Suppose somebody in an
organization is not busy. What happens? They are given more work to do. They may be given work that is part of their
current process, or it may be work from another unrelated process. What probably wouldn’t happen is an analysis
of why their current activities did not keep them busy.
Because of their roots in paper,
the activities of most information processes are performed in a serial order. Each activity is executed one-at-a –time and
the next not started until the prior is completed. Since paper was expensive to copy, parallel
processes were rarely used. Because of
the process rings-of-a-tree phenomenon, the serial process flow model is still prevalent
even after paper is eliminated.
A natural characteristic of any
serial process is that it creates bottlenecks.
An example of this is a highway. Where
one section of a highway can be smooth sailing, the next section can be stop-and-go,
followed by a section where traffic is again flowing freely.
A bottleneck in a serial
process slows the amount of work getting through just like a bottleneck in a
highway. If work starts to pile up
behind a bottleneck, the people downstream of the bottleneck will soon run out
of things to do. Then what happens? They are given more work to do.
Another characteristic of a
serial process is that bottlenecks can independently move up and down a
process. Watch the bottlenecks of a
highway and you’ll notice that some bottlenecks always occur at the same
location while other bottlenecks appear to move randomly up and down the
highway.
The same occurs for serial information
processes. One day a bottleneck could be
located at one activity in a process; and the next day at another. Bottleneck locations change as the mix of
work and the available resources change. If a bottleneck moves, people with plenty of
work to do, may suddenly not have enough.
Then what happens? They are given
more work to do.
As bottlenecks shift from one
location to the next, the cycle of handing out more work to keep people busy continues. If a bottleneck is temporarily removed so
that someone downstream now becomes too busy, nothing is usually done other
than to ask them to work harder. But the
bottleneck moves again so that the same person now has too little to do, they
are given more work.
Finally, the ongoing effort to
make sure everyone is busy results in a status quo where everyone always has
too much to do. This status quo is
called process spaghetti.
Figure
1.10
is a depiction of the bottleneck principle of process spaghetti. Process A, B, and C are each serial processes. Each box indicates one or more activities
executed by one or more people. The
first activity of each process is numbered A-1, B-1, C-1; the second A-2, B-2,
C-2, etc. Note that the same people are
responsible for executing the activities of A-2, B-2, and C‑2. In this example these people originally only
had activity B-2 to perform. But over
time they were given the work of A-2 and C-2 to keep them busy.
People downstream of a bottleneck
are given more work to stay busy until they too become a bottleneck. This principle applies until all process activities
are bottlenecked.
As process spaghetti begins to
develop it causes even more spaghetti.
Suppose the people responsible for activities A-4 and B-3 were only
recently given activity A-4 because of a bottleneck at activity B-2. Since it’s a new activity for them they want
to make sure it’s done well, so they decide to do all of the A-4 work before
any B-3 work.
Now under this scenario, the
people responsible for activity B-4 no longer has enough to do, because B-3 is
now bottlenecked. Then what
happens? They are given more work to do.
Process spaghetti once formed
is very difficult to unravel. Suppose
there was an effort to remove the bottlenecks from Process B in Figure
1.10. If the people responsible for activities A-2,
B-2, and C-2 were told to prioritize B-2 work over everything else, then a
logjam of work would suddenly flow to B-3.
The size of the bottleneck for the people responsible for A-4 and B-3
would increase. In addition, work
flowing to A-3 and C-3 would be reduced further; bottlenecking those processes. The net result of removing the B-2 bottleneck
could be a reduction in the value created by all three processes.
A lot of progress can be made
removing bottlenecks before anyone notices a change in created value. It can take months to clear a single process
of its bottlenecks without impacting too negatively otherwise seemingly
unrelated processes.
The time and cost it takes to
unwind process spaghetti is not the only issue.
It can also be difficult to sustain management commitment. Bottleneck removal projects can quickly run out
of steam when management runs out of patience waiting for a positive result
from a Process B; while hearing about all the problems cropping up with Processes
A and C.
No comments:
Post a Comment