NonConformances as a Tool to Measure Employee Performance

[Jim Wynne]

Good Morning!

I have enjoyed all these responses - I also enjoyed reading Don't Feed the Hog, and I learned quite a bit in this short exchange. To start with, I don't think I asked my question quite right, or that I provided sufficient background.

We are looking at continuing to develop an Employee Performance Management System (EPMS) - of which one part is Quality Performance.

Non-conformances are only one metric (of several) that we monitor. We do differentiate the type of non-conformace - Man, Material, Machine, or Method. Non-conformances do not always point at an operator. I believe we have done a pretty good job in advancing root cause, looking at practices and procedures - identifying improvements and implementing them. Often with the input and assistance of the shop employees. Much of what I read in Don't Feed the Hog is relevant and in practice. My interest here is in the Man related issues: I am less interested in the operator who set his machine up wrong (simple mistake), but caught the out-of-tolerance part at first article inspection, then I am in the next operator who set his machine up wrong (also simple mistake) and machined 100 pcs.

Perhaps a better question would be, "What makes a successfull program in developing or structuring the Quality component of an EPMS - which would be used in conjuction with several other facets, in order to evaluate employees? What other metrics could we consider?

I hope I have explained a little bit better about what I am after. I look forward to reading more..

I'm always a little wary when attempts are made to distill an exceedingly complex task down to an initialized (EPMS) neat little package. There is no way that I know of to do what you're trying to accomplish without missing something important, or without some form of subjective judgment gumming up the works.

The best you can do is look for patterns. If an individual continually has issues with producing acceptable output and other similarly-situated people don't, you need to deal with that individual. How that happens depends on the circumstances.

On the other hand, if there are recurring quality issues that are ascribable to operator error and the errors are more or less evenly distributed among operators, you most probably have some form of systemic issue that won't be remedied without management intervention in improving processes.

There cannot be a reasonable set of "performance" criteria that can be expected to be apropos to every individual in every job. Managers need to actually think about what they're doing and how each person's unique characteristics affect the work being done. This can't be done with a checklist and a set of one-size-fits-all criteria.

 

[Steve Prevette]

An interesting topic and conversation to date. Thanks, Wes, for pointing out some of my writings.

Understanding the differences between employees with respect to noncomformances can actually be useful. Be sure though, not to fall in the trap of ranking workers from top to bottom (as in the Red Beads) but do analyze for statistically significant differences.

Dr. Deming tells the story of a louper (sewing person) who had many more errors than their counterparts. Come to find out, the worker's eyeglass prescription was out of date, and new glasses solved the problem. Now, if we had marked them down in their performance evaluation, that would likely not have helped.

Let us say you have a worker who has actually figured out a good way to do something, that works, and prevents errors. They stand out ahead, but only due to they have failed to share the methodology with their coworkers and the company. Do you really want to reward someone for this behavior?

Sholtes and Scherkenbach have written a lot about performance appraisals. A thing they point out is that appraisals try to do too much in one bundle. They try to accomplish the following (all decisions that indeed must be made):
a. Feedback to the employee
b. Planning for the employee's career
c. Deciding who to hire / fire
d. Deciding how much to pay someone

Then there is Dr. Deming's equation of x + f(x). That is, the output of the employee is the sum of their contributions, plus their interaction with the system (other employees, equipment, managers). We may know from measurements that the output of the employee is 10. Now solve for x. Can't be done. What if their individual output is 30, but interaction with the system is -20?

Recommendation. Use your measurement data. Understand your measurement data. Understand your people. Understand the interaction of the data with the people. Analyze for causes. Perhaps there will be a rogue employee that you are better off getting rid of. Perhaps there are employees that just need a little encouragement. Perhaps there are some employee that honestly need some help. Perhaps there are employees that could (and should) be helping other employees. Use the data, but don't abuse the employees with it.