Book summary “Six sigma”: summary

Introduction

You face constant pressure to improve quality, cut costs, and satisfy customers while dealing with high process variation and hidden factory losses that drain profits. Traditional quality programs fail because they rely on intuition rather than data, leaving you with chronic defects and unpredictable performance.

Mikel J. Harry presents Six Sigma as a disciplined, data-driven management system that eliminates defects and variation to drive breakthrough financial profitability. The central idea unites business, operations, and process levels through rigorous measurement and statistical control.

The book guides you through principles, leadership structures, lean practices, statistics, and advanced design methods. You will discover answers to several core questions:

• How do you measure true process capability?
• What role do executive leaders and belts play in deployment?
• How do you eliminate non-value-added activities?
• When should you apply statistical tools and design experiments?

1. Principles of Six Sigma

Six Sigma replaces philosophical quality programs with a goal-driven management system focused on deterministic transfer functions and financial returns.

You can operate this system simultaneously at three levels. At the enterprise level, it drives business growth and customer satisfaction. At the operations level, it targets delivery time and cost of poor quality. At the process level, it minimizes variability, cycle times, and defects to generate verifiable savings like reduced headcount and lower overhead.

Total Quality Management relied on philosophy and local quality improvements rather than verifiable financial results. Historical studies showed that TQM failed to consistently improve quality or deliver broad financial impacts, with many programs stalling or failing to satisfy executive expectations.

A 1992 Rath & Strong survey found that only 20 percent of Fortune 500 companies were satisfied with their TQM results, and Florida Power & Light remains the only US company to win Japan's Deming Prize, later dismantling its strategy due to bureaucracy. Seventy-three percent of 300 electronics companies had quality programs, but 63 percent failed to improve quality by even 10 percent. Two-thirds of 30 quality programs studied by McKinsey & Co. stalled or fell short.

Six Sigma relies on deterministic reasoning where a dependent outcome variable is a function of underlying causal variables, expressed as Y equals f of X. In the real world, outcomes depend on a set of determinants represented as x1 through xn. Recognizing that variables do not exert equal influence allows you to identify the vital few factors that drive major changes in the output.

Six Sigma evolved across three distinct generations from Motorola's initial defect reduction to Asea Brown Boveri's cost reduction, and finally to comprehensive value creation. The first generation from 1984 to 1994 focused on defect reduction at Motorola. The second generation from 1994 to 2001 targeted verifiable cost reductions at Asea Brown Boveri. The third generation from 2001 onward focuses on the creation of value through utility, availability, and worth.

Motorola initiated the first generation of Six Sigma, while Asea Brown Boveri applied it within their large-scale transformer business during the second generation.

Sigma benchmarking reveals that the average global corporation operates at a four sigma level with massive hidden cost-of-poor-quality losses, whereas Six Sigma businesses achieve vastly superior performance. While corporate accounting systems often report cost of poor quality as a tiny fraction of sales, true benchmarking shows that average companies bleed 25 to 30 percent of sales in hidden losses. Operating at Six Sigma delivers an 1800-fold improvement over the global four sigma average.

Fifteen to 25 percent of every sales dollar is left on the table by typical corporations. Five to 6 percent of sales can be recovered in hard savings through Six Sigma. Four sigma average global performance translates to about 6810 defects per million opportunities. Twenty-five to 30 percent of sales represents the cost of poor quality for average companies. Eighteen hundred times better performance is achieved by a Six Sigma company compared to its four sigma counterpart. One percent of sales is the cost of poor quality experienced by a Six Sigma business. Three point four defects per million opportunities represents the elemental Six Sigma performance level for a single critical-to-quality characteristic.

2. Leadership and Deployment Planning

Successful Six Sigma deployments require executive leadership, a structured hierarchy of X-Belts, and metrics tied to valid performance criteria.

Six Sigma is fundamentally driven by leadership rather than mere tools or structures. You need leaders who act as change agents and create hope. Leadership provides the energy and impetus required to set a corporation on a new performance trajectory. A company with strong leadership and weak Six Sigma knowledge is better off than a company with massive Six Sigma knowledge and weak leaders.

