CSSC Green Belt key terms

The structured collection and analysis of what customers need, want, and expect from a product or service, including needs they never state. It converts customer language into requirements a project can measure and act on.

A product or process characteristic that measurably affects what the customer values, derived from customer input. CTQs translate broad wants into specific, testable requirements with defined targets and specification limits.

A document that formally authorizes a Six Sigma project and states the problem, scope, goal, team, timeline, and expected benefit. It is the team's agreement with the sponsor and changes only with sponsor approval.

A short, factual description of a performance gap: what is wrong, where and when it occurs, and how large the gap is. It contains data, avoids naming a cause, and does not propose a solution.

A one-page view of a process showing its Suppliers, Inputs, Process steps, Outputs, and Customers. Teams use it at the start of a project to agree on scope and boundaries before detailed mapping.

A diagram of the steps a process actually follows, including decisions, handoffs, and rework loops. The deployment version also shows who performs each step, which exposes delays and non-value-adding work.

The shorthand for a process relationship: the output Y is a function of input and process variables x. Green Belts define and measure Y first, then use data to find which x variables drive it.

The financial and strategic justification for a project, such as cost reduction, capacity gained, risk lowered, or customer retention improved. It explains why the organization should spend resources on this effort.

A scheduled checkpoint at the end of each DMAIC phase where the team presents its deliverables and evidence to sponsors and reviewers. Passing the gate confirms quality and releases resources for the next phase.

A precise written description of how a characteristic is measured: the instrument, the procedure, and what counts and does not count. It makes data consistent across people, shifts, and sites.

A set of studies that estimates how much of the variation in data comes from the measurement system rather than the process. It examines resolution, bias, stability, repeatability, and reproducibility before data are trusted.

A study that separates measurement variation into repeatability (one person measuring the same item twice) and reproducibility (different people measuring the same items). Results are judged against total or process variation.

A document that specifies what data will be collected, why, from which sources, by whom, how often, and how it will be recorded and verified. It prevents teams from analyzing whatever data happen to be available.

Choosing a subset of units, transactions, or time periods to represent a larger population. Random selection is what allows conclusions from the sample to be extended to the population with known confidence.

The measured level of process performance before any improvement is made. A trustworthy baseline uses valid data over a representative period so that later gains can be proven rather than merely asserted.

Indices that compare the spread of a stable process with its specification limits. Cp uses only the spread, while Cpk also reflects how far the process mean sits from the nearest specification limit.

A normalized defect rate calculated as defects divided by the product of units and opportunities per unit, then multiplied by one million. It allows processes of differing complexity to be compared on a single scale.

The probability that a unit passes through every step of a process without a defect, obtained by multiplying the first-pass yields of all steps. It exposes hidden rework that an average final yield conceals.

A statistical check, such as Anderson-Darling or Shapiro-Wilk, of whether sample data are consistent with a normal distribution. The outcome guides which capability indices and hypothesis tests are appropriate.

The underlying, controllable reason a defect or variation occurs, which if removed would stop the problem from recurring. It is confirmed with data rather than assumed from experience or opinion.

A team diagram that sorts possible causes of a problem into branches such as people, machine, method, material, measurement, and environment. It widens the search for causes before any are tested with data.

A statistical procedure that weighs sample evidence against a stated assumption about a population. The result is a probability used to decide whether an observed difference is real or plausibly due to chance.

The probability of seeing a result at least as extreme as the one observed if the null hypothesis were true. It is not the probability that the null hypothesis itself is correct.

A Type I error is rejecting a true null hypothesis, a false alarm controlled by alpha. A Type II error is failing to detect a real effect, controlled by beta; power equals one minus beta.

A test that compares the means of three or more groups by splitting total variation into between-group and within-group portions. A significant result indicates that at least one group mean differs.

A method for modeling the relationship between one or more input variables and a continuous output, producing an equation plus measures of fit and significance. It supports prediction and estimates the size of each input's effect.

A structured team review that lists how a process or product could fail, then rates each failure's severity, likelihood of occurrence, and chance of detection to set improvement priorities by risk priority number.

A planned set of trials in which several input factors are varied together so that their individual effects and interactions on the output can be estimated efficiently, with fewer runs than one-factor-at-a-time testing.

An iterative improvement cycle: plan a change and how it will be measured, do it on a small scale, check the result against expectation, then act by standardizing or adjusting. It suits rapid learning at limited risk.

A design feature or device that either prevents an error from being made or makes it obvious immediately, so a defect cannot pass to the next step. Good poka-yokes work without relying on operator attention.

Continuous, incremental improvement driven by the people who do the work, often through short focused events aimed at removing waste, simplifying steps, and shortening cycle time.

A diagram of all steps, both value-adding and not, plus the information and material flows needed to deliver a product or service. It shows where time is spent and where waste and delay accumulate.

A limited trial of a proposed change, run on a small scale or in a single area, to confirm the expected benefit and reveal side effects before full deployment across the organization.

A document listing what will be monitored, the measurement method, sample size and frequency, control limits or acceptance criteria, and the action to take when a signal occurs. It sustains the gain and transfers ownership.

The use of control charts and decision rules to separate routine common-cause variation from special-cause signals, so the team reacts only when the process has genuinely changed.

A time-ordered plot of a process statistic with a center line and statistically calculated control limits. Points outside the limits or non-random patterns indicate special-cause variation that needs investigation.

Control limits are calculated from the process's own variation and describe what it is doing. Specification limits come from the customer and describe what is required; capability indices compare the two.

A pre-agreed checklist telling the operator or process owner what to check and do when a control chart shows a special-cause signal, so responses are fast, correct, and consistent.

The person accountable for a process's ongoing performance after the project closes, responsible for maintaining the control plan, keeping documentation current, and sustaining the gains.