SUMMARY
The long-term development of a measurement “infrastructure” is a key building block for a full organization Six Sigma system. A huge benefit is an ability to monitor and respond to change.
Two activities toward measuring current performance.
There are basic concepts that you need to look into to fully understand the business process measurement.
Observe and then Measure. Get the “thing” being measured boiled down to an objectively observable event or behavior.
One example is the “10, 5, first and last” service rule in a hotel: Make eye contact with a guest at 10 feet; greet them at 5 feet; and be the first and last person to speak.
The Continuous versus Discrete Measures. Continuous measures are those factors that can be measured on an infinitely divisible scale or continuum such as weight, height, time, decibels, temperature, ohms, and money. Discrete items are characteristics or attributes (level of education; type); counts of individual items (numbers of orders processed); artificial scales (rating a record from 1 to 5). In other words, if you don’t see a number on some kind of measurement scale like temperature or time, you know you’re dealing with discrete measures.
Here are some measure examples of discrete, continuous and continuous converted to discrete.
| Discrete | Continuous | ---> Discrete |
|---|---|---|
| *number of typographical errors *rating of service *units delivered/day *number of claims in dispute |
*hold time per incoming call *minutes to board plane *quantity of gas in tank |
---> *number of calls on hold past 30 seconds ---> *delayed boarding incidents ---> *tank empty/full |
Establish a Process for Measurement. This is saying that measures should be continuously improved. Below is a five-step measurement implementation model.
Select what to measure
?
Develop operational definitions
?
Identify data source
?
Prepare a collection and sampling plan
?
Implement and refine measurement
The steps clearly show that selecting what to measure in your processes is very important.
Measurement selection criteria
| Value/Usefulness | Feasibility |
|---|---|
| link to high-priority customer requirements | availability of data |
| accuracy of data | lead time required |
| area of concern or potential opportunity | cost of getting data |
| can be benchmarked to other organizations | complexity |
| can be helpful on-going measure | likely resistance or “fear factor” |
As important as selecting what to measure is developing operational definitions. This means to establish a clear, understandable, and unambiguous description of what’s to be measured, or observed so that everyone can operate, consistently on the basis of definition.
In gathering your data, you need to prepare a collection and sampling plan. You may start with data collection forms in the form of well-designed spreadsheets and “checksheets.” You should keep your forms simple, labeled well, with space for date (and time), and collector’s name, consistently organized, and with stratified key data.
Stratification involves getting the baseline measure of performance against customer requirements and organizing data in layers or “strata.” This is also called the “slicing and dicing” of measures.
| Factors | Examples (Slice the data by …) |
|---|---|
| Who | *Department *Individual *Customer type |
| What | *Type of complaint *Defect category *Reason for incoming call |
| When | *Month, quarter *Day of the week *Time of day |
| Where | *Region *City *Specific location on product |
Sampling, on the other hand, means using some of the items in a group or process to represent them all. Keep in mind the rule of thumb that the larger your sample, the better your accuracy.
Having gathered your data, it is now time to establish baseline defect measures to determine internal performance.
Six Sigma allows you to adjust measures according to the number of “opportunities” for defects. The main steps in defining the number of opportunities are:
a. develop a preliminary list of defect types
b. determine which are the actual, customer-critical, specific defects
c. check the proposed number of opportunities against other standard
There are several ways to calculate and express measures based on defect opportunities.
1. Defect per Opportunity, or DPO – expresses the proportion of defects over the total number of opportunities in a group.
Formula:
Number of defects
# of Units x # of Opportunities
Service Example:
52 defects on applications = .052 DPO
250 apps. X 4 opportunity/app
• the service has a five-percent chance of having a defect in one categoryManufacturing Example:
52 defects on microchips = .052 DPO
750 chips x 150 opportunities/chip
2. Defects per Million Opportunities, or DPMO – indicates how many defects would arise if there were one million opportunities. In manufacturing, DPMO is often called PPM (parts per million).
Formula: DPO x 1,000,000 (10)
Service Example: Loan applications ? .052 x 106 = 52,000 DPMO
Manufacturing Example: Microchips ? .00046 x 106 = 460 DPMO
3. Sigma Measure – translate your defect measure—usually by DPMO—by using a conversion table (Chapter 2).
Sigma Measure Formula: Calculate DPMO, Consult Table
Service Example: Loan applications ? 52,000 DPMO = 3.1 Sigma
Manufacturing Example: Microchips ? 460 DPMO = 2.3 Sigma
COMMENTARY
The measurement techniques discussed in the chapter fell short of presenting the complete basic statistical quality control (SQC) tools for measuring performance, objectives, etc. Without the other techniques of SQC, Six Sigma measurement will not be complete and its implementation will be difficult. This will also leave the impression that Six Sigma is indeed highly technical and good only for technical people.
To illustrate, when it discussed sampling it should have likewise discussed measures of central tendencies and measures of dispersions. These are averages, mean values, standard deviation—figures that sampling represent and used in decision making. Second example is the discussion on defect opportunities. This is prioritization and in SQC a powerful tool to handle this is the Pareto Diagram. It also talks about stratification technique but it did not mention regression analysis and correlation coefficients, which will be used for a stratified data.
In sum, when you talk about measurement the best discussion should be the SQC tools. This is the basic tools used in TQM. This is one illustration that shows Six Sigma actually has its foundation laid down by past quality management systems, most especially TQM.
