Your FREE Six Sigma Green Belt Flashcards 2026 – 250+ Cards

Define, Measure, Analyze, Improve, Control — the core Six Sigma roadmap for improving an existing process.

A data-driven methodology that reduces process variation and defects to improve quality, targeting 3.4 defects per million opportunities.

The Greek letter for standard deviation; in Six Sigma it measures how much a process varies and how many sigmas fit between the mean and the spec limit.

The defect rate of a 'Six Sigma' process — 3.4 defects per million opportunities, allowing for the 1.5σ long-term shift.

Any output that fails to meet a customer requirement or specification.

A unit that contains one or more defects.

Defects Per Million Opportunities — (defects ÷ (units × opportunities)) × 1,000,000; a standardized defect rate.

Defects Per Unit — total defects divided by total units inspected.

The expressed and implied needs, wants, and expectations of customers, gathered to define what quality means.

A measurable characteristic of a product or process whose performance standard must be met to satisfy the customer.

Define, Measure, Analyze, Design, Verify — Design for Six Sigma, used to create a new process or product rather than improve an existing one.

An ongoing effort to make small, incremental improvements to processes, products, or services.

A methodology focused on maximizing customer value while eliminating waste (non-value-added activity).

Defects, Overproduction, Waiting, Non-utilized talent, Transportation, Inventory, Motion, Excess processing.

A step that transforms the product/service in a way the customer is willing to pay for, done right the first time.

A step that consumes resources but adds no value the customer would pay for — a target for elimination.

All the value-added and non-value-added steps required to bring a product or service to the customer.

A lean tool that diagrams the flow of material and information to expose waste and improvement opportunities.

A senior leader who sponsors a Six Sigma project, removes barriers, and secures resources.

An expert who trains and mentors Black Belts and Green Belts and leads the Six Sigma deployment.

A full-time Six Sigma leader who runs complex projects and applies advanced statistical tools.

A part-time practitioner who leads smaller projects and supports Black Belt projects using core DMAIC tools.

A team member with basic Six Sigma awareness who supports projects and process data collection.

The person accountable for a process's performance and for sustaining improvements after the project.

A core, cross-functional process that delivers value across the whole organization.

The total cost of defects — internal failure, external failure, appraisal, and prevention costs.

An organization-wide approach to continuous quality improvement that predates and feeds into Six Sigma.

A method that improves throughput by identifying and managing the single biggest bottleneck (constraint).

The justification for a project: the problem, its cost, the goal, and the expected benefit to the organization.

DMAIC improves an existing process; DMADV (DFSS) designs a new process or product to meet Six Sigma quality.

Pioneered at Motorola in the 1980s and popularized by GE in the 1990s as a quality-improvement methodology.

Conformance to requirements and fitness for use — meeting or exceeding customer expectations.

Plan-Do-Check-Act — Deming's iterative cycle for continuous improvement.

W. Edwards Deming, a quality pioneer known for PDCA, the 14 points, and emphasizing reducing variation.

Joseph Juran, who developed the quality trilogy (planning, control, improvement) and applied the Pareto principle.

The likelihood a customer continues to buy and recommend; a downstream goal of quality improvement.

Internal customers are the next step in your own process; external customers are the end users who buy the output.

A process is the set of activities that produces an output; the product/service is that output.

The rate at which a process produces completed output over time.

The step that limits the throughput of the whole process — the constraint.

The document that defines a project's problem, goal, scope, business case, team, and timeline; it authorizes the project.

A concise, fact-based description of what is wrong, where, when, and how big — with no causes or solutions.

A SMART target for the project metric (e.g., reduce cycle time from 10 to 6 days by Q3).

Specific, Measurable, Achievable, Relevant, Time-bound — the test of a well-written goal statement.

The boundaries of the project — what is and is not included, defining where the process starts and stops.

Suppliers, Inputs, Process, Outputs, Customers — a high-level map that scopes the process in the Define phase.

Classifies customer requirements as basic (must-be), performance (one-dimensional), and delighters (excitement) to prioritize features.

