CLBB cheat sheet

Six Sigma targets 3.4 defects per million opportunities (DPMO), which corresponds to a process centered at 1.5 sigma shift from specification limits.

DPMO = (number of defects / (units × opportunities per unit)) × 1,000,000.

Rolled throughput yield (RTY) = product of each process step's first-pass yield; five steps at 95% yield give RTY ≈ 0.95^5 ≈ 77%.

Lean's eight wastes: defects, overproduction, waiting, non-utilized talent, transportation, inventory, motion, extra processing.

DMAIC applies to improving existing processes; DMADV (Define, Measure, Analyze, Design, Verify) applies to designing new processes or products.

Belt hierarchy: Champion sponsors and removes barriers, Master Black Belt provides technical leadership, Black Belt leads projects, Green Belt supports part-time.

Cost of poor quality categories: internal failure, external failure, appraisal, and prevention; prevention and appraisal are costs of good quality.

Critical to quality (CTQ) translates the voice of the customer into measurable characteristics with defined specification limits.

Process capability Cp = (USL − LSL) / (6σ); it ignores centering and measures potential capability.

Cpk = min[(USL − mean)/(3σ), (mean − LSL)/(3σ)]; Cpk ≤ Cp always, and equality implies the process is centered.

Ppk uses overall sample standard deviation while Cpk uses within-subgroup sigma; Ppk ≤ Cpk for a stable process.

A process is considered capable at Cpk ≥ 1.33; Cpk ≥ 1.67 is often required for critical characteristics.

Sigma level ≈ 3 × Cpk when the process is centered (e.g., Cpk 1.33 ≈ 4 sigma).

OEE = Availability × Performance × Quality; world-class OEE is commonly cited as 85%.

Takt time = available production time / customer demand rate.

Little's Law: Work-in-process = throughput rate × flow time; a key relationship for process flow analysis.

KPIs must be SMART (specific, measurable, achievable, relevant, time-bound) and tied to a balanced-scorecard perspective.

Tuckman's stages of team development: forming, storming, norming, performing, and adjourning.

RACI assigns each task one Accountable party, plus Responsible, Consulted, and Informed roles; only one person is accountable per task.

Situational leadership matches style to follower readiness: directing, coaching, supporting, and delegating.

A team charter documents purpose, scope, sponsor, members, milestones, and decision authority.

Stakeholder analysis plots influence versus interest to choose engagement strategy (manage closely, keep satisfied, keep informed, monitor).

Conflict resolution modes from the Thomas-Kilmann model: competing, collaborating, compromising, avoiding, accommodating.

Effective meeting cadence: short daily stand-ups, weekly team reviews, and milestone tollgate reviews with the sponsor.

Groupthink countermeasures include assigning a devil's advocate, anonymous idea collection, and leader speaking last.

A project charter states the business case, problem statement, goal statement, scope, team, and milestones.

A problem statement must be data-based, quantified, and free of cause or solution language.

SMART goal statements specify metric, baseline, target, and target date.

SIPOC maps Suppliers, Inputs, Process, Outputs, and Customers at a high level to define scope boundaries.

Voice of the customer is gathered via interviews, surveys, focus groups, complaints, and observation, then translated into CTQs.

Kano model classifies requirements as must-be, performance, attractive, indifferent, and reverse.

A Pareto chart applies the 80/20 rule to prioritize the vital few problems.

Project scope is bounded by explicit in-scope and out-of-scope statements; scope creep is controlled through the charter.

A measurement system analysis (MSA) qualifies a gauge; Gage R&R variation should be under 10% of tolerance to be acceptable, 10–30% conditional, above 30% unacceptable.

Gage R&R partitions variation into repeatability (equipment) and reproducibility (appraiser).

Data types: continuous (variable) versus discrete (attribute: nominal or ordinal).

A process map or detailed flowchart identifies inputs, outputs, and decision points for data collection planning.

Data collection plans specify what, how, when, where, who, and how much data to collect.

Baseline capability uses Cp/Cpk or DPMO computed from measured data before improvement.

Sampling: random sampling supports representativeness; rational subgrouping keeps within-subgroup variation from common causes.

Stratification separates data by shift, machine, operator, or lot to reveal hidden variation before analysis.

Hypothesis testing steps: state H0 and Ha, choose significance level α, select test, compute statistic, compare to critical value or p-value, conclude.

A p-value below α leads to rejecting H0; failing to reject does not prove H0 is true.

Type I error (α) rejects a true null; Type II error (β) fails to reject a false null; power = 1 − β.

Two-sample t-test compares means of two independent groups when data are continuous and approximately normal.

ANOVA compares means of three or more groups; F = between-group variance / within-group variance.

Chi-square test of independence evaluates association between two categorical variables.

Regression estimates the relationship between a continuous response and one or more predictors; R² is the proportion of variation explained.

A cause-and-effect (fishbone) diagram organizes potential causes into categories such as Man, Machine, Method, Material, Measurement, and Environment.

Correlation does not imply causation; a statistically significant correlation may be spurious or confounded.

Design of experiments (DOE) varies factors systematically to estimate main effects and interactions.

