The operations best practices collectively termed Lean Six Sigma (LSS) fill several textbooks and form the basis for a professional certification.  They were developed by academics and industry practitioners across 80 years.  Looking back, the key results proceed logically from just two statistical insights!  First, repeated activities have variable results which can be described by a probability distribution.  The variability is inherent in the process and will continue until the process is fundamentally changed.  Second, the completion of a sequence of steps with variable time results can be described by a probability distribution derived from queuing theory.   Variability of the individual steps will accumulate, leading to bottlenecks and longer completion times than might be expected.

There are two core ideas.  Activities have variable results.  Sequences of activities have variable results with significant delays.  Both insights rely upon probability theory, which is based upon the repeated trials of random events.  Probability theory looks at the outcomes of the whole process.  It predicts average results and the dispersal of results.  As such, it describes the forest rather than the trees.  It predicts patterns of outcomes, not single events.  It treats variability and central tendency as equally important.

Six Sigma promotes the pursuit of near perfect quality.  The impact of variability on defects and quality is described in the work of Dr. Deming and Dr. Juran.  Lean is shorthand for Lean Manufacturing, which emphasizes the flow of product through a sequence of steps.  The impact of variability upon a sequence of dependent events is described in the works of Shigeo Shingo (Toyota Production System) and Dr. Goldratt.  In the last 30 years these breakthrough ideas have been applied, refined and extended by many others. 

The insights below are briefly summarized together with their major practical implications.  This approach allows the reader to quickly gain a sense of the range of topics addressed and their potential impact.  The power of these two statistical insights in the hands of an experienced operations team will hopefully become apparent. 

Quality, total quality management, zero defects, process re-engineering and ISO 9000 programs each played a role in developing the content of Lean Six Sigma.  These programs had mixed degrees of success due to their innovative nature, misunderstandings by businesses and practitioners, change management hurdles and conflict with the dominant short-term financial paradigm.  Lean Six Sigma has built upon their successes and modified the tools and applications to better fit within the real business environment.  Organizations which have reached a critical mass of Lean Six Sigma skills, processes and skilled staff have gained momentum, leveraging this effective approach to operations excellence.

1. Process steps can universally be modeled as inputs, processing and outputs.  Processing has inherent variability.  This variability applies equally to machine and human processing.  Variability is everywhere and can be described by simple statistics.  Defects are inherent in all variable processing steps.
2. Since variability is inherent in every process step, the responsibility for variable outputs lies primarily with the managers and engineers who designed the system.  Staff members are responsible for working within their capacity.  Supervisors are responsible for selecting, motivating and disciplining staff.  Process design is the largest driver of output and defects!
3. Process results can be measured in a simple fashion and graphed versus time.  Controlled processes can be determined statistically and the range of normal variability defined.  Outlier events can be distinguished from inherent process variability.  Outlier events should each be investigated.  Non-outliers should be ignored.  Reaction to random variability only increases variability and reduces average results!
4. Line employees can measure results, understand variability, determine and reduce sources of process variability.  Together with management they can analyze results and greatly improve processes.  Employee training and engagement is critical.
5. Since defects arise from variability, zero variability should be the goal of each process step.  Continuous improvement through many small steps is used to eliminate variability (close to zero).  Failsafe processes and radical simplification are highly valued.  Process step operators are best positioned to self-inspect the results of their work and prevent any defects from moving downstream.  Production lines should stop when a process step is broken.  Final inspection is the least cost-effective defect reduction program.
6. Supplier quality is critical.