A tool that organizes a large set of ideas or VOC data into natural groupings.

A diagram that translates broad customer needs into specific, measurable CTQ requirements.

Identifying who is affected by or can affect the project and planning how to engage each one.

Maps each task to who is Responsible, Accountable, Consulted, and Informed; exactly one Accountable per task.

Affinity groups unstructured ideas; a tree diagram breaks a goal into progressively detailed sub-tasks.

A management tool that maps cause-and-effect links among many issues to find key drivers.

A tool that weighs options against weighted criteria to choose the best one objectively.

Breaks a broad goal or requirement into successive layers of detail (objectives → tasks).

Shows the strength of relationships between two or more groups of items in a grid.

Shows the sequence and dependencies of project tasks and the critical path.

Maps what could go wrong in a plan and prepares countermeasures.

A bar chart of project tasks against a timeline, showing start, duration, and overlap.

A hierarchical decomposition of project work into manageable deliverables and tasks.

The longest chain of dependent tasks in a project; it sets the shortest possible completion time.

The needs and goals of the organization — cost, growth, profitability — balanced against the VOC.

What the process is actually capable of delivering, shown by its data and control charts.

The key output measure the project aims to improve (the 'big Y'), driven by process inputs (the x's).

The core Six Sigma idea: the output (Y) is a function of the process inputs/factors (x's); fix the x's to fix Y.

Comparing your process performance against best-in-class to set targets and find improvement ideas.

Comparing the expected costs of a project against its expected financial benefits.

Formally ending the project: confirming goals met, documenting results, and handing off to the process owner.

Forming, storming, norming, performing, adjourning — predictable phases of team development.

A structured method where team members generate and silently rank ideas to reach consensus.

A group technique that narrows a large list of options to a few by rounds of voting.

A group method for generating many ideas quickly without early criticism.

Agreed team norms for how members will work together, communicate, and make decisions.

Listing ways to cause a problem, then reversing them into prevention ideas.

Uncontrolled expansion of project boundaries; the charter and sponsor guard against it.

A 30-second summary of the project's problem, goal, and value for stakeholders.

Choosing projects tied to strategy with measurable, achievable goals and clear customer impact.

Surveys, interviews, focus groups, complaints, and observation used to gather VOC.

Critical-to-Quality, Critical-to-Cost, and Critical-to-Delivery characteristics derived from customer needs.

Sponsor/champion, team leader (Belt), members, process owner, and facilitator — each with defined duties.

A decision all team members can support, even if it is not everyone's first choice.

Lists driving and restraining forces for a change to plan how to strengthen or weaken each.

A significant checkpoint or deliverable date in the project schedule.

A tangible, verifiable output the project must produce.

A flowchart showing every step, decision, input, and output of a process as it actually runs.

A diagram using standard symbols to show the sequence of steps and decisions in a process.

A value stream map adds material/information flow and timing data; a process map shows the step sequence.

Variable data measured on a continuous scale (time, weight, length) — more information per data point.

Attribute data counted in categories (pass/fail, number of defects) — less information per point.

Categories with no order (color, machine ID).

Categories with a meaningful order but no fixed interval (small/medium/large, survey ratings).

A population is the entire group of interest; a sample is a subset measured to infer about the population.

Each item has an equal chance of selection, reducing bias in the sample.

Dividing the population into subgroups (strata) and sampling each to ensure representation.

The arithmetic average — the sum of values divided by the count.

The middle value when data is ordered; robust to outliers.

The most frequently occurring value in a data set.

The difference between the largest and smallest values; a simple measure of spread.

The average of the squared deviations from the mean; the square of the standard deviation.

A measure of how spread out data is around the mean (written σ); it equals √variance.

A symmetric, bell-shaped distribution defined by its mean and standard deviation.

In a normal distribution, ≈68% of data lies within ±1σ, ≈95% within ±2σ, ≈99.7% within ±3σ of the mean.

The distribution of sample means approaches normal as sample size grows, regardless of the population shape.