Bob Galvin of Motorola described the finite nature of human leadership contrasted with its almost limitless influence to inspire, set direction, and call for action.

Six Sigma deployment relies on a structured hierarchy of specialized roles known as X-Belts, each carrying distinct responsibilities for project execution and training. Executive management owns the initiative, Champions direct and align projects, Master Black Belts train and coach, Black Belts lead high-impact projects full-time, Green Belts assist part-time, and White Belts provide process-specific knowledge.

You must implement these deployment roles according to clear operational practices:

• Executive management establishes corporate-level goals and sets performance expectations.
• Champions plan, orchestrate, and direct X-Belts to ensure consistency with strategic goals.
• Master Black Belts manage project selection, mentor Black Belts, and mediate team conflicts.
• Black Belts lead data-centric process improvement projects yielding hard cost savings.
• Green Belts apply Six Sigma methodology part-time on one to two projects.
• White Belts provide cross-functional process knowledge and maintain primary and secondary metrics.

White Belts receive 2 to 4 hours of awareness training to prepare for their roles.

To initiate and sustain organizational change, performance metrics must satisfy six rigorous criteria rather than merely serving as measurement standards. Metrics must have a valid scale, a clear standard or goal, tied compensation and recognition, regular review throughout the enterprise, capability for horizontal and vertical pooling, and high correlation with business criteria.

You apply twelve criteria for performance metrics at business and process levels by following these practices:

• Ensure the metric has a meaningful, valid, and reliable scale such as frequency or defects.
• Establish a precise standard or numerical goal for the metric.
• Relate compensation and recognition directly to achieving the performance goal.
• Review the metric on a regular basis and distribute data to all impacted personnel.
• Enable the metric to be pooled horizontally and vertically across functions and levels.
• Verify that the metric is highly correlated with enterprise performance criteria.

One such target is a 0.05% monthly injury rate goal or 99.95% uninjured workforce rate.

The DMAIC methodology provides a structured, data-centric problem-solving roadmap where the measurement phase is critical and time-consuming. Unlike older quality programs that neglected measurement, Six Sigma mandates a dedicated measurement phase because unmeasurable processes cannot be improved. The framework relies on a five-phase iterative approach.

When Mikel Harry benchmarked problem-solving methodologies at Motorola, he discovered none of them had a separate phase devoted solely to measurement. Lord Kelvin made his famous statement on measurement in 1891.

Six Sigma is structured around three foundational pillars—DFSS, PFSS, and MFSS—which target design risks, ongoing operational processes, and overall management systems. Design for Six Sigma reduces risks in product and service design. Processing for Six Sigma minimizes operational defect risks. Managing for Six Sigma unifies them to achieve best-in-class performance and value entitlement.

DFSS and PFSS enforce a limit of 3.4 risk exposures per million value opportunities. MFSS targets a 78% annualized improvement rate for its learning curve.

Valid Six Sigma application projects must target chronic, data-centric problems and be appropriately sized to deliver hard financial benefits within a strict timeframe. Projects must avoid cost avoidance or common-sense fixes, focusing instead on accounting-quantifiable problems requiring a science-based approach.

You must select and size these projects carefully:

• Select persistent or chronic problems that require data-centric, science-based approaches.
• Size Black Belt projects to be completed within 2 to 3 months.
• Size Green Belt projects to be completed within 4 to 8 weeks.
• Validate project benefits through the sponsoring organization before closing.

Avoid jumbo-sized projects that lack clear boundaries or realistic completion schedules.

3. Lean Practices and Value Stream Mapping

Lean Sigma combines Lean methods for waste reduction with Six Sigma methods for variation reduction to deliver a comprehensive business management system.

You target process efficiency through waste reduction while applying Six Sigma to target process effectiveness through variation reduction. Taken together, they provide a systematic approach to reforming critical business processes and achieving quantum improvements.

