There are great lessons to be learned from monitoring systems that directly translate to people.
Take a look at this example graph:
When a CPU is running multiple tasks, it will optimize its performance to give - based on priority - adequate shares of time to each of its tasks. Some performance is required to operate (System Load) and some performance is required to buffer data in and out to operate the different parallel tasks (Context switching).
When the "System" Load is high, we typically state that either the hardware is not meant for the operating system or that the Operating System is ill-configured, usually the latter. Every operation invested into the System, and every operation invested into Context is an operation not available to complete any task.
People and machines
People do exactly the same. But how would such a diagram relate to people in organizations? Let's translate.
Every person must divide their working hours into their different assignments, plus everything that accompanies it:
Tasks and Projects
whether something is a machine's task, or a person's task - it's a task. From a different level of abstraction, being assigned to a project is a major task. Working on 2 or 3 parallel projects is considered the norm in many organization, so there's often multiple high level tasks going on.
Context
As people switch between projects, they have the same kind of context switching going on: close the thoughts on project 1, and pick up the thoughts for project 2.
Additionally, when they pick up project 2, other people will have made progress, so they need to remember where they left off and then learn what happened in the meantime. Until they are ready to do work on project 2, this will have shifted.
For example, if I work on project 1 on exclusively Mondays and on project 2 exclusively on Tuesdays, each time I pick up these projects, I need to catch up the events and changes of four days. Let's just say that I can do this in a single hour - it still means that my effective progression time has been reduced by 12.5%!
System
Adding it all up
Let's say I work on 3 parallel initiatives, and 45% of my capacity is usurped by the Operating Model.
Another 12.5% are taken by Context Switching.
That leaves roughly 1/3 of my capacity to do actual work on the project.
Given 3 initiatives of equal priority, only 10% of my capacity are dedicated to each single project!
Stopping the multitasking
Alternate Me who doesn't multitask is 850% faster to complete each project!
If project 1 takes Parallel Working Me 2 months, it will take Alternate Me 1 week.
If all three projects take Parallel Working Me 2 months, I have no results until 2 months have passed, then I have 3 projects to show.
Sequential Working me will have 1 project to show after 1 week - and 3 projects to show after 3 weeks!
Sequential Me can take an entire month of vacation, and still has capacity to complete a 4th project by time Parallel Working Me has completed 3 projects, working full-time.
Why do you run parallel projects?
Very thought provoking Michael. How do you deal with dead (waiting) times in projects? For example, the team wants to prepare a physical prototype for testing. In order to do this, a test mold must be constructed. This takes 3 months. If the person (or the team) is fully dedicated to this single project what does s/he do during this time?
ReplyDeleteThat's a very interesting question, and one that has no universal answer.
DeleteFirst things first, though: If the test mold "must be constructed", who is doing it, and why isn't the person doing this if that is the most important thing to do?
Note here: Work is never passive. Work is active. There is no "work must be done." There is only "work to do", so who's doing it?
Okay, now let's assume that a 3rd party agent is doing the mold. Of course that means that the work, from the person's perspective, is now in a "Wait" state, so then it does not require any capacity.
If, however, the work is still in a "Doing" state from that person, I don't understand the question about "what does s/he do during this time?" - Work maybe?
On a different level, we always need to challenge whether handovers and wait states are required. Every minute of "Wait" in the work means that business value is delayed. If our goal is to succeed as quickly as possible, a 3-month-wait looks like the first thing to investigate.
This comment has been removed by a blog administrator.
ReplyDelete