Six Sigma is a structured approach to quality and process improvement, aiming to increase process capability, reduce variation, and thereby sustainably reduce errors, rework, and costs of poor quality. At its core, Six Sigma provides a disciplined set of methods and tools to make performance deviations measurable, identify causes cleanly, and secure improvements so that they work stably in everyday life.
The name "Sigma" comes from statistics: it describes the standard deviation as a measure of variation. Often, Six Sigma is associated with a very high level of performance - classically with 3.4 errors per million opportunities (DPMO) as a reference value. What's important here: in practice, it's not about "perfection at all costs," but about targeted, economically sensible stabilization of critical processes.
Six Sigma combines customer orientation with facts, data, and statistical logic. Typical fields of application are:
- Voice of the Customer (VoC) & CTQ (Critical to Quality): what is really important to the customer - and how can it be measured?- Measurement systems & data quality: reliable data as a basis (e.g., measurement system analysis, clear operational definitions).- Process performance & process capability: how well does the process meet specifications - today and sustainably? (e.g., Cp/Cpk as key figures)- Cause analysis instead of symptom fighting: testing hypotheses, proving relationships (instead of "gut feeling").- Risk orientation: error and risk prioritization (e.g., FMEA) and robust countermeasures.- Sustainable safeguarding: standards, control plans, SPC/control charts, reaction plans - so that improvements remain.
DMAIC - improving existing processes (stabilizing, reducing variation, increasing performance)
DMAIC is the standard approach when a process fundamentally exists but produces too many errors, variation, rework, costs, or delivery fluctuations. The goal is to measurably improve performance and subsequently secure it sustainably.
D - Define (Define)- Clarify target image & benefits: what is the problem (from the customer's/business perspective), what is the goal (SMART), where does the process begin/end (scope)?- CTQ & requirements: which quality characteristics are critical (CTQs), which specifications/expectations apply?- Project setup: roles (champion/owner/belts), milestones, risks, communication and escalation channels.- Typical results: project charter, SIPOC/high-level process image, stakeholder analysis, initial hypotheses.
M - Measure (Measure)- Determine current performance: baseline (e.g., DPMO, yield, throughput time, scrap, complaints).- Secure measurement system: data is only as good as the measurement system (clarity of definitions, measurement capability, if necessary MSA).- Make process more transparent: data collection plan, process steps, handovers, influencing factors (X) vs. result factors (Y).- Typical results: data collection plan, baseline key figures, reliable data basis, process map/VSM excerpt.
A - Analyze (Analyze)- Prove causes instead of assuming: which X drive the Y? Where do variation/errors really occur?- Structure cause logic: cause trees, 5-why, Ishikawa, Pareto; subsequent validation (data-based).- Identify bottlenecks/key points: process steps or influencing factors that have the greatest effect.- Typical results: verified root causes, quantified levers, prioritized list of causes including evidence.
I - Improve (Improve)- Develop and test solutions: countermeasures, process changes, parameter optimization, Poka Yoke, standardization.- Demonstrate effectiveness: pilot/trial, before-and-after comparison, stability proof under real conditions.- Increase robustness: design solutions so that they also work with fluctuations (not just in the "best case").- Typical results: implemented improvements, proven effect, updated standards/work instructions.
C - Control (Control/Secure)- Secure in the line: monitoring, control charts/SPC (where sensible), reaction plans, audit/review routines.- Standard & ownership: process responsibility, training, visual management, KPI logic, and escalation fixed.- Sustainability: prevent the process from falling back (drift) - through clear triggers and measures.- Typical results: control plan, KPI/visual boards, reaction matrix, stable process performance over time.
Motto: DMAIC is suitable whenever a process "is there" but does not run stably or capably enough.
DMADV / DFSS - development or fundamental redesign (building quality "in")
DMADV (often in the context of Design for Six Sigma - DFSS) is used when existing processes/products structurally do not meet the requirements (e.g., technological limits, too high complexity, wrong process design) or when something new is being developed. The goal is to integrate quality and performance capability into design and process layout from the outset.