7. Although Dr. Deming preached against the tolerance of any level of variability or defects, the controlled process measurement systems lead to self-correcting and self-improving systems.  A goal is set, like zero defects.  The result of the process is measured.  The set of defects is examined, grouped and rank ordered.  Solutions to reduce the leading cause are identified, implemented and confirmed to cause improvement.  The cycle is repeated.  This patient, incremental process can achieve heroic results with enough repetitions.  The conflict with the financial paradigm’s law of diminishing marginal returns is apparent.
8. Once released from the short-term assumptions of the financial paradigm, engineers found that progress towards perfection was possible with defect rates cut in half repeatedly.  One defect per hundred dropped to 1/200, 1/400, 1/800, 1/1,600, 1/3,200, etc.  As with the half-life of a radioactive material, there is no end.  The limit is zero, which is approached but never reached.
9. More importantly, practitioners found that all metrics can be defined in terms of defect rates and addressed the same way.  Zero variability, cycle time, capacity limits, set-up time, work-in-process inventory, response time, travel distance, non-value added steps, unique parts, unique events, etc. can all be pursued this way.  In practical terms this may be the single most important idea.  Cumulative progress through simple actions towards perfection is valuable.
10. This experience lead engineers to challenge and reject accounting orthodoxy.  Local optimums are not global optimums.  Fully allocated cost is a misleading fiction.  Efficiency and utilization ratios matter in the long-run but not in the short-run.  Eliminating waste is always a good step.  Staff and engineers can use this simple rule to guide improvement steps.  Budgets and financial measures provide disincentives for progress.  Financial measures are misunderstood by staff (and engineers).  In the long-run, enabling staff and engineers to make processes better may generate the greatest total value.  In practice, engineers and cost accountants have made peace, coordinating their standards and measurements and ensuring that budget variance explanations are made within the context of overall results.
11. The cost of quality is often underestimated by a factor of 2-5.  Accounting for all prevention, detection and mitigation costs confirms this range.  Initial defect prevention, identification and correction is highly valued.
12. The cost of customer dissatisfaction from defective products is “unknown and unknowable”.   The same value applies to out of stock products.  Businesses rarely apply infinite value to customer dissatisfaction, but the point is well taken and a non-zero estimate is used.
13. Freed from the mysteries of standard cost accounting, diminishing marginal returns, efficiencies and utilization measures, operations staff sought to make customer demands more certain.  They discovered that all customer demands fit within a framework of quality (product and process), speed, flexibility (capacity and exceptions), value (price, features and benefits), information (transaction cost, risk management) and personal relations.  Operations staff soon discovered that customer demands are unlimited.  Customers want zero defects, perfect tracking and assurance, immediate delivery as needed (not as ordered), infinite capacity, payment to use the product, total customization, zero transaction cost, fully buffered risk and family level intimacy!!  Operations staff members have used this insight to guide their long-term improvement efforts and to negotiate service level agreements with the sales staff to limit the demands for a given year and avoid the need for exception processing beyond this greatest common denominator!
14. The operations approach to measurements fits well within the balanced scorecard framework of complementary financial, customer, operations and asset measures.
15. ISO 9000 evolved to define a set of standards which ensure the effectiveness of a comprehensive quality management system.  ISO 9000 is built upon the simple self-improving system of “say what you do, do what you say and be able to tell the difference”.  As such, ISO 9000 leaders and auditors place heavy emphasis on the ability of the system to be audited for the existence of well-defined processes, the consistent measurement of results,  the review of variances and development of corrective actions and improvements.  This overlapping superstructure has sometimes been opposed by staff, operating managers, engineers and senior management who do not see the systematic value in being able to assess and prove compliance.  Wise internal ISO 9000 leaders have reduced the administrative burden and simplified processes to deliver “good enough” assurances of process compliance.