A full factorial with k factors at two levels requires 2^k runs; a fractional factorial reduces runs at the cost of confounding.

PDCA (Plan-Do-Check-Act) is the iterative improvement cycle underlying kaizen.

Poka-yoke (error proofing) prevents defects by design rather than by inspection.

5S: sort, set in order, shine, standardize, sustain.

Kaizen events (rapid improvement workshops) target a defined scope over a short, focused period.

FMEA prioritizes failure modes by Risk Priority Number = Severity × Occurrence × Detection.

Standard work documents the current best method and is the baseline for further improvement.

Control charts distinguish common-cause from special-cause variation; a point outside control limits signals special cause.

X-bar and R charts monitor continuous data in subgroups; individuals (I-MR) charts are used when subgroup size is one.

p, np, c, and u charts monitor attribute data; p and np for proportion defective, c and u for defect counts.

Control limits are computed from process data (typically ±3σ), not from specification limits.

Western Electric rules flag runs of 8 points on one side of centerline, trends, and repeating patterns as special cause.

A control plan records the characteristic, specification, measurement method, sample size, frequency, and reaction plan.

The project is handed off with documented standard work, training, and ownership transferred to the process owner.

Benefits are validated by comparing post-improvement capability to the baseline and confirming sustainment over time.

Lean focuses on eliminating waste and improving flow; Six Sigma focuses on reducing variation and defects.

Value-added activities transform the product in ways the customer pays for; everything else is waste.

Value stream mapping depicts material and information flow and identifies cycle time, lead time, and queues.

Kanban is a pull signal that limits work-in-process and triggers replenishment.

Single-minute exchange of die (SMED) reduces changeover time by converting internal to external setup tasks.

Total productive maintenance (TPM) targets equipment availability through autonomous and planned maintenance.

Cycle time reduction increases capacity; the constraint (bottleneck) governs system throughput per theory of constraints.

Standard deviation measures spread; reducing σ narrows the distribution and raises capability without changing the mean.

Project sponsor authorizes resources and champions the project; the process owner sustains gains after handoff.

A team's decision-making methods include consensus, majority vote, and leader-decides; consensus requires all members to support the decision.

Ground rules and a team charter reduce ambiguity and set behavioral expectations.

Effective facilitation uses a parking lot for off-topic items, a timekeeper, and a scribe.

Motivation theory: Herzberg's hygiene factors prevent dissatisfaction while motivators drive satisfaction.

Maslow's hierarchy orders needs from physiological through safety, belonging, esteem, and self-actualization.

Feedback should be specific, behavioral, and timely rather than evaluative.

Cross-functional teams bring diverse expertise but require deliberate coordination to avoid silos and miscommunication.

The DMAIC tollgate at Define confirms the charter, scope, and CTQs before data collection begins.

Measurement system error is quantified by bias, linearity, stability, repeatability, and reproducibility.

Attribute agreement analysis assesses agreement among appraisers on categorical judgments.

Sample size increases precision; larger n reduces the standard error of the mean (σ/√n).

Normal distribution: approximately 68% of values lie within ±1σ, 95% within ±2σ, 99.7% within ±3σ.

Baseline data should be plotted over time (run chart or control chart) to confirm stability before capability is computed.

A histogram reveals shape, center, and spread but hides time order; a run chart reveals trends.

Defect opportunity counts must be defined consistently to make DPMO comparisons valid.

Before hypothesis testing, verify data normality; non-normal data may require transformation or nonparametric tests.

Paired t-test is used when two measurements are taken on the same subject or matched pairs.

Regression diagnostics include residual normality, constant variance, and independence.

Multi-vari studies separate positional, cyclical, and temporal variation.

DOE screening designs (e.g., Plackett-Burman) identify important factors; response surface methods optimize them.

Interaction means the effect of one factor depends on the level of another; main effects alone can mislead.

Piloting an improvement on a small scale validates the change before full deployment.

Improvement selection weighs impact versus effort; high-impact, low-effort changes are prioritized first.

Phase 1 of a control chart establishes limits; Phase 2 monitors against those limits.

Out-of-control signals should trigger a documented reaction plan, not ad hoc adjustment.

Rational subgrouping should capture only common-cause variation within a subgroup.

Process audits and layered process audits verify adherence to the control plan.

Mistake-proofing and standard work reduce reliance on operator vigilance for sustained control.

Control charts require recalculation of limits when the process changes intentionally.

Documentation of lessons learned supports knowledge transfer across the organization.

A closed project has validated benefits, transferred ownership, and a sustainment plan in place.

Deployment planning aligns projects to strategic goals through a portfolio review and project prioritization.

Project selection criteria include financial impact, customer impact, feasibility, and cycle time.

Cost-benefit analysis compares projected savings against implementation costs and payback period.

Change management addresses communication, training, sponsorship, and resistance to sustain improvements.

A balanced scorecard tracks financial, customer, internal process, and learning-and-growth measures.

Benchmarking compares process performance against internal or external best practice.

A lessons-learned repository and belt certification records institutionalize capability.

Leadership engagement is a leading indicator of deployment success; lack of sponsorship is a common failure cause.