Projects are getting larger and more complex, and therefore, more risky. Rigorous and systematic risk management might not uncover all the needed information to understand known unknowns adequately, or to uncover unknown unknowns. Time to market demands drive projects towards shorter schedules, encouraging PM's to place more tasks in parallel, make bolder assumptions regarding how the project will proceed, spread limited resources between as many tasks as possible, etc. All of these actions only add further to the project's preexisting risk level. What a PM needs is a means of assuring the attainment of their project's objectives.
All projects are planned to meet the same basic four objectives: deliver the desired scope and quality, for a particular cost, and within a specified time period. To achieve this, project managers employ a variety of techniques and tools. One very established tool is Earned Value Management. Now defined by ANSI/EIA-748-1998, this approach is over 30 years old. Earned value (EV) provides a means to forecast project performance through the use of cost and schedule performance indices, which is developed by tracking the accomplishment data associated with the project's tasks. Variations to plan highlight either risk triggers that are taking place, or risks that are manifesting themselves into problems. Failure to address either will result in the project's failure.
Then there is the “newcomer” to project performance measurement. Eliyahu Goldratt has taken his Theory of Constraints (popularized in his business novel The Goal) and applied them to project management in his most recent book Critical Chain. He sees this approach as a means to manage “project uncertainty,” particularly in the area of schedule performance. To do this, Goldratt uses a very different means by which to structure, manage and forecast a project's completion. True to its Theory of Constraints (TOC) roots, critical chain (CC) focuses upon insuring the project's end date is attained (the project's goal) and does so with a minimum number of metrics. In contrasting EV against CC, a student of TOC would say that earned value collects many “trivial” numbers instead of the “critical few” important to measuring the project's goal (i.e., the completion deadline).
At first glance, they seem to be mutually exclusive. This, combined with the “Siren Song”of less data and infrastructure then what is commonly associated with earned value, makes it very tempting to shift exclusively to Critical Chain. Might there be, however, a few more considerations before making your decision? Which is right to control schedule performance? Are they really mutually exclusive? Have you considered the possibility that they might be complementary?
Before exploring these questions, a little background is needed on each of these two approaches.
Background—Earned Value Management System
The Earned Value technique as we know it has over 30 years of application to complex government projects and programs. Its roots start in the U.S. Air Force Cost/Schedule Planning and Control Systems Criteria (C/SPEC), which were later embodied in the U.S. Department of Defense Cost/Schedule Control Systems Criteria (C/SCSC). Built around 32 “criteria” (there were 35 until a few years ago when three were eliminated), these criteria defined the essential elements of an Earned Value Management system. More recently, the American National Standards organization has captured these same criteria in its ANSI/EIA-748-1998 standard.
In its simplest form, EV uses the values that were estimated for the various project work packages as the amount that will be “earned” as work is completed. The process typically starts by determining all the project deliverables. These are converted into work packages, which are then estimated (in hours or dollars) for their resource requirements. Prior to baselining, the earned value method that will be used for this work package is selected based upon which method best approximates how the actual expenses are expected to accrue. (This keeps minor plan versus actual differences from exceeding the threshold requirements for a variance report.) As work is completed, actuals are collected in cost or control accounts. Simultaneously, as tasks are started, completed, and/or milestones met, the rules applicable to the EV method selected are applied. This creates the value for the work “earned” for a given reporting period, which is then compared against the actual expenses to determine the cost variance, and against the project's baseline curve for schedule, or more accurately, “work” variance. Variances exceeding threshold limits are investigated as to their root cause, problems identified, and the plan adjusted for these new realities.
The preceding steps produce three values each reporting period: baseline plan, earned value, and actual costs. The cumulative values can be used to create two performance indices showing how effective resources are being converted into results. The Cost Performance Index (CPI) reports the efficiency at which the project team converts resources into work and the Schedule Performance Index (SPI) reports the effectiveness of the team at converting the planned work into results. These past results are then used as a proxy of what the future performance will be and are used to forecast when the remaining work will be accomplished and for how many more resources.
