I T'S NOT HARD to get most project managers to agree that earned value systems are among the most useful and meaningful tools used to status, report, and analyze project cost, schedule, and performance. Unfortunately, their important benefits have too often been lost because learned value systems are thought to be too involved and costly to implement for many commercial projects. In part, this has occurred because earned value systems were invented to solve very real project measurement problems in cost-reimbursable programs. Thus, they have traditionally been cost-centric, reflecting the high priority that cost assumes in cost-reimbursable contract environments. But in many projects, time is the highest priority resource. Indeed, so important is time that it has become one of the most important elements of quality and value in modern business practice.
This article describes an earned value system that is a simple and practical program management tool when applied to time-centric or time-constrained projects. It is derived from “Earned Value—The Next Generation—A Practical Application for Commercial Projects,” which James R. Sumara and I presented at the 1997 PMI® Seminars & Symposium. The essence of the technique is to earn task starts and finishes. And, like all earned value systems, a principal benefit obtained and compelling reason to employ this tool is that the application of earned value stimulates project improvement.
John C. Goodpasture has 25 years experience in project and program management. He is the director of the project office for Lanier Worldwide Inc. and is responsible for project methodology, development and training of a professional project manager staff and for leading and facilitating projects assigned to the office.
Traditional Earned Value Measurements
Before looking at a new approach to earned value, let's review the traditional approach. The purpose of earned value is to measure accomplishment of the project's scope and objective and predict outcome at completion, using units of measure which are at the core of the value system of the project. The typical measurements are given in the PMBOK® Guide in Chapter 10.
Traditionally, the core value of the project has been money, and the focus has been cost. When used to measure project progress with cost metrics, the essential concept is: Each outcome is assigned a value during planning. This is the budgeted cost for the outcome. At the end of a measurement period, the accomplishment is evaluated. If it can be “claimed” that the planned accomplishment was indeed achieved, then the project is on schedule; otherwise it is behind or ahead of schedule in proportion to accomplishment. The degree of schedule variance is measured in dollars of scope-not-done-as-originally-planned, not time. As shown in Exhibit 1, the project, consisting of two tasks, is behind schedule by $15. Furthermore, the schedule “efficiency” is only 10/25ths, or 40 percent. To ascertain whether or not the project is over budget, the actual cost of what was accomplished is compared to the budget for the claimed accomplishment. In Exhibit 1, the project is $20 over cost, making the cost efficiency only 10/30ths, or 33 percent. Both cost and schedule are expressed in a common measure of value, which in cost-centric projects is money. What about simply measuring the variance between the budget and the actual cost for the period, rather than for each task? Although commonly employed, it really gives no information about project value attained, only about cash flow.
Although informative and powerful, there are many reasons why it is difficult, essentially impractical, to apply these conventional earned value measurements in commercial projects: no time-card or time-accounting system, poor timecard compliance even when it exists, no overhead calculation or cost allocation system, and many other reasons. So, in spite of recognized benefit, the redesign of the business to overcome these deficiencies is often impractical.
Cost-Centric Earned Value
Exhibit 1. This project is behind schedule and over cost.
Exhibit 2. Earned value claims depend on clear rules for claiming accomplishment.
Earned Start-Finish System
For many commercial projects, cost—at least product or service development and launch cost—has a much lower priority than time. Therefore, there is a need to focus an earned value management technique on the core value of time. Thus we come to the “earned start-finish,” or time-centric earned value system. In a time-centric system, the purpose is to earn the time elements of the project plan, specifically answering the two questions: Are tasks starting on time? Are they finishing on time? This start-finish time-centric system retains the most important features of the cost-centric approach: a focus on accomplishment; measurement of performance against plan; and a predictive outlook on completion.
Like the cost-centric system, the time-centric system depends on the information contained in the WBS, the project tasks, and project schedule. Each task within the project schedule is associated with a deliverable of the WBS, and measured by its successful completion. So, in order to be able to claim success, a task must not only start but must also finish with a result that is meaningful to the accomplishment of value in the project. But, how are claims to be measured? The recommended “accounting” rules are given in Exhibit 2. The strength of the earned start-finish technique depends upon the degree to which the project team's evaluation process for validating accomplishment “pushes back” on claimed performance. Thus, the claims are best done in a peer review meeting where the claimant makes the case. In this setting, the work package manager defends the claim of performance, demonstrating in concrete terms why it should be accepted that the task has started or finished. If it is, credit is given; if it is not, the task must wait to be claimed in the next measurement period.
