Putting this all together, here are the top three
things that I look at to determine good design.
Complex end to end process have to be partitioned
and responsibility for the overall process is shared. Wherever
there are organizational (ownership) lines drawn you have a high
potential for the process to fail (where interdependent activities
cross organizational boundaries). For example when order management
passes a work order to manufacturing and the customer calls manufacturing
directly to expedite at the cost of other orders. I look for mechanisms
that will prevent people from withdrawing from the boundary and
throwing problems 'over the wall'. These mechanisms include, education
in the total process, well defined end to end measures, well defined
feedback loops at border points, and documentation of informal
work practices. Sales, manufacturing, and the customer need to
see what is happening, understand the overall consequences, and
redesign the process to meet all of their needs.
The more stakeholders that can participate in the
design, the more stable the implementation, and the more adaptive
the design. I try to go at least two steps beyond the primary
design focus. By including customer processes and vendor processes
we can see how the overall design fits together. Many times this
can encourage innovative partnerships across traditionally difficult
boundaries.
This is a technique to explore future uses of the
process design. Everyone has an idea of the future, hopefully
a vision of what could be, and an idea of what could go wrong.
Drawing on these ideas you can stress test the design. When I'm
designing an order fulfillment process I might ask; what if I
open a sales office in China?, what if I expand my business to
provide order fulfillment processes as a service?, what if a large
order is canceled at the last minute?, what if there is a transportation
disaster?. The benefit of this is not so much predicting the future
(that these things will happen), but in exploring how well the
people and process will adapt. Will it become obsolete over time,
or adapt in a way that improves over time?