Earned Value Management Systems (EVMS) has extensive use in large government programs, particularly where progress payments were made. Prior to EVMS, it was difficult for the customer to validate the amount of tangible work that was accomplished for a given invoice. Sometime, payments had to be made with limited physical evidence of progress. By utilizing EVMS and good project management disciplines, objective values for the intermediate work products are determined prior to the project's start. These values can then be used to validate the reasonableness of the invoices being presented to the customer. The significant content of financial information, and its use as a means to audit invoices, has resulted in most EVMS systems being maintained/administered by the finance department. Requirements to maintain audibility results in a significant overhead, which was considered reasonable for large complex programs.
Background—Critical Chain Project Management
Critical Chain Project Management (CCPM) is one of the newest techniques to become available for PMs to manage their projects. Eliyahu M. Goldratt, an Israeli physicist, first popularized his “theory of constraints” (TOC) in the business novel The Goal. Critical Chain (CC) is a direct descendant of this material, so a thorough understanding of the principles behind The Goal makes CC far easier to understand.
The Goal used the Socratic method of learning to involve the reader into its concepts and to “learn” along with the novel's main character, Alex Rogo. In the story, Alex learns that his factory is simply a process with one or two constraints that limit production capacity and sales. If he wants to improve factory output/sales, the constraint's capacity must be improved. Improvements to non-constraints is actually wasteful in not one but two ways:
1. Capital investment has been sunk into improving something that already had sufficient capacity to exceed the throughput of the factory's constraint(s).
2. This improved non-constraint, like other processes, will be measures to insure full utilization. This incentivizes it to produce even more than before, resulting in the level of corporate assets tied up as work-in-progress inventory increasing, which also increases the need for storage space, inventory personnel, losses due to damages, obsolescence, etc.
This is where the book's title is highlighted … “the goal” for a typical business is to make money. The assumption that most organizations follow is that if sub-components are improved, then the organization as a whole must improve. This approach fails to consider the impact of a process constraint (i.e., only improvements here yield results). It also fails to recognize the role the constraint plays in generating revenue (i.e., every minute the constraint is not producing represents lost revenue for the company as a whole, which can never be made up).
This last point is key to understanding Critical Chain Project Management. Its very name is drawn from the adage “no chain is stronger than its weakest link.” As the weakest link defines the overall strength of a chain, the constraint defines the overall strength of a process. Improving the non-weakest link does not improve the chain's strength … it only consumes resources and makes the chain heavier! In The Goal, the weak link involved equipment capacity. In Critical Chain, the weak link involves limited resources available to the project team. In both cases, this limited capacity defines the maximum output of the affected process.
Application of these principles to project management is a simple extension. A project has a defined goal … to produce a given end result by a particular date. It follows a process, utilizing project management disciplines and resources, to accomplish its goal. Somewhere within this process, a limited number of constraints define the project team's throughput. The key is knowing where the constraint is, getting it working at maximum efficiency, having no constraint “down-time,” and not having the non-constraints overproduce.
The challenge facing those applying TOC principles comes not from needed changes in equipment and/or processes, but in the participant's behavior. In The Goal's factory setting, revenue was not improved through capital investment but through behavior changes in the constraining resource's users. The same is true with CCPM, where the focus is shifted to optimizing the component's performance relative to the entire system. Specifically, metrics is changed to reinforce desired behavior, which supports the achievements of the larger (system) goals. In CCPM, the behaviors that must be changed to make project management successful are:
1. Managing Constraining Resources: The constraint that dictates process output must be always utilized (i.e., its time must not be lost). This means it arrives when expected, never waits for required resources, and is not tasked with work that can be completed by others.
2. Protectionism: Overestimating a task's duration to make sure it is completed on time. This ties up resource time and places safety time at multiple places throughout the project schedule. This extra time needs to be removed and located in key points within the project.
3. Elimination of Rework: Projects are started with limited knowledge. If resources are allowed to start as early as possible, rework may be required because of new project knowledge, and/or this completed work might unduly influence future decisions for tasks not yet started. By delaying this work, it can incorporate all known information and/or have limited influence on more critical project components.
