Section 06. CapabilityAttribute

06. CapabilityAttribute

Overview

Early in any project (in Define for goal setting or more formally in Measure) it is crucial to understand the current level of performance of the process prior to making any changes. Many Champions and Belts mistakenly believe this is only to show how much is saved by the project or how big an improvement is made to justify continuation of the Lean Sigma program. These things are important but are only a small piece of the picture. The primary use of the measure is to ensure the gains are sustained after the improvements are in place. If the change is unmeasured it often is undone later (with all the best intentions) because it is not fully understood. However, if a measured and verified performance change is made, then there is less likelihood for future damaging "tweaks."

There are many performance metrics available (throughput, Overall Equipment Effectiveness, quality, and so on) of which Capability is just one. Capability comes in two forms depending on the type of data to which the tool is applied, but both look at performance versus specification(s). Here the focus is on the application to Attribute type data.

As always, whenever tools are applied to Attribute data, the tools are weaker, and here the same problem applies. Attribute Capability is expressed as Defects per Unit (DPU), calculated as:

It is necessary, therefore, to define the Unit and the Defect. A Unit is the physical output from the process (described as an entity in previous sections). It is something that is inspected, evaluated, or judged to determine "suitability for use." It is something delivered to Customers or users. Examples might include

• Invoice

• Shipment

• Customer Call

• Order

A Defect is anything that does not meet a critical Customer requirement or established standard. Examples might include

• Typographical error on an invoice

• Incomplete shipment

• Customer call dropped

• Missing order information

There can be multiple defect types for each entity and multiple defects of each type per entity.

DPU thus considers the total of all the individual defects (or errors) on each unit, or it can be applied to individual defect types, thus identifying which defect type is creating the largest loss. It is generally used to identify opportunities by prioritizing based on the commonest defect type (typically using a Pareto Chart).

DPU does not however take process complexity into account; to do this, the measure used is Defects per Million Opportunities (DPMO), defined as

The definition of Unit and Defect is as before. To understand Opportunities, it is useful to mention the difference between Defects and Defectives. A Defective Unit is any Unit containing a Defect. There can be multiple Defects in one Defective Unit and the Defectives Units are a result of Defects. It is impossible to reduce Defective Units without reducing the number of Defects. Opportunities therefore are the number of potential chances within a Unit to be Defective. It can often be difficult to identify all opportunities and it generally depends on the Customer. Examples might include

• Purchase Orders. Opportunities = Number of critical fields x 2 because the fields can either be empty or incorrect.

• Customer Call. Opportunities = Number of defect reason codes (for example, a missed call, a dropped call, an unresolved call, a call sent to a manager, a call that requires a call back, and a call that lasts longer than 15 minutes).

Opportunities are notoriously open to abuse. A process could be "artificially improved" if there are more Opportunities considered at the end of the project than at the beginning, so it is very important for the Team to be consistent in how they are defined.

Roadmap

The roadmap to calculating the Capability for Attribute type data is as follows:

Step 1.	For the metrics in question, define the goals and specifications. Ensure the validity of the metric (see "MSAValidity" in this chapter).
Step 2.	Define the Unit, Defect, and Opportunity as per the descriptions in "Overview" in this section.
Step 3.	Collect process data. At least 100 data points are required if the proportion of defects is greater than 5%. If the defect rate is less than 5%, then the sample size needs to be increased accordingly or it might be better to switch to a reliability-type metric, such as days between defects.
Step 4.	Calculate DPU and DPMO as per the equations in "Overview" in this section. From the DPMO calculate the sigma rating from a lookup table (see "Interpreting the Output" in this section).

Interpreting the Output

Capability is required to sign off on the Measure Phase of the project and is revisited in Control to validate performance.

DPMO is the capability measure primarily used to calculate process Sigma Rating. The infamous 3.4 Defects per Million representing "Six Sigma performance" is actually 3.4 DPMO. Some novice Belts worry about the required Sigma Rating and assume the project is only successful if the process has a Six Sigma defect rating when the project is complete. This is untrue; Six Sigma as an initiative is about breakthrough performance improvement. If a change is significant in terms of the savings it generates or the additional capacity or revenue it creates, then the Sigma Rating is really a secondary concern.