Accountiability

[James Robert Crow <jr1crow@mindspring.com>]

Hello Group:

How do we know that a problem exists?  Do we have data that shows that the
quality in in decline?  Is this data over an extended period of time or is
in for only a couple of months?  A lot of time is spend in organizations
trying to fix problems that don't exist.  Unless you have a definate trend
that shows that your warrenty costs are up, your inprocess failures are up,
your customer complaints are up then a problem may not exist in the true
sense of the word.  There is no deviation from the norm or special cause.

This does not mean that your quality is as good as it needs to be, it
simply means that there is no assignable cause.  In other words if we do a
root cause analysis we will be unable to find a root cause.  Nothing has
changed.  The problem is in the process, and that is where you must start.
Put together the data concerning the output of the process to see what is
happening at this time.  If you already have control charts in place then
you have this done already.

You can now begin defining you key cross organizational processes, flow
charting them and measuring the output at each stage of the process.
Remember one of Deming's things regarding systems is that the output of any
stage must be equal to the input requirements of the next stage. Where is
the work backing up?  Where are the constraints, or when you look at the
type of quality problems you are having at what stage of the process are
these problems occuring.  As you remove these constraints you will get a
smoother flow and better productivity.

If you do determine a trend then you must do a root cause analysis to
determine what has happened that might have caused this.  What has changed
in the system that could affect quality?  When and where did this change
occure?  Is there a correlation between this change and the downturn in
quality?

Finally let's address the people issue.  Perhaps the easiests way to fix
this is to simply get everyone together and ream them a new asshole.  Let
them know that the next person that screws up is history then be prepared
to back that up.  We will probably learn pretty quick that this is not very
effective and when our turnover goes up we will have even more quality
problems.

There will be variation in the workers just as there is variation in any
system.  If you can track errors by workers then you should be able to put
together a control chart to determine if any of the workforce is outside
the control limits, and if so why.  I would be willing to bet money that
your workers will fall within the limits which means that the errors are
being driven by the system.

Even so your error rate may still be too high.  Are you people being
properly trained, or do you have a follow-Jack method which is to say that
you have no method at all.  Follow-Jack training is a source of
suboptimization, it adds variation to the system.  People that have
received formal training are 30% more productive after one year.  Move from
control to development.
======================================================================