Exhibit 1. EVMS vs. CCPM Strength and Weakness Matrix
4. Higher Productivity—No Multitasking: Working on multiple tasks at the same time results in lost productivity. These workers must constantly remember where they were each time they restart the stopped task. By working on only one task at a time, this lost resource time is regained.
5. Higher Productivity—Working Toward Effort Estimate vs. Deadlines: Working towards deadlines results in what Goldratt calls “Student Syndrome” (i.e., “cramming for the exam”). We know about the deadline, but do not fully apply ourselves until much of the estimated task performance period has elapsed. Then we work overtime to try to make the deadline.
6. Capitalizing on Early Finishes: The law of averages calls for an equal number of tasks completing early and late. Late finishes are always passed along to subsequent tasks. Early finishes are not always captured unless the successor task is ready to and does start early. If only negative schedule variances are realized, then projects will end late. Positive variations must also be captured. The process starts by modifying a traditional schedule to consider the project's constraint. Estimates are reduced to remove the protection, and safety margins are centrally located in a “project buffer.” Tasks utilizing the constraining resource are linked, forming the project's critical chain (not the critical path). “Resource buffers” are placed to make sure constraining resources arrive when expected. “Feeding buffers” protect the critical chain by making sure tasks providing output to the constraint arrive early to on-time. Early completion opportunities are identified and successor tasks prepared to start early. Tasks are started as late as possible (still with their respective feeding buffer) so that rework is limited. Multitasking is never part of the base plan. During the project, participants are managed to the effort estimates, not the task's due date. The key management measurement is the size of the different buffers, particularly the project buffer, which indicates the project's overall health. What makes this metric different from many traditional project metrics is that it is directly measuring and providing continuous feedback regarding “the goal” of the project … when it will be finished.
Neither One is Perfect
Most people consider EVMS to be very complete. It focuses on extensively documenting completed events, which it then uses to forecast (predict) the project's future. To CCPM supporters, however, documenting the past is not the project's goal and that these activities contribute very little towards insuring its on-time completion.
CCPM users like its focus on the future and the limited number of required metrics. To EVRM practitioners, however, it might be considered overly simplistic. Goldratt is largely silent on the topic of budget. It is the authors' belief that this can be traced to The Goal and his dislike for managerial accounting and the wrong decisions to which it can lead. In the factory setting, if all costs are being addressed by planned sales, then new work utilizing existing resources, but not requiring the constraint, can be completed for any monetary gain greater than the material content. This same model would apply in Critical Chain. The product development team's costs are addressed by the company's production activities. Since these costs are now immaterial, he does not discuss, nor does he need to address them. While true in this case, it does not hold true in all project environments.
Likewise, other PMBOK® Guide disciplines, such as risk, communications, procurement, and integration management, are not specifically mentioned in CCPM. For example, Goldratt speaks often of “uncertainty” in Critical Chain, but most of this is centered around the uncertainty of on-time project completion. So while schedule risk is addressed via the buffers, their sizing is not based upon any risk management methodology. There is no discussion regarding proactive risk management activities within CCPM, only reactive buffer management. So while these areas are treated very superficially, it does not seem reasonable that they can be ignored. Good project management disciplines should be the foundation for any effective methodology.
Model to Contrast Earned Value and Critical Chain
Both EVMS and CCPM have their advantages and disadvantages. So, is one more appropriate than another?
In trying to answer this question and highlight the differences between these two approaches, a model was designed to contrast them. The various strengths and weakness were collected from the authors' personal experiences, literature, brainstorming and perceptions. This information was then entered into a 2x2 matrix, with CCPM on the horizontal and EVMS on the vertical, with strengths and weaknesses as binary attributes along each axis. Characteristics from each were then identified and linked to one another so that any existing patterns would be emphasized. The results are shown in Exhibit 1.
