	31	31	31	31	31	29	30	28
4	42	43	42	43	43	43	42	42	42
5	28	30	29	29	30	29	31	29	29
6	42	42	43	45	45	45	44	46	45
7	25	26	27	28	28	30	29	27	27
8	40	40	40	43	42	42	43	43	41
9	25	25	25	27	29	28	26	26	26
10	35	34	34	35	35	34	35	34	35

ANOVA at the current settings
Source	DF	SS	MS	F
Parts	9	3,935.96	437.33	162.27
Operators	2	39.27	19.63	7.28
Parts × operators	18	48.51	2.70	5.27
Repeatability (error)	60	30.67	0.51
Total	89	4,054.40

Data: Houf RE, Berman DB. IEEE Trans CHMT 1988;11(4):516–520, as listed in SAS/STAT 14.1 User's Guide, PROC VARCOMP Example 121.2 (specification 18–58 °C/W). Guidelines: AIAG MSA 4th ed. (2010), as quoted by Minitab.

Method, assumptions and limits

What it calculates: A crossed two-way random-effects ANOVA gauge study: variance components for parts, operators, the part-by-operator interaction and repeatability; %GRR of study variation, %Contribution, %Tolerance and the number of distinct categories. The what-if mode computes the same metrics directly from standard deviations you enter.
Equations: σ²_E = MS_E; σ²_PO = (MS_PO − MS_E)/r; σ²_O = (MS_O − MS_PO)/(p·r); σ²_P = (MS_P − MS_PO)/(o·r); negative estimates are set to zero. σ²_GRR = σ²_E + σ²_O + σ²_PO. %GRR = 100·σ_GRR/σ_total; %Contribution = 100·σ²_GRR/σ²_total; %Tolerance = 100·6σ_GRR/(USL − LSL); ndc = 1.41·σ_P/σ_GRR, truncated (minimum 1). When the interaction p-value exceeds 0.05 (Minitab's default "alpha to remove interaction"), the interaction is pooled into repeatability.
Variables and units: p = 10 parts, o = 1–3 operators, r = 3 trials; thermal performance in °C per watt.
Acceptance guidance: AIAG MSA 4th edition (2010), as quoted by Minitab: %GRR below 10% acceptable; 10–30% acceptable depending on the application, cost of the device, cost of repair and other factors; above 30% not acceptable; ndc should be 5 or more. These are guidelines, not regulatory limits.
Valid range: Balanced crossed designs only. At least two operators must stay selected. What-if inputs are clamped to the slider ranges.
Does not establish: Bias, linearity or stability of the gauge; the adequacy of any EUTECNIQ method; or analytical method validation under ICH Q2(R2) or USP, which is a separate exercise.
References: SAS/STAT 14.1 PROC VARCOMP Example 121.2; Houf and Berman 1988; Burdick, Borror and Montgomery 2003; Minitab support: "Is my measurement system acceptable?", number of distinct categories, default settings.
Verification: Reproduces every sum of squares, mean square and variance component printed by SAS (e.g., Var(part) 48.29259, Var(operator) 0.56461, Var(interaction) 0.72798, Var(error) 0.51111, PTR 0.20145). Independent reference: tests/numerics/ref/qual_quality-six-sigma.py.

In the published study, measurement took 19% of the study variation and ndc is 7: inside AIAG's "may be acceptable" band.

Control

A control chart separates routine variation from a real change

Live X̄ chart of capsule fill weight with run rules: simulated streamSimulated data

Subgroups of 5 capsules; mean 500 mg, SD 9 mg; illustrative specification 460–540 mg. Limits from the first 25 subgroups. Press "Shift the mean" and watch the rule numbers appear.Rules: Western Electric (1956) zone rules 1–4 and the six-point trend rule, as listed in NIST/SEMATECH e-Handbook §6.3.2; Nelson 1984. Seeded normal random numbers.

Rule numbers used on the chart

Number	Signal
1	One point beyond 3σ of the plotted statistic
2	Two of the last three beyond 2σ, same side
3	Four of the last five beyond 1σ, same side
4	Eight in a row on one side of the centre line
5	Six in a row steadily rising or falling

What each run rule looks likeSchematic

NIST/SEMATECH e-Handbook of Statistical Methods, §6.3.2 (WECO rules and trend rule).

Run rules detect small shifts sooner and raise false alarmsModel output

3σ rule: ARL = 1/p (exact, normal model; NIST §6.3.2.1 gives about 371 in control). Run rules: Monte Carlo, 40,000 runs per shift, known parameters (src/modules/qual_six/compute/arl.py). Check: Champ CW, Woodall WH. Technometrics 1987;29(4):393–399 (91.75, quoted by NIST).