D - Define (Define)- Goal & business case: why new design/redesign? What goals (cost, quality, delivery capability, safety, scalability)?- Customer/user requirements: VoC -> CTQ structure including priorities and measurement definitions.- Boundary conditions: standards/regulatory, capacity, automation, interfaces, site/layout limits.- Typical results: design charter, CTQ tree, rough concept scope, and stakeholder alignment.
M - Measure (Measure/Translate)- Make requirements measurable: CTQs into measurable specifications, tolerances, target values, acceptance criteria translate.- Define capability targets: what process capability is required to achieve CTQs stably?- Benchmark/references: internal/external comparison, lessons learned, known failure patterns/failure modes.- Typical results: measurable specifications, target capability, measurement and acceptance plan.
A - Analyze (Analyze/Evaluate Options)- Develop design alternatives: multiple concepts (not just "one idea").- Evaluate risk & feasibility: FMEA/design risks, complexity, costs, manufacturability, operability, maintainability.- Make trade-offs transparent: e.g., quality vs. cost vs. flexibility - with clear decision criteria.- Typical results: evaluated concepts, decision basis (conscious selection of the best design).
D - Design (Design/Detailed Design)- Work out detailed design: process steps, parameter windows, test concept, layout/material flow, standard work.- Robust design: error prevention (Poka Yoke), tolerance design, suitable process windows, interface clarity.- Prepare industrialization: work schedules, qualifications, maintenance, spare parts, HSE aspects.- Typical results: detailed process/product design, SOPs/standards, test & control concept, ramp-up plan.
V - Verify (Verify/Validate)- Proof under real conditions: prototype/pilot, pre-series, process acceptance, performance proof against CTQs.- Stability & handover: acceptance criteria met? Control plan, training, start-up/run-at-rate, releases.- Series readiness: clear criteria, from when "series-capable" or "rollout-capable".- Typical results: validated design, fulfilled CTQs, release/handover to line, stable series start.
Motto: DMADV/DFSS is sensible when you cannot "repair" quality but have to construct it cleanly - because the existing approach cannot stably meet the requirements in principle.
Six Sigma works best when it's not just "project work" but is clearly embedded in leadership, prioritization, and governance. Typical roles are:
- Sponsor/Champion: ensures priority, resources, barrier removal- (Master) Black Belt: methodological leadership, coaching, statistical depth- Green Belt/Yellow Belt: implementation close to the process, quick improvements- Process owners: secure handover, standards, and key figures in everyday life
Strengths of Six Sigma- More stable processes, fewer errors/rework, higher delivery reliability, and quality- Decisions based on data instead of assumptions- Improvements are sustainably anchored through control mechanisms
Frequent stumbling blocks- Too little data quality or unclear measurement definitions ("numbers without trust")- "Statistics as an end in itself" instead of focus on impact and business benefit- Lack of leadership involvement (project runs, line remains unchanged)- Projects that are too large without a clean scope and without quick benefits (quick wins)
Lean and Six Sigma complement each other very well:
- Lean reduces waste, shortens lead times, creates flow and standards.- Six Sigma reduces variation, increases process capability, and lowers error costs.
Operational Excellence is the overarching framework: value creation, stability, leadership system, and culture interact - Six Sigma provides the data-based in-depth logic where "stability and quality" are the bottleneck.
In the sense of the 5M Lean House, Six Sigma can be very well classified:
- Lean motivation: improvements are justified through customer requirements (CTQ) and business benefits.- Lean mindset: fact-based, hypothesis testing, understanding variation - instead of opinions.- Lean management: clear roles (champion/belts), prioritization, review routines, KPI logic.- Lean migration: structured project path (DMAIC/DMADV) and change accompaniment into the line.- Lean manifestation: standards, control plans, SPC/monitoring - visible, stable implementation in everyday life.
I use Six Sigma not as a "statistics program" but as a logic of effectiveness for critical quality and stability problems:
1. Cut the problem correctly: CTQ clear, scope small enough, benefit visible.2. Only deepen data where necessary: as much statistics as required - as little as possible.3. Combine Lean + Six Sigma: first create stability and standards, then reduce variation targeted (or vice versa - depending on the bottleneck).4. Securing in routines: control mechanisms are directly anchored in leadership, standards, and key figures - so that it doesn't remain with "project slides".