As the matrix shows, EVMS and CCPM share some strengths and weaknesses. Both require good discipline, high levels of resource planning and have a variance analysis process built in. Neither prevents the problem of scope creep, has a built-in (complete) risk management system, or has a process to address dissimilar participant goals. As expected, greater insight was gained by examining the differences, or where they were opposite to one another,
Examining the upper right corner that contains the EVMS strengths and the CCPM weaknesses, it is clear that EVMS's greatest value comes from its ability to objectively track and measure the value of the completed work. This allows benefits such as estimating model validation, and milestones to measure intermediate progress. This contrasts with CCPM's apparent reactive risk management style to schedule variations, use of point estimates, and limited history with large, complex projects.
The opposite condition brings out CCPM's strengths in the area of measuring, forecasting and insuring the project's completion date. It is also independent of the problems typically encountered when trying to use financial data to make managerial decisions, requires a project buffer, which helps address schedule variations, and only requires current schedule performance information with which to predict the project's completion date. EVMS is heavily dependent on financial data (when its analysis is based upon cost), and its use of cost/control accounts does encourage participants to think in terms of their limited area of responsibility. It also makes no overt effort to discourage multitasking and/or require management reserve.
EVMS vs. CCPM … Which One is Right for You?
Summarizing Exhibit 1 to its core elements, CCPM is focused on providing the user with a project end date that can be counted upon. EVMS collects large volumes of actual performance data to determine if the work is being accomplished as planned, and from this, it forecasts the project end date. With both providing a forecasted project completion date, these two approaches appear to some as doing the same thing. So which one is best?
Exhibit 2. EVMS vs. CCPM Conflict Resolution Diagram
To uncover the root of this conflict, the authors created a Conflict Resolution Diagram based upon the principles that are found in some of the other Goldratt publications. While certainly not experts at this approach, the diagram (Exhibit 2) clearly shows the apparent “Catch-22” into which the project manager is being placed. This dilemma, however, was created trying to address two issues with a single approach.
“How much real work has been completed so I know if I should pay this progress payment invoice?” was a critical question that the financial and contracting officers needed for large, complex government programs. In the early stages of some projects, physical evidence of progress against which to compare the invoice that has been presented for progress payment is limited. Earned value very effectively answers this question.
The other need was to have some idea as to the project's completion date. Earned value was seen as an economical means to fill this need. Most of the earned value users of this era had financial accounting backgrounds. They were already versed in many other decision-making techniques utilizing financial data and were comfortable with this. This satisfied the need for completion date forecasting.
What begins to stand out here is that projects have multiple interested parties with different, but not necessarily competing, needs. Those charged with ensuring fiscal responsibilities want to know if they are paying a fair price for what has been accomplished. This is not suboptimal to the organization's goal of completing the project for its anticipated benefits. If the project performance is unsatisfactory, the organization will want to have the lowest possible investment expended upon the failing approach. Having zero investment is unlikely, as suppliers do not typically undertake large, complex, high-cost projects and wait for a single, final payment at project completion.
The other interested party (group) is those charged with achieving the project's goal. They are the ones responsible to ensure the project's benefits are achieved. This too is an organizational level goal, as success benefits all, including the financial group.
These two different needs, often housed within two different functional areas of the customer's organization, are not competing or in conflict with themselves. The conflict comes from management's decision to satisfy these needs with a single approach. The assumption that is being broken here is that there must be only one approach.
Integrating Earned Value and Critical Chain
Once accepted that both needs are real, and that one of these requirements cannot be “eliminated,” it is clear that the project manager needs to address both effectively. Until such time as there is a single integrated tool, what's wrong with using both? With the less than stellar performance of projects in some industry segments, it is time to reconsider the decision to use a single approach, which then requires tradeoffs when applying it towards satisfying the other need for which it was not designed. Additionally, with computing costs continuing to decline, the penalty associated with operating dual systems is not of the same magnitude as it was in the past.
What are the issues associated with operating dual systems? Much of the foundation planning is the same. Both require the plan to be created in detail and an estimate of resources determined and leveled across the project. Task precedence must be identified and entered into the scheduling tool. It is not until the end of their respective planning phases that they diverge, with EVMS assigning earned value rules and values for intermediate deliverables and CCPM requiring the critical chain to be identified and buffers created/inserted. Intermediate progress report information (i.e., task completion information) would still be used by both approaches.