Data table: average run length by shift

Shift (σ)	3σ rule only	+ rules 2–4	+ rules 2–5
0	370.4	91.7	78.3
0.25	281.1	57.5	52.0
0.5	155.2	27.5	26.6
0.75	81.2	14.7	14.5
1	43.9	9.3	9.2
1.25	25.0	6.5	6.5
1.5	15.0	4.9	4.8
2	6.3	3.1	3.1
2.5	3.2	2.2	2.2
3	2.0	1.7	1.7

Run rules catch a 1σ shift after about 9 points instead of 44, at the price of a false alarm every 92 points instead of 370.

Capability

C_p and C_pk together fix the expected defect rate; the sigma level adds a 1.5σ convention

Normal process, two-sided specificationppm = 10⁶[Φ(−3C_pk) + Φ(−3(2C_p − C_pk))]

C_p, spread versus tolerance1.33 C_pk, distance to the nearer limit1.00 Measurement SD as % of process SD0%

Expected outside specification1,350 ppm

Beyond the nearer limit1,350 ppm

Z_bench (σ)3.00

Sigma level, 1.5σ convention4.50

Observed C_p with measurement error1.33

Observed C_pk with measurement error1.00

The capability plane: lines of equal defect rate, and the process they describeNormal-model calculation

Reference values (NIST §6.1.6, centered process)

C_p = C_pk	USL − LSL	Rejects (NIST)	This calculator
1.00	6σ	0.27%	2,700 ppm
1.33	8σ	64 ppm	63 ppm (C_p = 4/3)
1.66	10σ	0.6 ppm	0.57 ppm (C_p = 5/3)
2.00	12σ	2 ppb	0.002 ppm

NIST/SEMATECH e-Handbook §6.1.6; Montgomery 2019; 1.5σ shift: Harry and Lawson 1992, a convention questioned by Coskun and Ialongo 2020; Z_bench from the total tail area.

Method, assumptions and limits

What it calculates: For a normal process with a two-sided specification, the expected fraction outside the limits from C_p and C_pk, the benchmark Z, the sigma level under the 1.5σ convention, and the indices that would be observed if measurement error adds to process variation.
Equations: The nearer limit lies 3C_pk standard deviations from the mean and the farther limit 3(2C_p − C_pk). ppm = 10⁶[Φ(−3C_pk) + Φ(−3(2C_p − C_pk))]. Z_bench = Φ⁻¹(1 − p). Sigma level (convention) = Z_bench + 1.5. With σ²_observed = σ²_process + σ²_measurement, observed C_p = C_p/√(1 + g²), g = σ_m/σ_p; C_pk scales the same way when the gauge is unbiased.
Valid range: 0.5 ≤ C_p ≤ 2.5; C_pk is held at or below C_p (a larger value is impossible). Normality and statistical control are assumed; skewed data (for example, contaminant results near a detection limit) need other methods.
The 1.5σ convention: Six Sigma tables add 1.5σ to the long-term Z so that "6σ" corresponds to 3.4 defects per million. It is a convention from Motorola's practice, not a law of nature; EUTECNIQ would report measured within- and between-lot variation instead.
Does not establish: The capability of any real process, the stability of a process, or its shelf life. Capability from real data, with confidence intervals, is in the operational excellence library.
Verification: Reproduces the NIST table of rejects against C_p; independent reference: tests/numerics/ref/qual_quality-six-sigma.py.

A centered C_pk of 1.33 means about 63 ppm outside specification (NIST: 64 ppm); measurement error lowers the observed index.

Two normal distributions inside the same two-sided specification, eight standard deviations wide (Cp 1.33). Centered, the process puts 63.3 parts per million outside the limits. Moved one standard deviation toward the upper limit (Cpk 1.00), it puts 1,350 parts per million outside, almost all beyond the nearer limit. Computed from the normal distribution; a mathematical illustration, not process data. — Montgomery, Introduction to Statistical Quality Control, 8th ed. (2019), ch. 8; NIST/SEMATECH e-Handbook §6.1.6; 1.5σ convention, Harry & Lawson 1992. Mathematical illustration, not process data. Generated illustration, EUTECNIQ. Full-size SVG

Improve

A two-level factorial ranks seven effects from eight runs

Model in coded units (−1, +1)ŷ = b₀ + Σ b_ix_i + Σ b_ijx_ix_j + b₁₂₃x₁x₂x₃

Residual SD (ohms)0.304

Residual degrees of freedom5

Best predicted cornerX1 +, X2 −

Predicted impedance there (ohms)4.64

Cube of the eight runs and the ranked coefficientsPublished data (NIST)

Data table: NIST eddy current probe study (2³ full factorial)

Run (standard order)	X1 turns (90 / 180)	X2 distance (0.38 / 1.14)	X3 gauge (40 / 48)	Impedance (ohms)
1	−	−	−	1.70
2	+	−	−	4.57
3	−	+	−	0.55
4	+	+	−	3.39
5	−	−	+	1.51
6	+	−	+	4.59
7	−	+	+	0.67
8	+	+	+	4.29

Coefficients (NIST §5.6.1.5); an effect, high minus low, is twice the coefficient
Term	Coefficient
Mean	2.65875
X1	1.55125
X2	−0.43375
X3	0.10625
X1·X2	0.06375
X1·X3	0.12375
X2·X3	0.14875
X1·X2·X3	0.07125

NIST/SEMATECH e-Handbook of Statistical Methods, §5.6.1 (eddy current probe sensitivity case study): data, coefficients and residual standard deviations.