Models the count of successes in a fixed number of independent pass/fail trials.

Models the count of rare events in a fixed interval (e.g., defects per unit).

A bar chart of the frequency of data across intervals; reveals the shape, center, and spread of a distribution.

A graphical summary showing the median, quartiles, and outliers of a data set.

A line plot of data over time used to spot trends, shifts, or cycles.

A plot of two variables to reveal whether and how they are related (correlation).

A study that quantifies how much variation in the data comes from the measurement system itself.

A study of measurement-system Repeatability (same operator) and Reproducibility (different operators).

Variation when the same operator measures the same item multiple times with the same gage.

Variation when different operators measure the same item with the same gage.

How close a measurement is to the true value; bias is a consistent offset from truth.

How consistent repeated measurements are with each other (low spread), regardless of accuracy.

The smallest increment a measurement system can detect.

Whether a measurement system's bias stays consistent over time.

Whether measurement bias is consistent across the full range of the gage.

How well a process meets its specification limits, compared to its natural spread.

Potential capability = (USL − LSL) ÷ 6σ; compares the spec width to the process spread, ignoring centering.

Actual capability = min[(USL − mean), (mean − LSL)] ÷ 3σ; accounts for both spread and centering.

Cp assumes the process is centered; Cpk penalizes off-center processes. Cpk ≤ Cp always.

Long-term performance indices, like Cp/Cpk but using the overall (long-term) standard deviation.

The customer/engineering limits (USL, LSL) that define an acceptable output; set by requirements, not the process.

Control limits come from the process data (the voice of the process); spec limits come from the customer.

The proportion of units that pass without defects.

The fraction of units that complete a step correctly the first time, without rework.

The product of the first-time yields of every step; the probability a unit passes the whole process defect-free.

The number of standard deviations between the process mean and the nearest spec limit.

The current process performance captured before improvements, used to prove the change worked.

A precise, agreed definition of what is measured and how, so data is consistent across people.

A simple structured form for collecting and tallying data in real time.

A small sample of items collected together under similar conditions for a control chart.

Systematic error from a sample that does not represent the population.

The standard deviation divided by the mean; a unitless measure of relative variation.

Values dividing ordered data into four equal parts (Q1, Q2=median, Q3).

Q3 − Q1; the spread of the middle 50% of the data, robust to outliers.

A measure of how asymmetric a distribution is around its mean.

A measure of how heavy-tailed or peaked a distribution is compared to normal.

The number of standard deviations a value is from the mean: (x − mean) ÷ σ.

Models the time between independent events occurring at a constant rate.

An MSA for discrete data that checks whether appraisers rate items consistently and correctly.

The percentage of total variation consumed by the measurement system; under 10% is generally acceptable.

An MSA metric of how many groups the gage can reliably tell apart; ≥5 is desired.

Convert DPMO to a Z (sigma) value using a normal table or DPMO-to-sigma conversion.

Any chance for a defect to occur on a unit; used to compute DPMO.

A plan defining what data to collect, how, by whom, and how often, with operational definitions.

The process of finding the fundamental cause of a problem rather than treating its symptoms.

Asking 'why?' repeatedly (about five times) to drill from a symptom down to the root cause.

A cause-and-effect diagram that organizes potential causes by category (the 6 Ms) around a central spine.

Cause categories on a fishbone: Methods, Machines, Materials, Measurement, Manpower (people), Mother Nature (environment).

A bar chart ordering causes by frequency to highlight the 'vital few' that drive most of the problem.

Roughly 80% of effects come from 20% of causes; focus effort on the vital few.

A statistical relationship in which two variables move together; it does not prove causation.

A value from −1 to +1 measuring the strength and direction of a linear relationship between two variables.

A method that models how a response variable changes as one or more input variables change.

Models the response Y as a straight-line function of one predictor X: Y = b₀ + b₁X.

Points trending up = positive correlation; down = negative; no pattern = little/no correlation.

A method to decide, with a stated risk, whether sample data supports a claim about a population.

