Are you being misled by your progress Gantt chart?

by Tammo T. Wilkens, PMP

ON ONE PROJECT I WAS INVOLVED WITH, the client and the project manager were constantly misinterpreting the traditional progress bar shown on Gantt charts generated by CPM software. It got to the point that I removed the progress bar from the charts to avoid meaningless questions. However, management insisted on some form of progress display.

My position is that the traditional method of displaying schedule progress on a Gantt chart, based on percent complete of the activities, shows only a small part of the progress picture. It does nothing to answer the question, “How far behind or ahead of the planned schedule is the work?” Additionally it often misleads the project manager into thinking the opposite of what is true. There is a method, however, that communicates the whole schedule progress picture without encountering this misleading problem.

The Problem. Most, if not all, currently available scheduling software has a feature to display progress in a contrasting color or pattern over the duration of the activity in question, as in Exhibit 1; some people refer to this as the thermometer effect, owing to its appearance. This progress bar is often quantified as a percentage to signify that the activity is X percent complete. The tendency is then to interpret the progress bar as an indication of being ahead or behind schedule. This interpretation depends on whether the progress bar crosses over the data date on the Gantt chart. Unfortunately, even though the progress bar crosses over the data date, the activity could very well be behind the planned schedule. For example, consider an activity with a target completion one day before the data date (see the thin bar in Exhibit 2). If the activity is reported at that time to be 90 percent complete and has both completed and remaining durations of four weeks, the progress bar superimposed on the activity would extend well beyond the data date. It would appear that the activity is ahead of schedule. In reality it is behind schedule since the target finish is before the data date. The converse is also possible.

img

The preceding assumes that progress is indicated as a percent of the activity's duration. If, on the other hand, progress is displayed on the basis of days completed and remaining duration, the picture does not change much. All in-progress activities appear to be exactly on schedule. The progress bar is plotted up to the data date, as in Exhibit 3.

Earned Value Plays Its Part. The solution involves the use of earned value information to display not only the progress of activities, but also the status of the schedule compared to the plan or target in a quantified way. There are many reports and graphics available to display earned value-based schedule variances. However, none seem to communicate both cost-based schedule performance and traditional time-based schedule information in a concise manner on the same page. The method I use results in a simple Gantt chart display format that communicates the full story.

In order to better understand the meaning of “percent complete” with regard to activity progress, let us first look at what it means to say “this activity is X percent complete.” It is generally accepted practice to let percent complete represent the portion of physical work that has been completed up to the time in question, the data date. The trick is to obtain an objective measure of the physical work that has been completed. Most projects are made up of various activities of different kinds of work. Therefore, some uniform means of calculating percent complete has to be used if project reporting is to occur at any level above the most detailed breakdown of the work. This is because one cannot simply combine cubic yards of concrete placed with linear feet of piping installed and with number of electrical connections made, and so on.

To accomplish this, the project management profession has adopted the use of earned value. With this method, a uniform unit of measure is employed (usually workhours or dollars) to allow combining measurements of different kinds of work. The earned value of an activity is simply the budgeted value of the portion of the work that has been completed, expressed in hours or dollars. In the case of physical work, this is determined by measuring the installed quantities, be they cubic yards of concrete, feet of pipe, lines of software code, or number of tests performed. In the case of mental work, where progress is not as visually apparent, earning rules should be established to determine the percent complete of an activity.

Most scheduling software has a feature that displays progress in a contrasting color or pattern over the duration of an activity. This progress bar is often quantified as a percentage to signify that the activity is X percent complete. The tendency is then to interpret the progress bar as an indication of being ahead or behind schedule. But is this accurate?

Exhibit 1. Most scheduling software has a feature that displays progress in a contrasting color or pattern over the duration of an activity. This progress bar is often quantified as a percentage to signify that the activity is X percent complete. The tendency is then to interpret the progress bar as an indication of being ahead or behind schedule. But is this accurate?

Here's an activity with a target completion one day before the data date (the thin bar). The activity is reported to be 90 percent complete and has both completed and remaining durations of four weeks. In this case, the progress bar superimposed on the activity would extend well beyond the data date. It would appear that the activity is ahead of schedule. In reality it is behind schedule since the target finish is before the data date

Exhibit 2. Here's an activity with a target completion one day before the data date (the thin bar). The activity is reported to be 90 percent complete and has both completed and remaining durations of four weeks. In this case, the progress bar superimposed on the activity would extend well beyond the data date. It would appear that the activity is ahead of schedule. In reality it is behind schedule since the target finish is before the data date.

If progress is displayed on the basis of days completed and remaining duration, all in-progress activities appear to be exactly on schedule

Exhibit 3. If progress is displayed on the basis of days completed and remaining duration, all in-progress activities appear to be exactly on schedule.

