Before we look at each type of checking, we should grasp one essential princi- ple of quality management. The sooner—the earlier in the process—we elimi- nate error, the better.
The best choice is that, with good requirements specification, architecture, and design, we prevent the error altogether. So error prevention is the least expensive option. In order of time and cost, here are six ways to deal with error:
•
Do it right the first time.Put top-notch effort into quality definition, plan- ning, and design, so that the product or service is as free of error as possible from the beginning.•
Catch the errors in early reviews of the plans.We should invest a lot of our quality management effort in review—ideally, close, structured review— of plans and design documents.•
Doing good production work.If we work as a team to make sure that inputs from vendors are checked, equipment is working well, and people are fol- lowing procedures, we minimize error and waste in production.•
Checking production work.This would include all forms of inspection and testing, with correction of the defect or scrapping of the component to pre- vent the defect from reaching the customer.•
Letting the customer receive the error, and then doing a good job of fixing it.Here, the customer has to deal with the frustration of the error, but we do a good job of helping with the cost through warranties, service plans, and affordable, high-quality customer service.•
Letting the customer receive the error, and then not providing good cus- tomer support.In this case, the customer pays the price, and we almost cer- tainly lose the customer. We also risk loss of reputation and legal action against us.One way to fully appreciate the above list is to realize that, for every error, one of these six things will happen. We will address this fully when we discuss theCost of Quality in Chapter 14.
Another way is to realize that these six errors fall into three sequential stages— planning, development of the product or service, and delivery to the customer. Many studies across all industries have demonstrated that there is a cost and time ratio for planning:development:delivery of 1:10:100. This is called the 1:10:100 rule, and it states that each error will cost ten times more to fix in development than it would to fix in planning, and 100 times more if the error actually reaches the customer. Some experts, most notably Dr. Harold Kerzner, have discovered much higher ratios. Dr. Kerzner cites a client who found that, in a five stage proj- ect life cycle, the ratio was 1:5:25:100:1000. The cost and time ratios and conse- quences of errors of the 1:10:100 rule are laid out in Table 5-1.
Let’s take a closer look at the three basic methods of checking:
•
Review is the process of comparing a document, such as a requirements specification or a design plan to standards or requirements that govern the process or results required of that document. Reviews can be highly formal and strict, or they can be loose and informal. Generally, the higher cost of the time effort spent in a formal review pays off, because it means we catch more errors earlier in the process.Chances to Prevent and Correct Error and Their Consequences
Do good work Check, catch, 1:10:100 Time Cost Customer Stage to prevent error and correct error ratio consequences consequences consequences
Do it right the 1 Best option Best option Least error;
Plan
first time most satisfaction
Catch errors in 1 May delay Relatively low cost, errors Least error; early reviews beginning of are easier to fix in plans most satisfaction
of the plans development than while working
Doing good 10 Keeps Ten times more expensive Customer will not see
production production per error, but continuous error, but increased
work and delivery improvement here saves a cost may be passed on
Develop on time lot of money to customer
Checking 10 May delay Expenses can balloon at Customer will proba-
production product the end of development, bly get product late,
work delivery cost and reducing return and cost increase will
on investment probably be passed on
Customer 100 Time to We have to maintain a Hassle for customer,
receives error, repair for larger customer service cost may go to
we fix well customer team and repair facility, company or customer—
loss of future business customer likely
Deliver to be dissatisfied
Customer 100 Time for Very large customer Hassle, cost, and
receives error, multiple support cost consequences, customer dissatis- we do not repair efforts loss of future business, faction with product
fix well loss of repair income, loss and with company
of reputation, legal costs
•
Inspection is the act of examining an attribute of a product, service, or component and comparing it to its specification. Some comparisons are discrete, such as “Did a red, blue, or black t-shirt go into the box for the customer?” while others involve measurement. Where measurement is involved, we determine if the attribute is within specified tolerances.Statistical quality controlis a special case of inspection where we test only a sample of the product and extrapolate to statements about the entire batch of the product using statistical methods.
•
Testingis the process of actually doing something with a product, service, or component and seeing what happens. Key issues in testing include the design of experiments, the cost of testing, and the type of test. Tests should be designed to check the maximum number of features at the lowest costs.Destruct tests are tests that check a feature, but destroy the product in doing so. Clearly, destruct tests can only be used on prototypes or samples of our final product, not on every item we were going to sell!