The default claim of no difference or no effect, which the test tries to disprove.

The claim that there is a difference or effect — what you conclude if you reject H₀.

The probability of seeing data this extreme if H₀ were true; if p ≤ α, reject H₀.

The significance level — the accepted probability of a Type I error, often 0.05.

Rejecting a true null hypothesis — a 'false positive'; its probability is α.

Failing to reject a false null hypothesis — a 'false negative'; its probability is β.

A range, computed from data, that likely contains the true population parameter at a stated confidence (e.g., 95%).

A hypothesis test comparing means when the population standard deviation is unknown / samples are small.

Analysis of Variance — tests whether the means of three or more groups differ significantly.

Tests whether observed counts of categorical data differ from expected counts (e.g., independence).

Failure Mode and Effects Analysis — a structured way to identify failure modes and prioritize them by risk.

Severity × Occurrence × Detection in an FMEA; higher RPN = higher priority to address.

The three FMEA ratings (each 1–10): how bad, how likely, and how hard to catch a failure is.

A graphical study that classifies variation as within-piece, piece-to-piece, or time-to-time to localize its source.

When the effects of two factors cannot be separated, obscuring which one drives the response.

Correlation means two variables move together; causation means one drives the other — correlation alone never proves it.

Using charts (histograms, box plots, scatter, Pareto) to explore data before formal statistics.

Distinguishing common-cause (inherent) from special-cause (assignable) variation in the data.

A result can be statistically significant yet too small to matter practically — judge both.

When predictor variables in a regression are themselves correlated, distorting their estimated effects.

The difference between an observed value and the value predicted by a regression model.

The proportion of variation in the response explained by the regression model (0 to 1).

The probability of correctly rejecting a false null hypothesis (1 − β).

The probability of a Type II error — failing to detect a real effect.

The number of observations needed to detect an effect at a chosen risk and power.

Compares the means of two related measurements (e.g., before vs after on the same units).

Compares the means of two independent groups.

Compares two variances or tests overall significance in ANOVA/regression.

A table of categorical counts used in a chi-square test of independence.

Tests (chi-square, F) that compare process spread rather than averages.

Confirming a suspected root cause with data before fixing it, not just by opinion.

Side-by-side box plots used to compare distributions of groups quickly.

A structured method of changing input factors deliberately to learn their effect on the output.

An input variable deliberately changed in an experiment (e.g., temperature).

A specific setting of a factor in an experiment (e.g., 100°C and 150°C).

The output measured in an experiment to judge the effect of the factors.

The average change in the response caused by changing one factor from low to high.

When the effect of one factor on the response depends on the level of another factor.

An experiment testing every combination of factor levels to estimate all effects and interactions.

A DOE using a carefully chosen subset of runs to study many factors economically.

Changing one input at a time; inefficient and misses interactions — DOE is preferred.

A focused, short (often week-long) team effort to rapidly improve a specific process.

Sort, Set in order, Shine, Standardize, Sustain — a lean method for an organized, efficient workplace.

A device or design that prevents or immediately detects errors so defects cannot pass downstream.

A visual signal that pulls work or replenishes inventory only when needed, limiting overproduction.

Production triggered by actual downstream demand rather than a forecast (push).

Producing or delivering only what is needed, when it is needed, in the amount needed.

A lean method to drastically reduce equipment changeover/setup time.

The documented, current best way to perform a task, ensuring consistency and a baseline to improve.

The pace of customer demand = available production time ÷ customer demand; sets the rhythm of production.

The time to complete one unit or one cycle of a process step.

The total elapsed time from a customer request to delivery.

Improve throughput by exploiting and elevating the bottleneck before optimizing elsewhere.

Arranging equipment and workstations by product flow (a cell) to cut transport and waiting.

A small-scale trial of a proposed solution to confirm it works before full rollout.

Weighing each candidate improvement's expected benefit against its cost and risk to choose the best.

A weighted matrix that scores candidate solutions against criteria like impact, cost, and effort.

