Near Perfect Quality

Six Steps To Near-Perfect Quality

This article discusses the barriers to near perfect quality and tasks required to achieve it.

Links to more best practice articles and training below.

Before discussing the six steps we need to discuss what "near-perfect" is and why perfect quality (zero defects) is not just difficult, it is nearly impossible. But you can make imperfect quality highly improbable! As Douglas Adams would say, we need to create "an infinite improbability drive".

All processes produce variable results. In statistical terms, in order to produce no faults the tolerance must be infinitely large in relation to the process capability. Therefore however good your process, statistically there will occasionally be errors. In the normal distribution, plus or minus 3 standard deviations represents approximately 99.7% of the population (3 errors per 1000). So if your process produces normally distributed products or services and the tolerance required is equal to process variation (six standard deviations around the mean) then on 0.3% of occasions errors will arise. A common definition of a "capable process" in Statistical Process Control (SPC) is that tolerance should be >1.3 x process variation, (from which you can expect approximately 15 faults per million), (commonly expressed as Parts Per Million and abbreviated to PPM).

Plus or minus 6 standard deviations (tolerance = 2 x process capability) represents less than 1 fault per million, (the ratio commonly used in "Six Sigma") (where the Greek letter Sigma (σ) is the mathematical shorthand for standard deviation) . Also we disagree with the concept of "cost of quality" as a means of providing motivation for quality improvement. Where we have seen attempts to cost quality they have in our view been gross underestimates. The biggest single cost is losing a valued customer and we have not found a satisfactory method to cost that, but the effect will certainly be felt in next year's profit and loss account.

Secondly it is that is commonly assumed that quality can be achieved by precision processes & equipment staffed by super-humans. It is not! Quality is influenced by a number of factors as we will show below.

Thirdly we will argue that whilst this topic is commonly described as "quality control", in fact quality control only forms one of the six topics (the last of the six steps) and therefore alone quality control cannot reach near-perfect quality.

We will now discuss how to get close to perfect quality by examining the six topics in detail:

The Six Steps:

Step 1: Determine Customer Value

One way to judge quality is, "does it satisfy the customer need?". This used to be the case, but is no longer so. There is a significant difference between customer "needs" and customer "wants". Delighting the customer" is a common current phrase, and customer satisfaction surveys are a common means of assessing our ability to determine what the customer values (and thereby what the customer does not value). We need reliable methods of determining what the customer values. We use a number of techniques involving customers such as "paired comparison", "FAB analysis", "SWOT analysis" & "Quality Function Deployment" (future article) in trying to understand what customers value.

Having established what customers value we are now in a position to write down a specification of product or service which can satisfy this, "the design brief". This design brief must be expressed in language that the customer would recognise first of all, before we can convert this to a product or service specification which we aim to provide. We use a special form of FAB analysis in reverse as follows:

What Benefits is the customer seeking as a result of owning or using (our) product or service?
Therefore what critical Advantages does the customer expect to see in (our) product or service?
What Features will deliver that?
What Processes can deliver those features?

Unfortunately many processes designed to deliver features for products have not been subjected to this rigour! This can result in both over-engineering, or under-engineering (future article).

Also required performance characteristics have to be evaluated not just in functional terms but also in terms of service criteria and therefore other criteria need to be considered such as risk or innovation (for a full list see Previous Malpractice of the week 003: Buying cheap). These requirements then have to be then matched to our ability to deliver that in terms of our process capability. Requirements which cannot be reliably delivered are bound to end in tears.

This then brings us to defining a tolerance for our specification which can be met reliably. Again the main criteria is "would the customer notice if it was outside this tolerance". Only now can we set tight pass / fail limits to this.

Step 2: Culture (Also see Previous Malpractice of the Week 004: Creating the wrong culture)

Mostly ignored completely, the topic of quality culture is perhaps the most vital ingredient in delivering a quality product or service. If no-one thinks quality is important, it will not be!

There are a minimum of six ingredients to delivering a quality culture as follows:

Management must behave as if quality is important (Slogans and posters exhorting quality values simply will not do the job and are often counterproductive.)
Ownership: Not just ensuring that authority / responsibility / accountability is clear and a that a senior person has responsibility (ISO 9000), but also that individuals feel responsible for delivering good products and services and behave like that.
A superior standard / expectation must be set, in terms of product / service delivery. I.e. not percentage faults, but Parts per million (PPM). It is sometimes assumed that this standard is only applicable to highly automated mass production of fast moving consumer goods, where output is measured in millions, but our experience would suggest otherwise.
The workforce must be both engaged in decision making, planning and problem solving & empowered to act on quality problems.
Competence must be measured and training / cross-training / practice made a frequent activity which is encouraged. There must also be time set aside for this activity. We call it "play time" but it has a very serious intent. However it is a common but counterproductive phenomenon to see training budgets slashed as the first soft target when the financial going gets tough! What message does that send to employees?
Act on problems / near misses / non-conformance rigorously and tenaciously, so that they cannot happen again (Mistake-proof or re-train routinely). Problems should be seen as an opportunity to improve, not as a stick to beat people. In this context customer complaints or customer returns are viewed with shock & abhorrence & dealt with by operations with the highest priority, not as an inevitable nuisance factor dealt with (often very badly) by the "customer relations department"! The customer relations department in particular we view as a "Malpractice". (Future article)

Step 3: Planning

Before you can have control you have to have a plan. However whilst you often hear the words "production planning" as a job title or as a function, how many times have you heard the words "quality planning". Quality planning is a vital integrating role which aims to:

Facilitate collaboration between experts
Reduce risk
Reduce variables
Increase capability
Manage change.

