1.2.5 Basic Six Sigma Metrics

Basic Six Sigma Metrics Introduction to Six Sigma Metrics Six Sigma metrics quantify how well a process meets customer requirements and how consistently it performs over time. They translate defects, variation, and performance into numbers that can be compared, monitored, and improved. At the basic level, Six Sigma metrics focus on: - How often defects occur - How performance compares to specifications - How stable and capable a process is - How performance changes over time The central family of basic metrics includes: - Defects and defect rates (DPU, DPO, DPMO) - Yield (first pass yield and rolled throughput yield) - Sigma level (short-term and long-term) - Capability indices (Cp, Cpk, Pp, Ppk) - Basic performance measures (cycle time, lead time, throughput) The sections below build these concepts carefully and show how they interrelate. --- Units, Defects, and Opportunities Defining Units, Defects, and Opportunities At the core of Six Sigma metrics is a precise definition of what is being measured. - Unit One item processed by the system. It could be a product, a transaction, a report, a call, or any clearly defined work item. - Defect Any failure to meet a specified requirement. A single unit may contain multiple defects. - Defect opportunity A distinct chance for a defect to occur on a unit, based on defined requirements. This must be: - Clearly defined - Measurable - Mutually exclusive (no overlap between opportunities) For example, a loan application (unit) might have: - 1 opportunity for wrong customer information - 1 opportunity for wrong interest rate - 1 opportunity for missing signature Total opportunities per unit = 3. Counting Defects and Opportunities Accurate counts are essential before calculating metrics. - Total units (N): count of items examined - Total defects (D): total number of defects found across all units - Opportunities per unit (O): number of possible defect locations per unit - Total opportunities: N × O A clear operational definition for each defect type and opportunity prevents inconsistent counting and misleading metrics. --- Defect-Based Metrics: DPU, DPO, DPMO Defects per Unit (DPU) DPU shows how many defects occur on average per unit. - Formula DPU = D / N Where: - D = total number of defects - N = total number of units inspected Interpretation: - DPU = 0 means no defects were observed - Higher DPU = more defects per unit on average - DPU can exceed 1 when multiple defects per unit are common DPU is useful when: - The number of opportunities per unit is constant or not clearly defined - You want a direct sense of defects per item handled Defects per Opportunity (DPO) DPO standardizes defect counts by considering how many opportunities exist for defects. - Formula DPO = D / (N × O) Where: - O = opportunities per unit Interpretation: - DPO is a proportion between 0 and 1 - DPO represents the probability that a single opportunity is defective DPO allows comparison across: - Different products - Different processes - Different complexity levels (with distinct opportunities per unit) Defects per Million Opportunities (DPMO) DPMO scales DPO to a more intuitive metric: defects per one million opportunities. - Formula DPMO = DPO × 1,000,000 or directly: DPMO = [D / (N × O)] × 1,000,000 Interpretation: - DPMO ≈ 0: extremely few defects relative to opportunities - Higher DPMO: poorer quality - Often used to link performance to sigma level DPMO is one of the most commonly reported Six Sigma metrics because it: - Normalizes performance for complexity - Supports sigma level conversion - Allows benchmarking across industries and processes --- Yield Metrics: First Pass Yield and Rolled Throughput Yield First Pass Yield (FPY) FPY measures the proportion of units that pass through a step or process without needing rework, repair, or retreatment. - Formula (single step) FPY = (Good units out without rework) / (Total units entering the step) Interpretation: - FPY focuses on “right first time” - Reworked units are counted as defects for FPY, even if ultimately accepted FPY by step highlights where: - Most rework occurs - Process inefficiencies are concentrated Rolled Throughput Yield (RTY) RTY measures the probability that a unit passes through the entire process with no defects at any step, the first time. For multiple steps, each with FPYᵢ: - Formula RTY = FPY₁ × FPY₂ × … × FPYₙ Alternative form using DPU per step: - FPY for step i = e^(−DPUᵢ) when defects follow a Poisson assumption - Then RTY ≈ e^(−Σ DPUᵢ) Interpretation: - RTY is usually lower than any individual step’s FPY - RTY highlights the cumulative impact of defects across the whole process - Low RTY indicates that customers experience many combined chances of defects Yield metrics connect directly to cost, rework, and customer experience. --- Sigma Level as a Performance Metric Concept of Sigma Level Sigma level expresses how well a process is performing relative to its specifications in units of standard deviation. Conceptually: - Higher sigma level = fewer defects - Sigma level compares process mean and variation to specification limits There are two main viewpoints: - Short-term sigma Based on current, within-subgroup variation (less influenced by shifts and drifts). - Long-term sigma Incorporates the effect of process shifts over time (commonly approximated by subtracting 1.5 sigma from short-term). The commonly cited “Six Sigma” performance corresponds to: - About 3.4 defects per million opportunities (long-term), assuming a 1.5 sigma shift. Computing Sigma Level from DPMO In many practical settings, sigma level is derived from DPMO. 1. Convert DPMO to a defect probability: p = DPMO / 1,000,000 1. Interpret p as the proportion nonconforming (defective opportunities). 