Graphical Portrayal of the System. Exhibit 3 is an example project that shows data as cumulative planned starts and finishes and includes the cumulative actual starts and finishes. Using two information elements, the slope of the cumulative curve, and the planned start and finish dates for the remaining tasks, starts and finishes can be predicted for those tasks not yet complete.
For instance, from Exhibit 3, the following data is available: At the beginning of the first quarter, there is a variance of four task starts, 16 actually performed vs. 20 planned. Four starts were not earned. However, after the end of the first quarter, 28 task starts are claimed, and credited. The slope of activity in the first quarter is greater than planned, as shown by the steeper curve. In fact, 10 starts were planned in the period, and 12 were claimed, for a total of 28 starts claimed at the end of the second quarter. Within the resolution of the graph, for the third quarter, with only seven tasks remaining to start, and a productivity of 12/quarter, it is predictable that the project will complete all of the starts on plan. A similar analysis comes from the slope of the actual finishes. The actual curve is consistently seven tasks behind. The project is predicted to finish near the end of the next quarter.
Predicting Outcomes. The earned start-finish system is more limited as a management tool for prediction than is the cost-centric system. Limitations arise because cost is not included in the earned value equation, and so not all of the information potentially available in a project is used. But, this limitation can be mitigated by having a high degree of granularity in the project scheduling, thereby reducing the predictive errors in estimating the future from past performance.
Predicting future performance in one task from the past performance of another task depends on some correlation between tasks. At a minimum, project tasks are related through the project work package environment, the common resource pool, common project methodologies, predecessor conditions, and many other conditions of the project. As with all predictions, the extrapolation assumes that recent past behavior will be repeated in the near future. Of course, that is exactly the point! Do we want the past to repeat? If the predicted project outcome is not acceptable to the project manager or sponsors, then the earned value analysis stimulates the need for change to happen. From my own experience, I find that rarely does a project actually conform to the prediction. This is because, in the face of an unacceptable forecast, the project team takes the actions necessary to defeat the predictions and to put the project on a winning track. Thus, one of the best uses of all earned value systems is stimulating project improvement activity.
Exhibit 3. Performance claims are different from plan; the future is predictable from the past.
Exhibit 4. Real “finish” data shows the differences between planned performance and claimed performance.
Reader Service Number 053
Applying the System
Consider a project we call Process 2000. Process 2000 is controlled with a hierarchy of schedules. The summary level schedule is a networked Gantt chart of approximately 500 tasks. Only finish-start and start-start relationships are allowed; no finish-finish, start-finish, or constraint-date tasks are allowed. This facilitates identification of project dependencies, helps identify the true start and finish, and enables good statistical risk analysis. Exhibit 4 shows an example of the data from this real project and illustrates the finish performance of Process 2000 over a six-month period.
Like the cost-centric earned value system, the performance claimed is for all work completed in the evaluation period. It includes work planned to be completed in the period being evaluated, and work completed but not planned for that month. The latter can arise from work delayed from a prior evaluation period or from work accelerated into the current evaluation period. This accounts for the quite normal situation that there are claimed starts and finishes in periods for which there was no planned start or finish. These situations can be seen in the data. Furthermore, from the data presented, the actual association between a planned start and claimed start is lost in abstracting the data for the chart.
Conclusions and Lessons Learned From a Real Project
Consider this thought from poet Henry Wadsworth Longfellow: “Great is the art of beginning, but greater the art of finishing.” I found that it's much easier to start a task, or at least to claim credit for starting, than to finish. Consistently, we found starts more close to plan than finishes. Here is the problem with starts: It's too easy to call a meeting, circulate a memorandum or e-mail, or make a telephone call. But are these substantive when compared to what is necessary to kick off the work package or the task? This is a judgment call, and for this reason is best left to a peer review for a decision. But finishes are another matter. The consequences of a finish are often more concrete, and the subsequent tasks that depend on a predecessor to finish are more demanding.
I found that a constant and consistent focus on getting started and getting finished did indeed stimulate project improvement. Many tasks were reconsidered more carefully in the network of the project plan as the metric became better understood. The predictive nature of the curves did indeed stimulate project improvements to overcome missed milestones. Happily, these improvements invalidated predictions based upon past performance extrapolation.
THIS SYSTEM IS SIMPLE to apply, has relatively few requirements for tools, requires a minimum of training to be effective, but conveys powerful concepts about accomplishing valuable work in a project environment. Almost always, this leads to an improved level of project performance and greater satisfaction of the project's sponsors. ■