CSSC MBB key terms

A structured process for collecting what customers need and value, then translating those statements into measurable requirements a project can act on. It anchors scope and success criteria to real customer priorities.

A measurable characteristic derived from customer needs that defines what must go right for the customer to be satisfied. CTQs convert vague VOC statements into specifications with targets and tolerances.

A framework that classifies features as must-be, performance, or attractive based on how satisfaction responds to their presence. It helps teams avoid over-investing in basics while finding differentiators.

A planning method that maps customer requirements to design characteristics, then cascades priorities through subsystems, parts, and process controls. The house of quality matrix makes tradeoffs explicit and traceable.

The total cost incurred because work is done wrong, including scrap, rework, warranty, inspection, and lost goodwill. Quantifying it builds the business case and prioritizes improvement projects.

The five-phase improvement cycle: Define, Measure, Analyze, Improve, Control. Each phase has deliverables and a tollgate review that confirms the team is ready to proceed.

The five-phase design cycle: Define, Measure, Analyze, Design, Verify. It is used when no existing process can meet requirements and a new design is required.

A methodology for designing new products and processes so capability is built in from the start rather than improved later. It emphasizes requirement flow-down, predictive modeling, and verification against CTQs.

A document that authorizes a project and states the problem, goal, scope, business case, team, milestones, and constraints. It is negotiated with the sponsor and prevents scope drift.

A high-level map showing Suppliers, Inputs, Process, Outputs, and Customers for a process. It defines boundaries and highlights where measurement or ownership is unclear.

A formal checkpoint at the end of each DMAIC or DMADV phase where the sponsor and belt confirm deliverables are complete and valid before releasing resources for the next phase.

A set of principles and tools aimed at eliminating waste and improving flow by producing only what the customer pulls. It complements Six Sigma by shortening lead time while reducing variation.

A visual map of material and information flow for a product family, with cycle times, inventory, and wait times annotated. It exposes where value is added and where lead time is consumed.

Available production time divided by customer demand over the same period. It sets the pace a process must meet so output matches demand without overproduction.

A relationship stating that average work in process equals throughput multiplied by average flow time. It lets teams predict how inventory reductions shorten lead time.

Continuous, incremental improvement carried out by the people who do the work, often through structured rapid events. It focuses on small changes sustained over time rather than one large redesign.

A workplace organization method: sort, set in order, shine, standardize, and sustain. It removes clutter and establishes visual standards that make abnormal conditions obvious.

A mistake-proofing device or design that prevents an error from occurring or makes it immediately detectable. Effective poka-yokes are inexpensive, reliable, and part of the process, not an added inspection step.

Documented, current best practice for how a task is performed, including sequence, timing, and required inventory. It provides the baseline against which improvements are measured and sustained.

The time between completion of one unit and completion of the next by a given process step. Comparing cycle time to takt time reveals bottlenecks and capacity gaps.

A study that quantifies how much variation in observed data comes from the measurement system rather than the process. It is done before trusting data for capability or hypothesis testing.

A study separating measurement variation into repeatability (same appraiser, same part) and reproducibility (different appraisers). Percent study variation or percent tolerance indicates whether the system is acceptable.

The variation observed when one appraiser measures the same item multiple times with the same instrument. It reflects equipment and method consistency rather than differences between people.

The variation between appraisers measuring the same items with the same instrument. Large reproducibility points to training, technique, or operational-definition problems.

The difference between the average of many measurements and the reference or true value. It is a systematic offset, correctable by calibration, unlike random variation.

How bias changes across the operating range of the measurement instrument. A gauge may be accurate near one end of its range and biased near the other.

A study of agreement among appraisers and against a known standard when judgments are categorical, such as pass/fail or defect type. Kappa statistics summarize agreement beyond chance.

A precise, unambiguous description of how a characteristic is measured, including instrument, method, and decision criteria. It ensures different people classify the same item the same way.

Indices comparing process spread and centering to specification limits. Cp uses spread only, while Cpk accounts for how far the mean sits from the nearer limit.