Method, assumptions and limits

What it calculates: Least-squares coefficients of the 2³ full factorial (orthogonal, so each coefficient is the same whichever terms are kept), predictions at the eight corners, and the residual standard deviation of the chosen model.
Equations: b_t = (1/8) Σ x_t,i y_i with x = ±1; residual SD = √(Σ(y − ŷ)²/(8 − 1 − k)) for k terms.
Why this dataset: It is a public, fully documented two-level factorial with published coefficients, so every number can be checked. The method transfers directly to extraction or drying factors (for example solvent ratio, temperature, time); this page does not claim results for any botanical process.
Valid range: Predictions inside the cube only; a two-level design cannot detect curvature, which needs centre points or a response-surface design.
Verification: Matches NIST's coefficients and its residual standard deviations for each nested model (1.74106, 0.57272, 0.30429, 0.26737, 0.23341, 0.19121, 0.18031 ohms). Independent reference: tests/numerics/ref/qual_quality-six-sigma.py.

Two main effects carry the study: adding X1 and X2 cuts the residual SD from 1.74 to 0.30 ohms (NIST).

Explanation, details and sources

A factorial study at laboratory scale finds which factors matter; carrying the chosen settings to pilot and commercial scale, and validating the process there, is covered on Scale-up and tech transfer.

Analyze

FMEA ranks failure modes before they happen; the 2019 AIAG-VDA handbook replaced the RPN with action priority

Seven steps of a failure mode and effects analysisSchematic

AIAG, "AIAG and VDA release new automotive FMEA handbook," June 3, 2019; ICH Q9(R1) (2023) for risk-management formality.

A worked FMEA with its risk map is in the operational excellence library.

See the quality system these tools would serve

Programs, release decisions and design-space controls, each on its own page.

Quality system QbD and control strategy USP, ICH and COA

Part	Op 1 t1	Op 1 t2	Op 1 t3	Op 2 t1	Op 2 t2	Op 2 t3	Op 3 t1	Op 3 t2	Op 3 t3
1	37	38	37	41	41	40	41	42	41
2	42	41	43	42	42	42	43	42	43
3	30	31	31	31	31	31	29	30	28
4	42	43	42	43	43	43	42	42	42
5	28	30	29	29	30	29	31	29	29
6	42	42	43	45	45	45	44	46	45
7	25	26	27	28	28	30	29	27	27
8	40	40	40	43	42	42	43	43	41
9	25	25	25	27	29	28	26	26	26
10	35	34	34	35	35	34	35	34	35

Part	Op 1 t1	Op 1 t2	Op 1 t3	Op 2 t1	Op 2 t2	Op 2 t3	Op 3 t1	Op 3 t2	Op 3 t3
1	37	38	37	41	41	40	41	42	41
2	42	41	43	42	42	42	43	42	43
3	30	31	31	31	31	31	29	30	28
4	42	43	42	43	43	43	42	42	42
5	28	30	29	29	30	29	31	29	29
6	42	42	43	45	45	45	44	46	45
7	25	26	27	28	28	30	29	27	27
8	40	40	40	43	42	42	43	43	41
9	25	25	25	27	29	28	26	26	26
10	35	34	34	35	35	34	35	34	35

Two FDA rules are law; every other quality standard on this map is voluntary

No single certificate covers law, food safety, laboratory competence and management review

Specifications and production records drew 57% of FDA's 1,938 Part 111 citations in fiscal 2025

DMAIC orders the work: define what matters, prove the measurement, then improve and hold the gain

A critical-to-quality tree turns "what the label says" into ten measurable characteristics

A gauge R&R study splits observed variation into the parts and the measurement system

A control chart separates routine variation from a real change

Cp and Cpk together fix the expected defect rate; the sigma level adds a 1.5σ convention

A two-level factorial ranks seven effects from eight runs

FMEA ranks failure modes before they happen; the 2019 AIAG-VDA handbook replaced the RPN with action priority

See the quality system these tools would serve

C_p and C_pk together fix the expected defect rate; the sigma level adds a 1.5σ convention

Part	Op 1 t1	Op 1 t2	Op 1 t3	Op 2 t1	Op 2 t2	Op 2 t3	Op 3 t1	Op 3 t2	Op 3 t3
1	37	38	37	41	41	40	41	42	41
2	42	41	43	42	42	42	43	42	43
3	30	31	31	31	31	31	29	30	28
4	42	43	42	43	43	43	42	42	42
5	28	30	29	29	30	29	31	29	29
6	42	42	43	45	45	45	44	46	45
7	25	26	27	28	28	30	29	27	27
8	40	40	40	43	42	42	43	43	41
9	25	25	25	27	29	28	26	26	26
10	35	34	34	35	35	34	35	34	35