1. Look up the corresponding Z-value (standard normal deviate) for the non-defective proportion (1 − p). 1. Adjust for the assumed long-term shift if applicable: - Short-term Z ≈ Long-term Z + 1.5 In calculators or tables, these steps are embedded. Conceptually, sigma level is just the Z-score associated with process performance. Sigma Level and Specification Limits Where continuous data and specifications exist, sigma level can be computed directly from process parameters, using capability indices (see later section): - Short-term sigma level ≈ Zbench calculated from: - Distance from mean to nearest spec limit - Divided by short-term standard deviation Sigma level is then another expression of how much of the distribution falls outside specifications. --- Defects, Defectives, and Yield Relationship Defect vs Defective Unit Understanding the difference matters for selecting correct metrics. - Defect A single nonconformance (e.g., scratch, missing field). - Defective unit A unit that has one or more defects that cause it to fail requirements. Key points: - DPU and DPMO are based on defects - Yield and proportion nonconforming are based on defectives (units failing) A process may: - Have many minor defects per unit but still meet key specifications - Have few defects, but each is critical, making many units defective Relationship Between Defects and Yield Under a Poisson assumption for defects per unit: - Let DPU = λ (average defects per unit) Then: - Probability a unit has zero defects ≈ e^(−λ) This is an estimate of yield when any defect makes a unit defective. Thus: - Yield ≈ e^(−DPU) - DPU ≈ −ln(Yield) This relationship connects: - Defect-based measures (DPU, DPMO) - Unit-based measures (yield, defective rate) It is especially useful when: - Estimating RTY from DPU values across multiple steps - Moving between defect counts and customer-level impact --- Process Capability Metrics: Cp and Cpk Basic Capability Concepts Process capability metrics relate process variation and centering to specification limits. They assume: - A stable process (in statistical control) - A known, meaningful specification: - Lower specification limit (LSL) - Upper specification limit (USL) - Approximately normal distribution for the characteristic measured Two main families exist: - Cp, Cpk (using short-term variation, σ) - Pp, Ppk (using overall, long-term variation, s) This section covers Cp and Cpk as core basic metrics. Process Capability Index Cp Cp measures potential capability assuming the process is perfectly centered between the specification limits. - Formula Cp = (USL − LSL) / (6σ) Where: - σ = short-term process standard deviation (often estimated from within-subgroup data) Interpretation: - Cp compares the specification width to the natural process spread (±3σ) - Cp > 1: process spread is narrower than the spec spread - Cp < 1: process spread is wider than the spec spread Limitations: - Cp ignores where the process mean is located within the specs - High Cp does not guarantee low defect rate if the process is off-center Process Capability Index Cpk Cpk accounts for both variation and centering. It measures how close the process is running to its nearest specification limit, in sigma units. - Formula Cpk = min[(USL − μ) / (3σ), (μ − LSL) / (3σ)] Where: - μ = process mean - σ = short-term standard deviation Interpretation: - Cpk ≤ Cp always - If the process is perfectly centered, Cp = Cpk - Lower Cpk indicates more risk of defects on the nearer side Cpk can be connected to sigma level by: - Sigma level (short-term) ≈ 3 × Cpk (for two-sided specs, approximate) - Or directly through the Z-value from the nearest tail probability implied by Cpk Cpk is one of the most widely used basic Six Sigma metrics for continuous data with specs. --- Performance Metrics: Pp and Ppk Process Performance Index Pp Pp measures overall process performance using long-term variation (overall standard deviation, s) instead of estimated short-term σ. - Formula Pp = (USL − LSL) / (6s) Where: - s = overall standard deviation of all observations Interpretation: - Pp reflects total variation including shifts, drifts, and special causes (if present) - Pp is often less than Cp when the process experiences drift over time Pp is informative when: - The process may not be fully stable - Long-term performance is more relevant than short-term capability Process Performance Index Ppk Ppk is the long-term counterpart to Cpk. - Formula Ppk = min[(USL − μ) / (3s), (μ − LSL) / (3s)] Interpretation: - Ppk shows how the process