posted 28 February 1999 10:52 PM               

In a message dated 2/18/99 12:45:04 AM, [email protected] writes:

>I am working with a group whose notion is we can correlation analysis to
>determine relationship of failures of our service to major indicators/metrics
>in our processes. Doing this we believe could refocus our process management
>efforts in areas showing high correlation and the regression will provide
>us
>some form of predictability of future service based on which variables
>we
>control.
>
>I know I'm somewhat vague but I promised the group I would not share more
>info
>until we are further in our analysis.
>
>Can anyone provide any references or insight on this approach?
>
>Thanks,
>
>Thom Handa ([email protected])
>Dallas, TX
>

This is a tough subject and has been studied mightily over the years though maybe not directly. Service is a vague topic. Is it a touchy feely service like consulting, or uptime service on a machine. Regardless, the critical issues is what is the customer concerned with and can it be measured. I suggest it can, but the accuracy of these kinds of measurements can and do vary widely. And not all customers see things in the same way. So putting the above mentioned vagueness aside, I get the feeling that QFD maybe a help to you.

In essence, QFD (Quality Function Deployment) is a tool to map out product characteristics with customer wants and needs, working to determine the value the customer puts on the characteristics. With this knowledge, the goal is to develop a technical metric that can be controlled and manipulated to improve the characteristic the customer is interested in. QFD is not a small undertaking, and can be ripe for abuse (political manipulation, part of it's design was to thwart this). QFD id a kind of pseudo scientific method, attempting to add precision to a notoriously imprecise process.

I hope this helps. P.S. - My expertise in this is the result of some reading for application in my organization and a 3 day class.