It is the planning envisaged by Deming in the "Plan Do Check Act" (PDCA), or "Plan Do Study Act" (PDSA) cycle as it is sometimes called. Planning includes the establishment of:

A new product introduction process
A Specification Management process (Managing Product or Process computer programs or documentation changes (change control / version control)
People capability & developing methods of judging levels of competence
Process capability & process improvement using Taguchi type methods for understanding the influence of variables
Setting control limits (and charts for Statistical Process Control (SPC) / Six Sigma)
Design for ease of Operations, Delivery, Installation, Commissioning, Maintenance & Disposal ("Design for Whole Life Cycle") (DFL) (Not just "design for manufacture" (DFM), or "design for assembly" ("DFA"), which are only parts of the whole product life cycle). In particular the WEEE regulations require focus on end-of-life considerations.
Mistake-proofing / Prevention using "Pokayoke" type methods (the quality equivalent of Total Productive Maintenance (TPM)
Right first time set up / make ready
Supplier Quality Assurance (SQA)
Traceability / shelf life / Product recall facilities

Step 4: Best Method

There is some confusion as to what determines the best method. Therefore it is often either not achieved, or not used by everyone. There are three separate issues here:

There are usually trade offs between speed / reliability / quality / equipment life, or simply personal preference, etc. Unless the “best method” is determined, then used routinely, there will be inconsistent or sub-optimal results.
How do we know the best method? This involves learning, which can be achieved in two ways (plus a good deal of discussion) as follows:
- Observation of the best internal or external performers ("Sixth Sigma" or Benchmarking)
- Experimentation (Taguchi style)
How do we ensure that everyone does it that way? This involves documenting the method and training everyone to do it that way ("Standard working" future article)

Step 5: Institutionalise / Corporate Learning

Once the best method has been determined the next step is to ensure that this method is institutionalised permanently into working practice. The 19th century Austrian mathematician and physicist Ludwig Boltzmann showed that "everything returns to nothing" with his equation for exponential decay as follows:

The best methods will "return to nothing" under this universal systems law as the business, people & environment changes, unless the method is institutionalised. One of the biggest criticisms of managers is that problems have to be solved many times because the lessons have not been comprehensively learned and the revised practice not corporately implemented (as opposed to, by individuals). This can be achieved in the following ways:

Make people remember it by recording / celebrating the improvement
For each process there needs to be:
- An owner (See Previous Best Practice 001: Ownership)
- A procedure ("Standard working") (See Previous Best Practice 003: Procedures & Documentation)
- A record of who can do it to what degree of proficiency (A "skill matrix") (See Previous Best Practice 006: Scarce Skills Management)
- A training programme to ensure that continuity is provided (See Previous Best Practice B002: Education and Training)
- An archive procedure (See Previous Best Practice B019: Archiving)
- A closed loop control system so that you know it is working well
- A periodic strategic review
Procedures / documentation such as the procedure and training manuals must be updated to reflect the new method
Education & Training programs must be established to not only train people in the new method but they must also understand how the method has been determined and why it is the best, if you want people to use this method.
Audits must be performed to ensure people are continuing to use the method
Kill the old method. If it is still possible to do it the old way someone will try! You can prevent this by a number of strategies, depending on the process in question.

Step 6: Control

Last but no means least (and the one you probably thought you were doing) is the establishment of a quality control system. However we think again that there is a common misunderstanding of this role. Firstly this is commonly interpreted to mean "do not let our poor quality performance be noticed by real customers". In far too many cases however you are undoubtedly aware that even this undemanding standard is not achieved! Secondly we do not believe in the philosophy of Total Quality Management or Total Quality Control (TQM) (or "total anything" as a matter of fact). The conventional wisdom of Total Quality Management (See "Lean Manufacturing") is "let's improve everything". Time and resources are too precious for this! We believe in a focused approach to improving the things which differentiate your product and processes, or which contribute to a quality culture (above). All this can be achieved as follows:

Closed loop control systems (future article)
Good measurement, metrology and recalibration control
Problem management which is focused on accident prevention, not just clearing up the wreckage
Root cause analysis showing possible causes of problems, faults, non-conformance, or near misses
Understanding variables & if they cannot be trivialised establishing testing stages to circumvent them
Control of work in process / stock / faulty work
Driving Improvement selectively using Focused Improvement Systems (Previous Best Practice of the Week 021: FRACAS). To avoid this confusion we tend to use the term "ongoing improvement" rather than "continuous improvement".

In conclusion therefore, we insist that near perfect quality is only possible by:

Competent people,
Using the best methods consistently,
Who want to do a good job,
Who are not prevented from doing that by competing managerial pressures (e.g. output / short term cost constraints) (See Previous Malpractice of the Week 006: Hitting the Numbers), or cultural barriers (See Previous Malpractice of the Week 004: Creating the wrong culture)