An example of percentage milestones for a design drawing activity might be 10 percent for the initial calculations, 20 percent for the first draft sketch, 40 percent for the first CAD output, 50 percent for the first internal review, 65 percent for the second CAD output, 75 percent for submittal to the client, 85 percent for incorporating client's comments and 100 percent for final issue for construction. As these milestones are reached, the appropriate percent complete from these earning rules are then applied to the activity.

To establish the basis for earned value, the project is broken down by some useful scheme into more and more detailed levels. This becomes the work breakdown structure. The purpose of the WBS is to provide the framework for collecting detailed information into higher levels for review and analysis by individuals at the various levels of the project organization. The next step is to allocate the entire project budget to the elements at the lowest level of detail. Expressed in workhours or dollars, these budgets provide the basis for the earned value calculations.

The use of this earned value technique allows the project manager and the project team to look at any level of detail within the WBS of the project. They can then obtain a meaningful measure of the percent complete for each component at the level under review. With proper WBS design, each element in the WBS will have a specific individual responsible for that element. In this way, any questions on progress and cost performance can be answered by someone who is intimately familiar with the work.

img

Exhibit 4. The activity started some time ago but shows only a small amount of progress. It appears to be in serious trouble, requiring all kinds of remedies and work-arounds. However, the activity was started way ahead of schedule; note the thin target bar. It simply wasn't worked very intensely, which may not be a problem.

img

Exhibit 5. If we were to superimpose on the normal Gantt chart in Exhibit 4 another bar that reflects the “aheadedness” or “behindedness,” it might look like this. Here the red band shows that the activity is actually ahead of schedule to some degree.

img

Exhibit 6.If we apply the technique shown in Exhibit 5 to all the activities, a project schedule might look like this. The red background gives a vivid display of the general status of the project schedule. And it is based on the objective application of earned value: the objective quantification of an activity's percent complete.

The Process. Back to our problem of activity percent complete display. We know now that we can get a reasonably accurate measure of the percent complete of an activity. We apply that as input during the update cycle of our project and plot the progress as usual. The result is similar to Exhibit 1. However, what we desire is a simple graphic indication of how far behind or ahead of schedule the activity is. We understand now that the traditional display doesn't tell us, it only appears that it does. Conversely to Exhibit 1, which appears to be a favorable situation, we can have the condition shown in Exhibit 4. The activity started some time ago but shows only a small amount of progress. It appears to be in serious trouble, requiring all kinds of remedies and workarounds. However, as Exhibit 4 shows, the activity was started way ahead of schedule. It simply wasn't worked very intensely and shows little progress. Perhaps, while the project team saw an opportunity to do some work on the activity, there was no reason to bring all the planned resources to bear on it since it has plenty of time to be completed.

If we were to superimpose on the normal Gantt chart in Exhibit 4 another bar that reflects the “aheadedness” or “behindedness,” it might look like Exhibit 5. Here the red band shows that the activity is actually ahead of schedule to some degree. If we then apply this technique to all the activities, a project schedule might look like Exhibit 6. Note that the red background gives a vivid display of the general status of the project schedule. And it is based on the objective application of earned value. Remember that earned value is the objective quantification of an activity's percent complete. There is no guesswork because the progress is either measured by physical quantities or established from predefined earning rules. The mood of the day does not come into play when assessing progress.

The specific technique described here involves the calculation of the Schedule Performance Index (SPI) and applying that to a formula which determines the dates for the ends of the red curtain for each activity. The basic formula for the end-date of what we'll call the “Progress Curtain” is the “Curtain Finish” (CF). It is calculated as:

CF = DD + (SPI - 1) * CD

where:

DD = Data Date

CD = Completed Duration, in workdays

SPI = Schedule Performance Index

What this does is calculate a point in time relative to the data date that is equal to the number of days completed multiplied by the amount that the SPI is off expected performance. Thus, if the SPI is less than 1.00, the multiplier becomes negative and the curtain finish date will fall to the left of the data date. This indicates being behind schedule and is in line with the concept of an SPI less than 1.00. The smaller the SPI, the further to the left of the data date will be the curtain finish. The converse is also true. An SPI greater than 1.00 will add days to the data date. The larger the SPI, the greater will be the curtain finish. The beauty of this is that it is based on sound earned value principles and displays an objective picture of the project status.

I chose the completed duration as a basis for calculating the curtain finish so that the CF will not fall ahead of the actual start date of the activity. This leads to the condition where two activities with identical SPIs may have different curtain finishes. An alternative could be to use an arbitrary number of days (e.g., 30, or one update cycle length) for all activities. This is left to the reader to experiment with.