Capability-like indices computed with the overall standard deviation rather than within-subgroup variation. They describe how the process actually performed and are less sensitive to subgroup assumptions.

The number of standard deviations between the process mean and the nearest specification limit, often stated with a 1.5-sigma long-term shift convention. Higher values indicate fewer defects per million opportunities.

Total defects divided by total defect opportunities, multiplied by one million. It normalizes quality across processes with different complexity and converts directly to a sigma level.

The probability that a unit passes every process step without a defect, computed as the product of each step's first-pass yield. It exposes hidden factory losses that final inspection misses.

The upper and lower bounds on a characteristic set by design or the customer, not by the process. They define acceptability and must never be confused with control limits.

A time-ordered plot of a statistic with control limits computed from process variation. It distinguishes common-cause from special-cause variation and guides whether to act on the process.

Statistically derived boundaries, typically three standard deviations from the center line, describing expected variation of the plotted statistic. They come from the data, not from specifications.

The deliberate choice of samples within which only common-cause variation should exist, so that shifts between subgroups are detectable. Poor subgrouping inflates or masks control-chart signals.

Sensitivity rules that flag patterns such as points beyond three sigma, runs of eight on one side, or trends on a control chart. They detect small shifts that a single-point rule misses.

A procedure that compares observed data to a stated null hypothesis and quantifies how likely the result would be if that hypothesis were true. The choice of test depends on data type and assumptions.

The probability of observing a result at least as extreme as the data if the null hypothesis is true. It is not the probability that the null hypothesis is true.

Rejecting a true null hypothesis, also called alpha or producer's risk. Its probability is set by the significance level chosen before the test.

Failing to reject a false null hypothesis, also called beta or consumer's risk. It increases when samples are small or when the true effect is close to the null value.

The probability of correctly detecting a real effect, equal to one minus beta. It depends on effect size, variability, sample size, and the chosen alpha.

The number of observations needed to detect an effect of practical interest with specified confidence and power. It is calculated before data collection, not after.

A method that tests whether means of three or more groups differ by comparing between-group to within-group variation. It generalizes the two-sample t-test and underlies many designed experiments.

A model that relates one or more predictors to a continuous response and estimates the size and significance of each effect. Residual analysis checks whether model assumptions hold.

A planned set of runs that varies factors together so main effects and interactions can be estimated efficiently. It is more informative than changing one factor at a time.

An experiment in which every combination of factor levels is run, allowing estimation of main effects and interactions. Full factorials grow quickly; fractional factorials trade resolution for fewer runs.

The mixing of effects so that two or more influences cannot be separated in the analysis. In fractional factorials some confounding is accepted by design; in observational data it must be addressed.

A sequence of designs and models used to find factor settings that optimize a response, often after screening. It fits curved models and uses steepest ascent or desirability functions to locate the optimum.

A condition in which predictor variables are highly correlated, making individual coefficient estimates unstable and hard to interpret. It can be diagnosed with variance inflation factors.

A document listing the process steps, characteristics to monitor, measurement method, sample size and frequency, control method, and reaction plan when out of control. It sustains gains after the project closes.

The ongoing use of control charts and capability studies to keep a process stable and capable. It shifts attention from detecting bad output to monitoring the process that produces it.

A documented review of what worked and what did not during a project, shared so future teams avoid repeated mistakes. It is captured before team disbanding, while knowledge is still available.

A statistic measuring agreement between raters or between a rater and a standard after correcting for chance agreement. Values near one indicate strong agreement; near zero indicates chance-level agreement.

Comparing an instrument's readings to a known reference standard and adjusting it to reduce bias. It addresses systematic error but does not by itself improve repeatability.

The quantified justification for a project, including expected financial and non-financial benefits, costs, and risks. It ties the improvement to strategic goals and is revisited at tollgates.

A step-by-step diagram of how a process actually operates, including decision points, handoffs, and rework loops. It is used to find non-value-added steps and measurement points.