CSSBB cheat sheet

Hoshin Kanri (policy deployment) cascades vision to 3-5 year objectives to annual targets to projects, using an X-matrix that links objective, initiative, owner, and metric.

Balanced scorecard perspectives: financial, customer, internal process, and learning and growth; lagging financial results are paired with leading drivers.

COPQ = prevention + appraisal + internal failure + external failure costs; failure costs usually dominate and are the target of Six Sigma projects.

Hard savings change the profit-and-loss statement (labor, scrap, energy); soft savings are avoided or capacity costs and require finance validation.

ROI (%) = (net benefit / project cost) x 100; NPV discounts future cash flows at the hurdle rate; payback period = investment / annual cash flow.

Benchmarking types: internal, competitive, functional, and generic; the performance gap = best-in-class result minus current result.

Deployment roles: executive leadership, deployment champion, master black belt (technical leader), black belt (full-time project leader), green belt (part-time), process owner (sustains gains).

Planning sequence: SWOT/PEST plus VOC leads to mission, vision, and goals, then Hoshin deployment, then project selection, then tollgate and dashboard review.

Change management elements: readiness assessment, stakeholder power/interest analysis, communication plan, force-field analysis, and reinforcement of new behavior.

The process view cuts horizontally across functions; processes are classified as core (customer value), support, and management/leadership.

SIPOC = Suppliers, Inputs, Process (5-7 high-level steps), Outputs, Customers; used to scope a process before detailed mapping.

Little's Law: cycle time = WIP / throughput, equivalently WIP = throughput x cycle time.

Takt time = available production time / customer demand; it sets the required pace of the process.

DPU = defects / units; DPO = defects / (units x opportunities); DPMO = DPO x 1,000,000.

RTY = FPY1 x FPY2 x ... x FPYn; when defects are Poisson, RTY is approximately e raised to the negative DPU.

Sigma level from DPMO with the 1.5 sigma shift: 6 sigma = 3.4 DPMO, 5 sigma = 233 DPMO, 4 sigma = 6,210 DPMO, 3 sigma = 66,807 DPMO.

Cp = (USL - LSL) / 6 sigma-within; Cpk = min[(USL - xbar), (xbar - LSL)] / 3 sigma-within; Pp and Ppk substitute the overall standard deviation.

OEE = availability x performance x quality; Kano categories are must-be, performance, attractive, indifferent, and reverse; QFD's house of quality converts VOC into technical requirements.

Tuckman stages: forming, storming, norming, performing, adjourning; conflict peaks during storming and is a normal part of development.

Thomas-Kilmann conflict modes: competing, collaborating, compromising, avoiding, accommodating; collaborating suits high-stakes, high-trust issues.

RACI: exactly one Accountable per activity; Responsible does the work, Consulted gives input, Informed is notified.

Nominal group technique sequence: silent idea writing, round-robin recording, clarification, then rank voting; it reduces dominance and groupthink.

Multivoting narrows a long list over successive rounds; consensus means every member can support the decision, not that all prefer it.

Charter topics for a team: purpose, scope, deliverables, roles, milestones, decision rules, ground rules, and communication plan.

Typical ground rules: attendance and punctuality, one conversation at a time, confidentiality, data-based decisions, and no-blame discussion.

Virtual team enablers: explicit goals and roles, a set meeting cadence, a shared document repository, written decision logs, and rotating meeting times.

Team launch logistics: sponsor kickoff, charter review, role clarification, meeting schedule, minutes, and an action-item log.

Charter components: business case, problem statement, goal statement, scope in/out, milestones, team and roles, sponsor, and expected benefits.

Problem statement quantifies the gap using what, where, when, and how much, and contains no cause, no blame, and no solution.

Goal statement format: metric + baseline + target + deadline + direction, for example reduce defect rate from A to B by a stated date.

VOC to CTQ chain: customer need leads to a quality driver, which leads to a measurable characteristic with an operational definition.

PERT expected time = (O + 4M + P) / 6; activity variance = [(P - O) / 6] squared.

The critical path has zero float and fixes the shortest possible project duration; crashing adds resources to shorten it.

Stakeholder power/interest grid actions: manage closely, keep satisfied, keep informed, monitor.

DMAIC tollgates end each phase with a deliverable review and sponsor approval before the next phase starts.

