PM Post: A Relay Race, not a Train Schedule
Most projects are managed by carefully watching the calendar, comparing where we are today against some baseline schedule. That schedule typically consists of a series of start and due dates for consecutive tasks, with due dates of predecessors matching start dates of successors. Like a train schedule, if a task arrives at its completion on or before its due date, that portion of the project is considered to be “on track.” Successor resources plan other work and their availability around those dates. If the predecessor is finished early, the successor resource may not be available to pick up the handoff. Even if the resource is available, there is commonly little or no urgency for the successor to start (or to focus on it exclusively), since we’re “ahead of schedule,” and that resource will typically tend to other priorities.
The problem with this common practice is that while it is important for trains to arrive at and depart from their stations (their milestones) at appointed times, project value is more often tied to the absolute speed from beginning to end. The sooner the entire project is completed; the sooner project benefits can be accrued. A more appropriate metaphor to guide projects is a relay race, in which resources are encouraged to pick up the input handoff as soon as it is available, “run with it” in a full, focused, sustainable level of effort, and hand off the output as soon as it is complete.
This behavior is exacerbated in environments where schedules are built upon estimates that are considered commitments by the resources, and therefore contain a substantial amount of localized safety in each task to protect that commitment. If a project is deemed “on track,” and a resource realizes that there is chance of completing the work well within the “safe” estimate, the desired sense of urgency is again diminished. As a result, resources are momentarily comfortable sharing their time among several tasks or issues, extending out the time that they would otherwise be able to hand off their output to the next leg of the relay race.
Milestone schedules, like training schedules, become, at best, self-fulfilling prophecies, at least in terms of expectations of speed. They may still (and often do) take longer due to being derailed by Murphy’s Law because they have wasted what might have been early finishes which are now not available to offset tasks that take longer than anticipated.
Critical Chain Schedules, Resource Behaviors and Risk Mitigation
Schedules using the Critical Chain approach to project management address this question of lost safety in two ways. First, the usual system of task due dates itself is eliminated. The only dates in a critical chain schedule are launch dates for chains of tasks that have no predecessors, and final due dates associated with deliverables that are external to the project and which are protected by project buffers. Start dates of tasks are linked directly to the completion of their predecessors, and communicated through the buffer management project control process, discussed in more detail later in this paper. If you have no due-dates, you have gone a long way in eliminating due-date behaviors and in repealing Parkinson’s Law.
Secondly, the safety is moved out of the tasks to the buffers, thereby eliminating the idea of commitment that needs to be protected on one hand or that is good enough on the other. With the underlying assumption that the work of a task will take as long as it takes, no matter what the schedule model assumes, resources are directed to work on tasks without distraction until complete and handoffs are delivered. At least tasks won’t be delayed by outside influences. More importantly, management also must support the ability to do so, avoiding unnecessary distractions or conflicting priorities.
If resources run their leg of the relay race in an effective and efficient manner, some tasks will take longer than anticipated in the schedule and some will take less. The project is in a position to take full advantage of early finishes. In this way, the cumulative risk associated with due-date behaviors is replaced by the consumption and replenishment of buffers.
[Adapted from a paper originally written and presented in 2002. Still valid, worth repeating at least once a decade. At least…]