A program that involves operators in maintaining equipment to reduce breakdowns and defects.

A drawing of the physical path of a product or person to expose wasted motion and transport.

Using visual signals (boards, color, labels) so status and abnormalities are obvious at a glance.

Smoothing production volume and mix to reduce batching, inventory, and overburden.

Running experimental trials in random order to spread out the effect of unknown influences.

Repeating experimental runs to estimate experimental error and improve precision.

Grouping experimental runs to remove the effect of a known nuisance variable.

Extra runs at the mid-level of factors used to detect curvature in the response.

An advanced DOE that models curvature to find optimal factor settings.

A fractional design used to identify the few significant factors from many.

The 'real place' where work happens; lean practitioners go to gemba to observe directly.

A self-contained group of resources arranged for one-piece flow of a product family.

Moving a single unit through steps without batching, cutting WIP and lead time.

Cutting changeover time so smaller batches and more flexibility become economical.

Prevention (stops the error) and detection (flags the error) poka-yoke devices.

A detailed plan of tasks, owners, and dates to roll out the chosen solution.

A document specifying how each key process input/output will be monitored and what to do if it goes out of control.

Using control charts to monitor a process over time and distinguish common- from special-cause variation.

A time-ordered plot with a center line and upper/lower control limits used to detect special-cause variation.

Lines at ±3σ from the center line; points beyond them signal a special cause.

The process average (or target) on a control chart, between the control limits.

Inherent, random variation present in a stable process; do not react to individual points.

Variation from an identifiable, assignable source that makes a process unstable; investigate and remove it.

A process showing only common-cause variation — stable and predictable.

A process showing special-cause variation — a point beyond the limits or a non-random pattern.

Reacting to common-cause variation as if it were special; it increases variation.

A control chart pair for continuous data in subgroups: X-bar tracks the average, R tracks the range (spread).

Like X-bar & R but uses the standard deviation (S) for spread; preferred for larger subgroups.

A control chart for continuous data collected one point at a time (subgroup size of 1).

An attribute control chart for the proportion defective with varying sample sizes.

An attribute control chart for the number defective with a constant sample size.

An attribute control chart for the count of defects per unit with a constant sample size.

An attribute control chart for defects per unit with varying sample size.

Variables charts (X-bar/R, I-MR) plot continuous data; attribute charts (p, np, c, u) plot counts.

Choose the chart by data type (continuous vs attribute) and subgroup size.

Pattern rules (e.g., a run of 7, points beyond 3σ) that flag a likely special cause.

A small sample chosen so variation within it is only common-cause, isolating special-cause between subgroups.

A slow, gradual shift of the process mean over time, often caught by control charts.

Embedding gains via control plans, standard work, training, and monitoring so the process does not regress.

The documented, approved method for performing a task to keep the improved process consistent.

An Out-of-Control Action Plan: predefined steps to take when a control chart signals a problem.

A periodic check that the improved process and its controls are being followed and still work.

Poka-yoke controls that hold the gains by preventing the defect from recurring.

Transferring the controlled process and its documentation to the process owner at project close.

A simple stoplight technique using zones around the target to decide whether to keep running or adjust.

Documented insights from the project used to improve future projects and share knowledge.

A visual display of key process metrics over time used to monitor sustained performance.

A control plan, updated SOPs, training, monitoring charts, and a documented handoff to the owner.

Larger subgroups make X-bar charts more sensitive to small shifts in the mean.

Seven consecutive points on one side of the center line signals a non-random pattern.

Several consecutive increasing or decreasing points signals a possible special cause.

Patterns within the A/B/C sigma zones of a control chart that flag instability.

Control limits are recomputed when a verified, permanent process change occurs.

An out-of-control signal when no special cause exists — a Type I error on the chart.

Recompute Cp/Cpk only once the process is in statistical control, never before.

Ensuring operators are trained on the new standard work so improvements hold.

Assigning each control metric an owner responsible for monitoring and acting on it.

Spreading a proven solution to similar processes or sites to multiply the benefit.