Five fundamental laws guide Lean Sigma improvement efforts, addressing customer focus, process flexibility, focus on delay causes, work-in-process limits, and product complexity costs. These laws establish a hierarchy and framework for reducing human labor, inventory, and product development time by targeting the most critical performance variables in any operational or transactional system.

• Law 0 (Law of the market): Customer critical to quality defines quality and is the highest priority, followed by return on invested capital and net present value.
• Law 1 (Law of flexibility): The velocity of any process is proportional to the flexibility of the process.
• Law 2 (Law of focus): 20 percent of the activities in a process make up 80 percent of the delay.
• Law 3 (Law of velocity): The velocity of any process is inversely proportional to the amount of work-in-process, also known as Little's law.
• Law 4 (Fourth law): The complexity of the service or product adds more non-value-added costs and work-in-process than either poor quality or slow speed.

Taiichi Ohno observed a Detroit supermarket where workers immediately restocked merchandise sold to customers, inspiring the manufacturing pull system.

You must identify and eliminate eight categories of deadly waste, known as TIM WOODS, which drain resources and cripple business processes. Overproduction is considered the most serious waste because it forces excess inventory, motion, and transport.

• Identify and eliminate Transport waste (moving people, products, and information).
• Eliminate Inventory waste (storing parts and documentation ahead of requirements).
• Eliminate Motion waste (bending, turning, reaching, lifting, or ergonomic strain).
• Eliminate Waiting waste (waiting for parts, information, instructions, or equipment).
• Eliminate Overproduction waste (making more than is immediately required by customers).
• Eliminate Overprocessing waste (tighter tolerances or extra unrequested features).
• Eliminate Defects (rework, scrap, and incorrect documentation).
• Eliminate Skills waste (underutilizing employee capabilities and inadequate training).

Value stream mapping requires a hands-on, end-to-end evaluation of material and information flows starting from the customer and working upstream.

You must personally walk the entire door-to-door value stream carrying a stopwatch, avoiding reliance on standard times or delegated reports, and draft the map by hand in pencil before transferring it to software.

• Select a product family.
• Have one person personally lead the mapping activities.
• Start the door-to-door level by taking a quick walk along the entire value stream.
• Include both material and information flows starting with the customer.
• Bring a stopwatch and do not rely on standard times or unverified information.
• Map the whole value stream yourself by hand in pencil before using software.

Mr. Hank, a corporate communications senior executive, personally walked through the actual pathways and information flows with a stopwatch, starting at the shipping end and walking upstream to map the release and shipping of the Corporate Communications Newspaper.

Takt time measures the pace of customer demand by dividing available production time by customer demand. You determine effective available working time per period by taking a workday of 8 hours or 480 minutes, subtracting a 30-minute lunch and 45 minutes of meetings, leaving an effective working time of 405 minutes.

• Determine the effective available working time per period.
• Determine total customer demand for the same period.
• Divide available working time by customer demand to establish Takt time.

4. Statistics and Process Capability

Statistical tools, data collection, and probability distributions establish baseline performance and measure true process capability.

Data must be measured using one of four scales—nominal, ordinal, interval, or ratio—which determine the mathematical operations and statistical tools that can be validly applied. Company market values in billions of dollars such as company A at 239 and company C at 384 illustrate ratio-scale data. You can evaluate variability using the range, variance, standard deviation, coefficient of variation, and interquartile range. Six long-distance calls selected from a month of calls had lengths ranging from 9 to 53 minutes, yielding a range of 44 minutes.

• Identify the data to be analyzed, such as errors, defects, or consumer complaints.
• Select the categories to be used such as products and sort data into these categories.
• Draw two vertical axes and a horizontal axis to make the graph, arranging bars in decreasing or increasing order.

You use capability indices like $C_p$ and $C_{pk}$ to measure process potential and centering relative to customer specification limits. A process has a mean of 0.738, standard deviation of 0.725, LSL of 0.500, and USL of 0.900, resulting in a potential process capability index $C_p$ of 0.92. You apply a 1.5 sigma mean offset to short-term capability models to account for transient and temporal errors over the long term.