So, how would these two techniques be integrated? The key is for the PM to use them for the purposes in which they were best suited. EVMS should be used for contract performance issues, such as:
• Supporting progress payments by documenting cost incurred vs. value of work completed to date.
• Reporting progress against key project/program milestones.
• Validating the estimating methodology on which the project was based, so that future estimates are more accurate and any estimating risks yet to be encountered within the project are identified early.
• Variance analysis for project performance issues.
• Providing information to those charged with financial oversight.
CCPM should be used to get the most from the project's process relative to work to be performed:
• Managing the constraint(s) for maximum throughput, benefiting the project and the organization.
• Protecting the constraint from lost time by ensuring work arrives ahead of its needs.
• Ensuring critical resources arrive when expected.
• Frequent monitoring, forecasting, and protecting of the project's completion date.
• Providing information to those charged with overall project performance.
Both systems can still benefit from increased emphasis in the areas of:
• Scope and change control and a thorough understanding of the project's scope/requirements
• Risk management
• Use of three-point estimates and Monte Carlo simulations.
In summary, EVMS should be used to report the past achievements and CCPM should be used to manage/report on the future. This directly addresses the project manager's requirement to be responsive to the specific needs of these two “customers.” There was nothing identified within these two techniques that could be considered mutually exclusive or preventing this combined approach. Additionally, the behaviors discussed in Critical Chain background section provide improved productivity benefits under both techniques.
Conclusions
The key challenge for the project manager having both of these systems available on the same project is the discipline necessary to use them for their best-intended purpose (i.e., EVRM for the past and CCPM for the future). As long as the project manager does this, there should be no added risk to the project by having both in use. It is more likely that having both reduces risks on large, complex projects by having a clear picture of the future, while having an accurate record of the past from which to do variance analysis.
These two approaches also effectively fit into the process improvement aspects of a project's lifecycle. While often viewed as linear, a project can also be considered from a cyclical point of view by overlaying upon Deming's Plan-Do-Check-Act (PDCA) cycle. This is easy, as all projects are looking to incorporate feedback with which to improve future efforts. The planning needed to implement these two techniques has already been discussed and falls into the “Plan” section. Goldratt's behaviors provide a means to gain improved performance in the “Do” portion of the PDCA cycle. Having independent approaches that address the two customer “Check” needs is not unreasonable for high investment projects. This provides the needed feedback from which to create new actions (“Act” Phase) upon which to base planning updates. This improvement cycle continues until project completion.
Not all projects and their management will share the belief that the cost of administering both techniques on a project will be worth the potential benefits. These individuals will be willing to assume the risk that comes with applying just one technique to satisfy the different stakeholder needs. For them, there are many challenges yet to be faced. During a meeting hosted at Wright-Patterson Air Force Base on March 9 and 10, 2000, the authors were afforded the opportunity to meet with others trying to answer this very question. This list of issues is long and many answers remain to be worked out (i.e., Do you “cost-out” buffers, and if so, how do you arrive at the resource value to be used?). These types of questions will need to be resolved before a single tool can be employed. It is the belief of these authors, that this compromise is unnecessary and the associated risks are not warranted for the limited savings. Of course, this would become a moot point should someone develop a single, integrated tool.
The matrix of strengths and weaknesses found in Exhibit 1, upon which this paper's conclusions are based, is certainly not an all-inclusive list. We tried to identify as much information from as many sources as reasonable. Some readers are certain to identify additional items that were omitted, while others might take exception to the binary strength and weakness categories. Additionally, this is not a static environment. There is evidence that the traditional prescriptive guidelines of EVMS are softening toward more descriptive ones. Similarly, CCPM is becoming more refined with greater use. At a high level, however, we do not see these issues changing our general conclusion … they do not compete, but support key project objectives and each meets the needs of specific stakeholders in their own way. The challenge will be in changing the behaviors of the team, particularly with regards to managing to effort, not milestones. This will require the greatest effort.