Pareto logic: a small number of causes produce most of the effect, so Pareto is used to prioritize before root cause analysis.

Accuracy (bias) is closeness to the true value; precision is repeatability; a gauge can be precise yet biased.

Gage R&R decision rule by percent study variation: below 10% acceptable, 10% to 30% marginal and application-dependent, above 30% unacceptable.

Repeatability equals equipment variation (EV); reproducibility equals appraiser variation (AV); GR&R combines both with part variation.

Rule of ten: measurement resolution should be no more than one-tenth of the process variation or of the tolerance.

Sample size for a mean: n = (z x sigma / delta) squared; for a proportion: n = z squared x p(1-p) / delta squared.

Sampling methods: simple random, stratified, systematic (every kth unit), and cluster; SPC uses rational subgroups.

Baseline capability: Z = min[(USL - xbar), (xbar - LSL)] / sigma; Zbench is derived from the overall process yield.

Continuous data carry more information per observation than discrete data; discrete data require larger samples to detect the same shift.

The data collection plan specifies what, how, who, when, where, and how many for each measure, plus the operational definition.

Test logic: assume the null hypothesis; if p-value is less than alpha, reject the null; alpha is Type I error, beta is Type II error, and power = 1 - beta.

t-tests compare one or two means (paired or independent, equal or unequal variances); ANOVA compares three or more means with F = MS-between / MS-within.

Chi-square tests goodness-of-fit and independence for discrete data.

R-squared is the proportion of response variation explained by the model; adjusted R-squared penalizes unnecessary terms.

Regression assumptions are checked on residuals: normal, mean zero, constant variance, and independent.

Correlation r ranges from -1 to +1; r near zero means no linear relationship, and correlation does not establish causation.

A multi-vari study separates within-piece, piece-to-piece, and time-to-time variation to identify the dominant family of variation.

FMEA risk priority number = severity x occurrence x detection, each rated 1 to 10; act on high RPNs, high severity, or high detection ratings.

For non-normal data, transform (log or Box-Cox) or use nonparametric tests such as Mann-Whitney or Kruskal-Wallis.

A full factorial 2-to-the-k design needs 2-to-the-k runs; a fractional factorial 2-to-the-(k-p) reduces runs at the cost of confounding.

Resolution III confounds main effects with two-factor interactions; resolution IV confounds main effects with three-factor interactions; resolution V keeps main effects clear of two-factor interactions.

DOE pillars are randomization, replication, and blocking; these protect against lurking variables and provide an estimate of experimental error.

Response surface designs (central composite, Box-Behnken) model curvature and locate optimum factor settings.

Eight wastes: transport, inventory, motion, waiting, overproduction, overprocessing, defects, and unused talent (TIMWOODS).

5S sequence: sort, set in order, shine, standardize, sustain; sustain requires audits and habit to hold the gain.

SMED converts internal setup (machine stopped) into external setup (machine running); the target is changeover time under ten minutes.

Poka-yoke devices are either prevention (the error cannot be made) or detection (the error is caught immediately at the source).

TPM pillars include autonomous maintenance, planned maintenance, and focused improvement, aimed at zero breakdowns and higher OEE.

Theory of constraints sequence: identify the constraint, exploit it, subordinate everything else to it, elevate it, then repeat.

Chart selection: continuous data use Xbar-R (n = 2 to 9) or Xbar-S (n of 10 or more); individual measurements use I-MR; attribute data use p, np, c, or u.

p = fraction defective with variable subgroup size; np = number defective with constant size; c = defects per unit with constant opportunity; u = defects per unit with variable opportunity.

Control limits are plus or minus three sigma from the centerline and are computed from the process; specification limits come from the customer and are never drawn as control limits.

Out-of-control signals include 1 point beyond 3 sigma, 2 of 3 beyond 2 sigma, 4 of 5 beyond 1 sigma, 8 in a row on one side, 6 trending, and 14 alternating.

Cp and Cpk describe capability relative to specification; control charts describe stability over time, and capability is only meaningful for a stable process.

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

Reaction plan steps: contain the affected output, identify and correct the cause, verify the correction, and record the event.

Sustainment tools: standardized work and SOPs, visual controls, poka-yoke, layered process audits, and a fixed SPC review cadence.

The Control phase transfers ownership to the process owner with documented procedures, training, and a monitoring dashboard.