ICBB cheat sheet

Project charter scope: a Six Sigma project addresses one problem, one metric (Y), and a bounded process; multi-Y charters are split before Measure begins.

Voice of the Customer translates to Critical to Quality (CTQ) requirements; each CTQ needs an operational definition, a metric type (continuous, discrete), and a specification limit or target.

SIPOC maps Suppliers, Inputs, Process (4-7 high-level steps), Outputs, Customers; it defines project boundaries and is not a detailed process map.

Cost of Poor Quality categories: internal failure, external failure, appraisal, prevention; COPQ is used to justify project selection and prioritize improvement.

SMART goal test: Specific, Measurable, Achievable, Relevant, Time-bound; the goal states the baseline, target, and date (e.g., reduce defects from 8% to 2% by Q4).

Pareto principle (80/20): a small set of causes produces most defects; Pareto charts rank categories descending and drive project focus.

Kano model classifies features as must-be, performance, and delighter; must-be features are unspoken and cause dissatisfaction only when missing.

Team roles: Champion sponsors and removes barriers, Black Belt leads the DMAIC project, Green Belt supports part-time, Process Owner sustains the control plan after handover.

Project financials: hard savings are measurable budget or cost reductions; soft savings are avoided costs or capacity gains and require finance sign-off.

Data types: continuous (variable) data uses Xbar-R, Xbar-S, I-MR charts; discrete (attribute) data uses p, np, c, u charts.

p chart plots proportion defective with varying subgroup size; np chart plots number defective with constant subgroup size; c chart counts defects per unit with constant opportunity; u chart counts defects per unit with varying opportunity.

MSA for continuous data: Gage R&R reports %Study Variation and %Tolerance; under 10% acceptable, 10-30% conditional, over 30% unacceptable.

Attribute agreement analysis reports within-appraiser and between-appraiser agreement; kappa below 0.70 indicates weak agreement beyond chance.

Process capability: Cp = (USL - LSL) / 6s measures potential; Cpk = min((USL - mean), (mean - LSL)) / 3s measures actual centering; Cpk is always less than or equal to Cp.

Pp and Ppk use the overall standard deviation; Cp and Cpk use within-subgroup sigma; Ppk is generally lower than Cpk when the process drifts.

Sigma level from DPMO: 1.5 sigma shift convention means 3.4 DPMO equals 6 sigma; 66,807 DPMO equals 4 sigma; 6,210 DPMO equals 4.5 sigma.

Sample size for a mean: n = (z * s / E)^2, where E is the desired margin of error; doubling precision requires four times the sample.

Measurement system resolution must be at most one-tenth of the tolerance range (10:1 rule) or the system cannot detect process variation.

Hypothesis test decision rule: p-value less than alpha rejects the null hypothesis; alpha is the Type I error rate (false alarm), beta is the Type II error rate (missed effect).

Power = 1 - beta; power increases with larger effect size, larger sample size, and lower alpha; 80% power is a common planning minimum.

Test selection: 1-sample t for one mean, 2-sample t for two independent means, paired t for before/after on the same units, 1-proportion test for one defect rate.

ANOVA tests equality of three or more means; F = between-group variance / within-group variance; a significant F requires post-hoc comparison (Tukey) to find which groups differ.

Chi-square tests association between two categorical variables; expected cell counts should be at least 5 for validity.

Regression: R-squared is the proportion of variation explained; check residual plots for randomness, and inspect p-values of individual predictors for significance.

Correlation coefficient r ranges from -1 to 1; correlation does not establish causation, and a lurking variable can drive both variables.

Multi-Vari study stratifies variation into within-unit, between-unit, and temporal (lot-to-lot, shift) components to locate dominant variation sources.

FMEA: Risk Priority Number = Severity x Occurrence x Detection; act first on high-severity items regardless of RPN ranking.

DOE screening designs: full factorial runs 2^k; fractional factorial 2^(k-p) reduces runs at the cost of confounding (aliasing) main effects with interactions.

Resolution III designs confound main effects with two-factor interactions; Resolution IV confound two-factor interactions with each other; Resolution V confound two-factor interactions with three-factor interactions.

PDCA (Plan-Do-Check-Act) is the Deming cycle; in DMAIC, Improve pilots the change (Do) and verifies results before Control standardizes it.

Lean waste categories: overproduction, waiting, transport, overprocessing, inventory, motion, defects, plus unused talent.

5S sequence: Sort, Set in order, Shine, Standardize, Sustain; it is a workplace organization method applied to stabilize the improved process.

Poka-yoke is error-proofing that prevents or detects defects at the source; devices can be prevention-based (contact, fixed-value) or detection-based.

Kaizen is continuous incremental improvement by the people doing the work; kaizen events are short, focused, cross-functional workshops.

Pilot selection: run the improvement on a limited scope with measurable baseline and exit criteria before full deployment to limit risk.

Control plan elements: process step, characteristic and specification, measurement method, sample size and frequency, control method, and reaction plan with owner.

SPC chart selection: use I-MR for individual measurements, Xbar-R for subgroups of 2-9, Xbar-S for subgroups of 10 or more, p or u for attribute data.

Control limits are set at mean plus or minus 3 sigma from process data, not from specification limits; specification limits belong to the customer.

Western Electric rules: 1 point beyond 3 sigma, 2 of 3 beyond 2 sigma, 4 of 5 beyond 1 sigma, 8 consecutive points on one side of the centerline signal special cause.

Process behavior: a process in statistical control with Cpk below 1.0 requires fundamental process change, not tighter monitoring.

Standardization documents: updated work instructions, control charts, reaction plans, and a response owner are required for handover to the process owner.

Ongoing verification uses audits, control chart review, and periodic capability re-checks (Cpk, Ppk) to confirm gains are sustained.

Mistake-proofing in Control: add detection or prevention devices so the failure mode cannot recur without operator vigilance.

Project closure requires documented savings verified by finance, lessons learned, and transfer of the control plan to the process owner.