performs against specs over the entire time window, including sources of variation that Cp/Cpk may not capture - Ppk ≤ Pp, and usually Ppk < Cpk when the process is unstable or shifting Comparison of indices: - Cp, Cpk: focus on current inherent capability (short-term) - Pp, Ppk: describe actual performance over time (long-term) Understanding both sets is important for interpreting improvement results and long-term risk. --- Yield, Capability, and Sigma Level Connections From Capability to Defects Capability indices can be used to estimate defect rates when data are approximately normal. For a two-sided specification: - Let Zlower = (μ − LSL)/σ - Let Zupper = (USL − μ)/σ Then: - Probability of defect on lower side = Φ(−Zlower) - Probability of defect on upper side = Φ(−Zupper) - Total defect probability ≈ Φ(−Zlower) + Φ(−Zupper) Where Φ is the cumulative standard normal distribution. Cpk is effectively the smaller of: - Zlower / 3 - Zupper / 3 Thus: - sigma level (short-term) related to Cpk ≈ 3 × Cpk - corresponding DPMO can be estimated from the total tail probability From Defects to Capability Conversely, if you know the defect proportion (or DPMO) for a continuous characteristic: - Compute the Z-value for the non-defective proportion - Infer an approximate Cpk: - For symmetric two-sided specs, Cpk ≈ Z / 3 - For one-sided specs, use the Z corresponding to that single tail This allows: - Estimating Cpk from defect data when full specification-based analysis is unavailable - Consistency between defect counts and capability indices --- Basic Process Performance Metrics Cycle Time, Lead Time, and Throughput While these are not defect metrics, they frequently accompany Six Sigma metrics because they reflect process performance and are often improved together with quality. - Cycle time Time required to complete one unit or one process cycle from start to finish. - Lead time Time from customer request to customer delivery (often includes waiting and queue time). - Throughput Number of units processed per unit time. These metrics connect with defect-based metrics in several ways: - Reducing defects often reduces rework, which: - Lowers cycle time - Increases throughput - Excessive variation in cycle time is itself a process performance issue that can be analyzed with Six Sigma tools Although not specific to defects, understanding these time-based metrics is important for interpreting yield and capability outcomes in a broader performance context. --- Data Types and Metric Selection Attribute vs Variable Data Choosing the correct metric depends on the type of data. - Attribute data Count-based, such as: - Number of defects - Number of defective units - Pass/fail - Yes/no Typical metrics: - DPU, DPO, DPMO - Yield, proportion nonconforming - Sigma level derived from DPMO - Variable data Measured on a continuous scale, such as: - Time - Length - Weight - Temperature Typical metrics: - Cp, Cpk, Pp, Ppk - Z-values and sigma level based on mean and standard deviation - Defect proportion derived from specifications and distribution Consistent alignment between data type, metric, and analysis method is essential for valid conclusions. --- Practical Use of Basic Six Sigma Metrics Setting Up Metrics When establishing metrics for a process: - Define units, defects, and opportunities clearly - Identify specifications (LSL, USL) for critical continuous characteristics - Decide which metrics will be: - Primary (e.g., DPMO, Cpk) - Supporting (e.g., RTY, cycle time) - Establish consistent data collection methods: - Sampling plans - Time windows - Measurement system checks Interpreting Metric Changes When monitoring metrics over time: - Check if changes are statistically meaningful before acting - Relate metric changes to: - Process changes implemented - Environmental or operational shifts - Use multiple metrics together: - Improvement in DPMO but deterioration in cycle time may indicate more inspection or rework - Higher Cpk with no change in Ppk may indicate short-term gains without long-term stability - Ensure that improved metrics reflect better customer outcomes: - More good units on first pass - Fewer defects escaping to customers - More consistent and predictable performance --- Summary Basic Six Sigma metrics quantify process quality, capability, and performance using a consistent statistical language. Core elements include: - Defect-based metrics DPU, DPO, and DPMO describe how often defects occur, normalized by units and opportunities. - Yield metrics FPY and RTY quantify “right the first time” performance at steps and across entire processes. - Sigma level Expresses performance in standard deviation units and is closely tied to DPMO and process capability. - Capability and performance indices Cp and Cpk (short-term) and Pp and Ppk (long-term) relate process variation and centering to specification limits. - Supporting performance metrics Cycle time, lead time, and throughput complement quality metrics and link quality improvements to operational performance. Together, these metrics provide a coherent framework for measuring, comparing, and improving process performance in a data-driven way.