RE: Measurement Problem

["Alan Meekings" <alan@landmarkconsulting.co.uk>]

Ket,

You ask:

"The client wishes to eliminate completed works orders that are handed in
more than 2 weeks late after completion (the work is carried out remotely).
They are approximately 80-100 works orders issued per week.

2 weeks is critical to the client, as it affects their forwards planning if
they have any more lateness then 2 weeks.

What should we initially start to measure?

a) No of works order per week more then 2 weeks late
b) Average lateness per works order per week
c) Max and minimum late work order per week."

My suggestion is that you measure the time to hand-in works orders after
completion, not the "lateness" of their hand-in.

Essentially, you've defined 'failure' as "more than two weeks late".  This
is classic Six Sigma thinking - namely look at everything in terms of
'failure', and then trying to get 'failures' down below the Six Sigma level.

Unless you're measuring something like the number of free-fall parachutes
that don't open in use, or the number of babies who die from being dropped
in hospital, this approach is not very helpful (advocates of Six Sigma
please note).

My guess is you're more likely to gain useful insights from looking at the
raw data than arbitrary definitions of 'failure'.

Regards,

Alan

Email: alan@landmarkconsulting.co.uk
Mobile: 07785 258 741