Critical Chain

Critical Chain Project Management (CCPM), developed by Eliyahu Goldratt (applying his Theory of Constraints to project scheduling), starts from the same task network as the Critical Path's Critical Path Method but schedules it differently — accounting explicitly for shared resource constraints, and restructuring where schedule safety margin lives.

The critical chain vs. the critical path

The critical path is the longest task-dependency chain, assuming unlimited resources. The critical chain is the longest chain once resource contention is also accounted for — if two tasks on separate, non-dependent paths both need the same specialist at the same time, one has to wait, which can make a resource-constrained chain longer than the pure task-dependency critical path. CCPM schedules against this resource-aware chain, not the naive network diagram.

Where the safety margin goes

The central CCPM insight: individually padded task estimates don't protect the project schedule the way they're intended to, because of two behavioral effects:

• Student syndrome — work on a task tends to start close to its deadline regardless of how much buffer was built into the estimate, consuming the padding before the task even begins.
• Parkinson's Law — work expands to fill the time available, so a padded estimate reliably gets fully used even when the task could have finished sooner.

CCPM's fix: estimate each task at something closer to its median (50% likely) duration — deliberately without individual padding — then pool the safety that would otherwise be hidden in every task into a small number of explicit, shared buffers at the chain level instead.

Buffer types

• Project buffer — placed at the very end of the critical chain, protecting the overall project finish date; consumed only when tasks on the critical chain actually run long, not by every individual task's private padding.
• Feeding buffer — placed where a non-critical-chain path feeds into the critical chain, protecting the critical chain from being delayed by a late-running feeding path.
• Resource buffer — an advance warning (not idle time) alerting a needed resource that a critical-chain task is approaching, so they're ready to start immediately rather than causing a handoff delay.

Buffers are actively monitored (often with a simple red/yellow/green consumption-rate chart) — a buffer being consumed faster than the project's percentage-complete is an early warning sign the schedule needs attention, well before the buffer would be fully exhausted.

Common pitfalls

• Adopting CCPM's buffers without removing individual task padding — the whole model depends on task estimates being realistic-not-padded; keeping both individual padding and a project buffer just adds total schedule length without the intended benefit.
• Applying CCPM without addressing multitasking — Goldratt's broader argument is that a resource juggling multiple concurrent tasks (aiming to keep everything "moving") actually delays every one of them compared to focusing on the critical-chain task first; CCPM's discipline expects resources to prioritize critical-chain work above other work when they conflict.
• Confusing critical chain with critical path when reporting status — they can identify a genuinely different bottleneck chain once resource constraints are considered; using critical-path terminology/tools while actually trying to manage resource contention misses the problem CCPM is built to solve.
• Treating buffer consumption purely as a lagging metric — the value of the buffer chart is as an early-warning trigger for intervention, not just a retrospective record of what already happened.

Learn more

• Eliyahu M. Goldratt, Critical Chain (1997) — the original text introducing the method (as a business novel, in Goldratt's usual style).
• Critical Path for the underlying unconstrained-resource scheduling method CCPM builds on and diverges from.
• Project Management for the broader scheduling/risk context.