Principle 6: Variability (Item 7)

All systems and processes exhibit variability, which impacts on predictability and performance.

Why this is important

Is there any benefit in getting your processes to the point where they are capable of consistently producing products (or services) better than your customer specifies? Yes there is. And to see why, we must consider the `cost of quality' issue.

Cost of quality

In the old thinking, people thought that the only importance of variation was the need to force the product to comply with its specifications. The `cost of quality' was thought to be restricted to the aggregation of all costs required to meet the specification. It was assumed that there were no cost advantages, only additional costs, in having the product consistently better than the specification. The location of the midpoint of the range within the specification was incidental. "So long as you are in spec, that is OK".

Taguchi, a Japanese statistician developed an alternative approach to evaluating the `cost of quality'.

Taguchi proposed that all variation from the ideal result causes additional cost, at some point in time; to your company, to your direct customers, to your end-users, to the community. He proposed that all variation will impact your costs and performance.

He also found that costs increase in proportion to the square of the deviation from the ideal result. This finding indicates that you should aggressively work to minimize your variation, even when you are well within specification.

Taguchi's finding means that there are a competitive and cost advantages in reducing the spread (width) of variation to be well within your customer's specifications. When you do, you reduce your costs, your customers' costs or both. When you reduce you customer's costs, you have a more satisfied customer and a competitive advantage. Working aggressively to reduce variation in output gives a real win/win.

Reduction in variation was essentially the strategy that saw Japanese manufacturers gain such considerable market share during the 1980s. At that time, American manufacturers were still supplying equipment that had a much higher failure rate than their customers wanted. The Japanese manufacturers had recognized that customers really wanted a lower failure rate and aggressively worked to reduce their failure rate. The competitive use of tools such as Statistical Quality Control and Six Sigma grew out of that strategy. It was not until American manufacturers also began to work to make products more reliable, ie more within specifications, that the loss of market share to Japanese companies was slowed.

Measuring your Cost of Quality

Measuring your Cost of Quality is extremely useful. Most companies measure what they have put in place to catch poor quality – audit, QA, inspection, standardization. Important as this is, it is only part of the true cost. It is also responding to the old thinking and fails to recognize Taguchi's equation.

As well as those traditional measurements – which should be going down – you must add

• the cost of your rework (in most companies, this is huge)
• the cost of having your customers annoyed because what you delivered took them five telephone calls to get fixed
• the cost to your future income of being stuck fixing old problems
• the cost of lost production when your bottlenecks work on faulty material
• the cost of employing people just to fix things that should have worked in the first place (this can be a huge part of your support staff)
• the cost of having your `sales' staff work on solving customer problems (when they are fixing problems with old customers, they are not bringing new ones)
• the cost of warrantee payouts
• the cost of `money back' guarantees to customers
• the cost of returned goods
• the cost damages claims and liability insurance
• the cost of your complaints department and PR department
• the cost of lost sales due to perceived unreliability
• the cost of overtime due to production fluctuations
• the cost of scrapped parts and finished goods
• All these costs are incurred when you do not have consistent, reliable products and services. These should all be going down.

A simple example. An insurance claims department processing an average of 25 claims an hour, might have a distribution of processed claims ranging from 0 to 50. What is the cost of that variation? And the cost of its downstream effects? What are the costs of

• idle time (on the part of those waiting when no files were being handled)
• extra staff (or overtime) needed to handle the times when the log jam suddenly breaks and 50 files arrive
• the rework generated when 50 files are handled all at once.

We don't need to be consistent

Service companies often say that "This is manufacturing stuff. It does not apply to us. We are a service company". Wrong! Much of the work done each day by support staff in service companies is rework. Fixing problems for the second, third or tenth time that should have been done right to start with.

• Invoices without error - ever
• Trains running on time - always
• The doctor (dentist, lawyer, physio, accountant) seeing you at exactly the appointed time - always
• Enrollments at college or university completed without error. Results distributed without error. Consistency in marking.
• A library can always find the books on its shelves, inter-library loans work without error
• A bank never makes an error in a transaction or on your statement. There is no queue at the teller.
• At the fast-food counter, there is no queue, it takes 30 seconds to give your order and 2 minutes to get it. Your change is always correct.

Even if you think the direct service part of your business is already consistent, there will probably be problems in administration and support.

Non operational variation

We have all seen examples of companies who are very carefully controlling the variation in some parts of their company but have forgotten about it in other parts. For example, the high quality appliance producer that has worked consistently over many years to reduce variation in their products and who is regarded as an industry leader. But who is sent bankrupt by poor foreign exchange trading where the losses amounted to ten times the operating profit for the rest of the company.