 

1. Shingo, Toyota and Goldratt looked deeply at the apparently simple idea that a process is composed of a sequence of dependent events.  One step cannot be completed until the prior step is complete.  The first result is that variation (delays) accumulates.  Randomly shorter and longer times do not naturally offset each other.  Later steps pay for the delays of earlier steps.  Waves of delayed product sweep through the system, interspersed with periods of no product.  More steps, options and variability combine to make average total process cycle time several multiples of the theoretical average time!
2. As with Dr. Deming’s focus on the “willing worker” who is unfairly expected to pay attention and work harder to avoid variability and defects, Shingo and Goldratt note that harder working and more responsive employees can play only a very small role in offsetting this accumulated variability.  These results are not based upon Theory X or Theory Y assumptions about the nature of employees.   They rely solely upon the science of variation.  The process is the 90% answer!
3. Both statistical insights indicate that management should focus on the process rather than individual events or behavior.  The whole is more than the sum of the parts.  The forest is more than the sum of the trees.  Improved efficiency in one step might not help total output at all.  Improved cycle time at one step might not lead to shorter overall completion time.  Labor efficiency or utilization may not help short-term results.  Short-term machine utilization is irrelevant.  Releasing work may delay the whole process.  A perfectly balanced production line is inherently inefficient.  No forecast is accurate enough to optimize production scheduling.
4. The scientific management, factory automation and IT approaches to optimizing a single step are called into question.   Single step improvements may not benefit end results.  If they require large processing batches, they probably hinder overall results!  The flow of product from step to step, across functional borders, may be the most important opportunity for improvement.  These improvements can be made locally with little or no capital investment!
5. Simple, visual measures and signals can best guide the flow of parts and product from station to station.  Complex planning and control systems to track product and forecast completion are costly and ineffective!
6. The probability of a final defect rises to certainty as error rates exceed 1% and the number of steps exceeds 50.  A two percent defect rate across 30 steps has a 46% total defect rate!  Individual process error rates must be reduced.  The number of steps must be reduced.  Processes must be combined into fewer steps.  Modular product design is indicated.
7. Work-in-process (WIP) inventory can be used to buffer variability between work steps.  It is less effective at buffering than is generally presumed.  In practical terms, WIP serves to hide the variability of individual process steps.  Hence there is no pressure for improvement.  The counterintuitive approach is to reduce inventory by 20% and monitor the process step variability to see if product can flow through.  If not, invest in steps to reduce the variability.  If so, reduce the inventory by 20% again.  Repeat!  Zero WIP is the goal.  In practical terms, one hour or one load or one container of WIP is common.
8. Large batches inherently increase the lumpiness of flow between workstations, increase total production time and result in inefficiencies due to delays.  As with WIP, the recommended approach is to reduce batch sizes until “unit of one” processing is achieved.  In practical terms this can be one container, one pallet, one shift or one unit.
9. Smaller batch sizes also play a role in reducing lead-times to recover from out of stock conditions.  They play a role in maximizing required production during a capacity constrained situation.  As such, effective organizations systematically invest in ways to reduce batch sizes for individual production steps and finished goods orders.
10. The fixed costs of an operation usually determine the minimum order quantity.  In manufacturing this is driven by the set-up times.  Manufacturers have found that set-up times can be reduced by 80-95% or more!  The historic role of large batches and volume discounts is greatly diminished.
11. The accumulation of process step variability indicates that downstream capacity must be significantly greater than beginning capacity.
12. Systems to control the release of product at the beginning of a production process or between any two adjacent steps can improve the overall capacity of the system by reducing the investment of time and materials in product that is not immediately required!
13. Due to set-up costs, unique parts and learning curve effects, high volume products are typically costed too high while low volume products are costed too low.  Efforts to reduce set-up, training and purchasing fixed costs reduce these differences.
14. Although world-class operators reduce lot sizes, WIP, lead-times, defects and step variability to surprising levels, few have been able to design a wide variety of products and process flows with true “unit of one” processing, zero WIP, 0.1% defects and one hour cycle times for random orders.  Hence, many operators develop focused factories to continuously produce small batches of related A volume products, classic assembly lines with cellular manufacturing designs, travel distances and shared work teams for B volume products and flexible low automation job shops for C volume products.  Customers cooperate to keep focused factories at 100% capacity.  They pre-notify parts requests for B volume products.  They accept longer lead-times and higher costs for C volume products that cannot flow within the focused factory or cellular manufacturing lines.
15. The production of finished goods for safety stock when there is demand for finished goods not in stock is inherently counterproductive.  Current demand product should be the first priority at all times.
16. Since finished goods demand cannot be adequately forecast, organizations should invest in reducing their total production time so that they can produce to order with a very short cycle time.  This approach is called a pull, just-in-time or build to order system in contrast with a push system that builds stock to meet forecast demand.  Production cycle time, even when multiple factories are involved, should be reduced from weeks to days to hours.
17. Since raw materials cannot be stocked for all possible finished goods demand patterns, suppliers should also reduce their cycle times to hours so that all components have an infinite effective supply!
18. In practice, manufacturers maintain finished goods buffers for immediate sale.  They reduce production cycle times to days.  They hold inventory buffers between plants even if they are able to minimize them within a factory.  Manufacturers hold a few days or weeks of component supplies.  They partner with suppliers and trade-off other factors to achieve historic usage quantity (2x) parts supply cycle times of less than one week.
19. Firms have also learned that improvement efforts must be differentiated among continuous process improvement steps done locally with minimum capital investment, Kaizen events that quickly change a contiguous set of steps with minimum capital investment, IT resources or adjacent impact and process re-engineering which requires time, capital, IT and other functional resources. 
20. Firms have developed a number of process evaluation methods to identify improvement opportunities.  Process results are benchmarked against world-class results in related and unrelated industries.  As noted in step 9, processes are compared against the ideal of zero cycle time, travel distance, non-value added activities, etc. 

 

 

The so-called quality revolution became a business operations revolution.  Like the probability based revolutions in quantum physics, chemistry, mathematics, ecology, evolution, meteorology, marketing research, target marketing, network communications, game theory, portfolio theory and options valuation, the discipline of operations management has been revolutionized as well.  Variability is inherent in process steps.  Variability is magnified across steps in a process.  Customers have infinite and infinitely variable demands.  The sales team lobbies for customers.  Product managers seek infinite product variety.  Finance requires predictable short-term financial results.  Suppliers want predictable demands.  Operations managers have embraced the Lean Six Sigma program and actively invested in training, staffing and projects to create the assets needed to anticipate, meet and manage these inherently competing demands. 

 

The net benefits of a comprehensive Lean Six Sigma program are substantial.  The staffing and project implementation costs are modest and well-defined.  The benefits, steps and risks are documented in this series of Lean Six Sigma articles. 

 

The received wisdom in strategic planning indicates that firms should pursue operations excellence, product innovation or customer intimacy.  The Lean Six Sigma approach directly delivers operations excellence, supports systematic product innovation and meets current levels of customer needs.  World-class firms in a wide variety of industries have proven the value of this business strategy.