Tip of the Week #26                     Tip Index

Go to the Prior Tip "Monte Carlo Stopping Rule"
Go to the Next Tip "5-Steps in the Creative Process"
Return to MaxValue Home Page

Critical Chain
by Eliyahu M. Goldratt, 1997, North River Press, softcover, 246 p., $19.95 from the publisher or bookstores.

This is a 'must read' book for managers and others interested in effective project management. There are several important concepts and insights to be obtained here. The lessons in the book are so good that one can easily tolerate several flaws (my opinion).

Critical Chain is a novel about a young business professor. He and his Executive MBA class learn together, using the Socratic method, as they discover and apply radical new project management concepts. The principal side characters are a young task force charged with dramatically reducing a modem manufacturer’s time-to-market cycle.

As with Goldratt’s earlier novel-style books (best known are The Goal and It’s Not Luck), reading the story is an entertaining way to learn. Plan to invest a whole day of reading and reflecting, if possible. This book is worth it.

Key Ideas

• An information technology saying: "A project will run out of time but will never run out of excuses." (p. 26)
• Why are projects late? People blame uncertainty and unforeseen risks. Top management blames externalities, while project participants more-honestly place direct blame on internal politics and management.
• People usually build safety into their estimates when estimating times-to-complete for project activities. Typically, the estimate is judged at an 80% confidence point. That is, the estimator believes there is an 80% chance that an activity will be completed within this time.
• With all this safety built into the scheduling estimates, why are most projects late? Because (1) most safety is wasted and (2) because people do not focus on critical chain activities and resources.

The Critical Chain Approach

The critical chain is the sequence of activities—with explicit consideration of constraining resources—that determines the length of the project. The critical chain would be the same as the customary critical path except that some  resources may be unavailable when needed. The critical path constraint model is modified to include sequencing scarce resources.

• Use medians for activity estimates (instead of highly confident estimates). Determine the critical path (the series of activities that control the project completion time).
• Put safety into a project buffer at the end of the project. Also put buffers at the end of "feeding" activity paths that merge into the critical chain. The constraint is the critical chain, and this is where most attention should be focused.
• Do not accept "work performed" as project "progress." As in other of Goldratt’s teachings, the measurement is often the cause of the problem. Project reporting should focus on efficiently accomplishing those activities on the critical chain. Remind people about when their work on critical chain activities is to begin, and ensure that this work begins promptly, has full attention, and is uninterrupted.

Additional Ideas

I found these additional concepts interesting:

• "End-of-month syndrome." In many organizations, people start the month with good cost control (controlling inventory, scheduling without overtime, etc.). However, throughput control (meeting delivery commitments) becomes paramount near the end of the month. [This seems analogous to changes in my work style as a deadline approaches.]
• We naturally recognize uncertainty in project activity times. Unfortunately, early completion is usually not reported. Of course, late completion (delay) is passed on to the next activity. Safety is regularly wasted by: (1) procrastinating the start of non-critical activities ("student" effect), (2) multi-tasking (inefficiently working on multiple activities simultaneously), (3) dependency effects (not reporting advances, and passing on delays).
• Synchronizing (coordination) is another time waster. This provides another reason to avoid unnecessary multi-tasking.

Goldratt is recognized as the creator (or at least promoter) of "Theory of Constraints" (TOC). His landmark book is a novel called The Goal (over 2 million copies, authored with Jeff Cox), is about de-bottlenecking manufacturing operations. His process, whether talking about inventory or time, is essentially the same:

1. IDENTIFY the constraint
2. EXPLOIT the constraint (make sure the activity being worked upon, or the constraint, is being done as best possible with the present resources).
3. SUBORDINATE other activities to the constraint. Don’t start activities (production) prematurely.
4. ELEVATE the constraint: Look for ways to shorten the time required at the constraint.

The analogy between production management and project management is appropriate and effective.

Some Issues

Despite some great ideas, Goldratt comes across to me as eccentric. Maybe the mystique is part of his appeal. Here are several of what seem to be problem areas in Critical Chain that detracted from the good material:

• Goldratt doesn’t recognize that critical paths and critical chains can be dynamic. Monte Carlo simulation software is widely available for calculating probabilities that activities and resources will be critical. Critical Chain and The Goal would have been more timely in earlier days before Monte Carlo scheduling and materials requirements planning (MRP) software became available.
• He describes the (biased) median, not the (unbiased) mean, as the single-point activity forecast. Perhaps he is sanitizing his book with an easier-to-explain though incorrect estimator statistic. Correlation between activities is also omitted.  Of course, I recommend that the expected value (mean) be used for forecasting. In the context of project schedule safety, perhaps activity time estimates for critical path scheduling should be conditioned upon not having low-probability, high-consequence contingencies occur.  Activity path slack times (safety or contingency) are buffers for normal schedule variations
• Goldratt says that we can measure project uncertainty by the length of the project buffer. This is a rather arbitrary indicator of the range of the project completion time unless one has information about the probability distributions. If we know (or judge) probability distributions, then we can do a much better job of forecasting by using them.
• Being famous for identifying the right performance measures, surprisingly, he doesn’t offer a way to measure the quality of the activity estimates. How are we to improve our project forecasts without a quality measure? [My preference is to use ratios of actuals/estimates as feedback.]
• Goldratt takes a seemingly unjustified shot at the bedrock of modern financial analysis: net present value (NPV). Without much explanation, his character proposes "flush," a product of days times dollar amounts, as a way to characterize tradeoffs between time and money. Besides not measuring value, his approach omits the obvious need to incorporate compounding.

Despite the several distractions, this book provides insights to important issues. Reading Goldratt has always been provocative. I’m recommending this as essential reading to everyone working in project management.

This book has been receiving wide acclaim.  A review, expanding greatly on Goldratt's novel, is "Bringing Discipline to Project Management" by Jeffrey Elton and Justin Roe, Harvard Business Review, March-April, 1998, pp. 153-159.  They caution that projects are seldom production processes.  The challenge is to manage a portfolio of projects in a sea of innovation.  New technologies, changing requirements (e.g., Internet software), and people issues confound logical approaches.  "Critical Chain starts with a set of talented and driven project managers and assumes the resource constraints are inside the work of the project, not in its leadership.   In truth, leadership may be the larger constraint."

See Tip 39 on a related book,
Project Management in the Fast Lane: Applying the Theory of Constraintsc

—John Schuyler, November 1997.  Revised March 2000 and Jan. 2006

Copyright © 1997-2006 by John R. Schuyler. All rights reserved. Permission to copy with reproduction of this notice.