• Construct a high-level flowchart of the process.
• Collect data on the number of defects in units for each process step.
• Calculate DPU and in-process yield for each step.
• Multiply together all in-process yields to find rolled throughput yield.

You can analyze discrete data by calculating the probability distribution of a discrete random variable, which must satisfy two specific rules. In a machine breakdown analysis where $x$ is the number of weekly breakdowns, the probabilities for 0, 1, 2, and 3 breakdowns are .15, .20, .35, and .30 respectively, summing to 1.0. You find the expected value by multiplying each value by its probability and summing the results.

• Assign a probability $P(x)$ to each value of $x$ such that $0 \le P(x) \le 1$.
• Ensure the sum of all probabilities equals 1.

Probability distributions model different operational scenarios. The binomial probability distribution calculates the probability of obtaining exactly $x$ successes in $n$ independent trials. An electronics firm finds that 5 percent of manufactured digital video recorders are defective in a random sample of 10 units. The Poisson distribution models random occurrences based on a known average rate. An ATM outside a bank is used on average by 5 customers per hour, and you can calculate the probability of 8 arrivals during a one-hour closure.

• Identify the average rate of occurrences lambda for the specified interval.
• Ensure lambda and the actual occurrences $x$ are defined for the same interval, adjusting lambda proportionally if necessary.
• Substitute lambda, $x$, and constant $e$ into the Poisson formula to calculate the probability.

5. Advanced Analysis and Design for Six Sigma

Hypothesis testing, experimental design, and Design for Six Sigma ensure robust processes and prevent defects before launch.

The method of least squares determines the best-fitting regression line by minimizing the error sum of squares.

You can compute the least squares regression line by following a direct analytical procedure:

• Compute means of explanatory and response variables
• Compute SS_xy and SS_xx sums of squares
• Compute slope b and intercept a using sample formulas

Seven households sampled with incomes ranging from 27 to 88 hundreds of dollars and food expenditures yielded a computed intercept a = 3.6461 and slope b = .2296.

Design for Six Sigma is a disciplined approach to designing or redesigning products, services, and processes from the ground up to ensure planned, measurable quality and meet customer requirements. The expected sigma level for a DFSS product or service is at least 4.5 sigma, translating to roughly 3.4 nonconformities per million opportunities over the long term.

You can perform stochastic optimization by combining Monte Carlo simulation with metaheuristic optimization algorithms to automatically search for controllable input settings that yield the best performance while respecting system constraints.

A Six Sigma team used simulation and optimization with OptQuest for plastic injection-molded parts, reducing per-part manufacturing costs from $2.03 to $1.16 while improving process capability to 4 sigma across 800 optimization runs.

10 Key Ideas

1. Mikel J. Harry presents Six Sigma as a disciplined, data-driven management system that eliminates defects and variation to drive breakthrough financial profitability.
2. Six Sigma relies on deterministic reasoning where a dependent outcome variable is a function of underlying causal variables, expressed as Y equals f of X.
3. While corporate accounting systems often report cost of poor quality as a tiny fraction of sales, true benchmarking shows that average companies bleed 25 to 30 percent of sales in hidden losses.
4. Six Sigma deployment relies on a structured hierarchy of specialized roles known as X-Belts, each carrying distinct responsibilities for project execution and training.
5. Five fundamental laws guide Lean Sigma improvement efforts, addressing customer focus, process flexibility, focus on delay causes, work-in-process limits, and product complexity costs.
6. Value stream mapping requires a hands-on, end-to-end evaluation of material and information flows starting from the customer and working upstream.
7. Data must be measured using one of four scales—nominal, ordinal, interval, or ratio—which determine the mathematical operations and statistical tools that can be validly applied.
8. You use capability indices like $C_p$ and $C_{pk}$ to measure process potential and centering relative to customer specification limits.
9. You apply a 1.5 sigma mean offset to short-term capability models to account for transient and temporal errors over the long term.
10. Design for Six Sigma is a disciplined approach to designing or redesigning products, services, and processes from the ground up to ensure planned, measurable quality and meet customer requirements.