The result is shown in Exhibit 6. The red curtain displays the schedule status against the data date. This permits a quick-glance assessment of the project schedule status that is based on an objective determination, using earned value techniques. It correlates with the earned value reports that might be published. And, most important, it presents the true progress picture of the activities.

Special Conditions. In practice, a few formula overrides are helpful. It often turns out that a few activities wind up with a huge SPI, in the order of 10 to 25 or even greater. This is the result of special cases, such as activities that are started very much ahead of schedule. In such cases it is graphically best to limit the curtain finish to some agreed-to amount. I have found it convenient to engage this formula override when the SPI exceeds a value of 2.00. In my experience, the vast majority of SPI values tend to fall between 2.00 and .30. See Activity C in Exhibit 6 for an example of an activity with an overridden SPI.

Another case for providing a formula override is when an activity is reported started on a data date prior to the planned or target start. The target start determines the calculation of the planned progress. This is known as the Budgeted Cost of Work Scheduled (BCWS) in earned value applications. I have found it useful to set the SPI to a default value of 1.50 in these cases. This overcomes the problem of a computer program not calculating an SPI when there is no BCWS. Remember that SPI equals Budgeted Cost of Work Performed (BCWP or Earned Value) divided by the BCWS. If the BCWS is 0, then the calculation of the SPI would default to 0. Thus I set the SPI at 1.50 on my projects. This seems a reasonable assumption for an undeterminable SPI when the activity did start ahead of schedule (see Activity J in Exhibit 6). The full logic applied to these calculations are shown in the sidebar.

Putting This Method Into Practice

This summarizes the calculations and overrides used to achieve the results presented in the article. The use of this process depends on the capabilities of the scheduling software employed. Primary features required are the ability to calculate user-defined variables, to calculate differences between dates, and to plot custom bars dependent on the custom-defined variables.

1. Clear all previous Curtain Dates (PRST and PRFN)

Select all activities with an actual start date.

This is to clear out old dates and start with a clean slate.

The terms PRST and PRFN are either a user-defined variable or activity code that can be used to plot schedule bars. They are date-type values that define the start and finish of the Progress Curtain.

PRST =″ ″

PRFN =″ ″

2. Set the Curtain Start Date (PRST) to Project Start Date

Select all activities with an actual start date.

This is to provide for each activity a new curtain beginning.

PRST =SD

3. Calculate SPI for in-progress activities

Select all activities with an actual start date.

The SPI is a variable that the user can calculate and use in further calculations (see 4 below). Note: If you use a computer calculated SPI, you must also be able to override that calculation (see 4 and 5 below).

SPI = BCWP / BCWS

4. Set SPI to 1.5 for activities with future Target Start Dates

Select all activities with target early start greater than the data date.

This is to override a default calculation to 0 (zero) when the target start date is in the future.

SPI = 1.5

5. Set Curtain Finish for activities with SPI <2.0

Select all activities with an actual start date.

This is the primary calculation (A, B and C are temporary variables).

A = DD – AS (Data Date minus Actual Start)
B = SPI – 1  
C=A*B (Deviation From Data Date)
PRFN = DD + C (Curtain Finish)

6. Set Curtain Finish for activities with SPI > 2.0

Select all activities with an actual start date.

This is to override an unusually large SPI.

A=DD–AS (Data Date minus Actual Start)
PRFN = DD + A  

For reference, the SPI and Cost Performance Index (CPI) are shown in the table in Exhibit 6. This provides the added benefit of cost performance information in the same report. For example, Activity F in Exhibit 6 shows a wonderful schedule condition. However, the CPI is a dismal .78. The responsible manager for that activity should use this information to probe further and decide if the schedule performance is worth the cost premium.

BY ADDING A THIRD ELEMENT (“aheadedness” or “behindedness” in addition to the current bar and progress bar), the project manager and other interested parties can obtain, at a glance, an objective display of the progress condition of the project, or any portion of it as defined by the WBS. It is earned value that makes this possible by providing a unified unit of measure: workhours or dollars. While I have presented my personal preference of display and formulas, other display formats and adjustments to the formulas are possible. However, these are left for the intrepid reader to pursue. ∎

Note: Although slightly changed for PM Network, this paper was presented at PMI ’96 in Boston, Mass., USA.

 

Tammo T.Wilkens, PMP, FIE., has practiced project management since the late ’70s. As manager of engineering planning at Ebasco Services in New York, he helped develop Ebasco's earned value-based project control system. He is project control manager for Los Angeles Metropolitan Transportation Authority's Red Line Subway project.

This material has been reproduced with the permission of the copyright owner. Unauthorized reproduction of this material is strictly prohibited. For permission to reproduce this material, please contact PMI.

PM Network • August 1997

Advertisement

Advertisement

Related Content

Advertisement