Knowing the Score
Two of the best known indicators of project performance are
probably the SPI, or Schedule Performance Index, and the CPI or Cost
Performance Index. These two indicators have been popularized due to their description
in the PMBOK®
and other project management sources. These indicators are key sources of
information on two of the most important project performance areas: schedule
performance and cost performance. There are many more areas of project
performance that the project manager must monitor to assess project health but
when you are reporting progress to your sponsor, they are unlikely to be
enthused over the quality of the software produced if the project is 50% behind
schedule and 50% over budget! The point here is that these 2 indicators are more basic than the others.
These indicators are also valuable sources of information
because of the data available to create them. First of all, let's take a moment
to review how these indicators are calculated. Those of you already familiar
with the calculation can skip to the next paragraph. The Schedule Performance Index
is calculated by comparing the amount of scheduled work completed with the
amount you planned to complete: Budgeted Cost of Work Performed/Budgeted Cost
of Work Scheduled. An index greater than 1.0 indicates a project ahead of
schedule, an index of less than 1.0 indicates one that is behind schedule. CPI
is calculated by comparing the budget for the work performed against the actual
cost of that work: Budgeted Cost of Work Performed/Actual Cost of Work
Performed. A CPI of greater than 1.0 means the project is under budget and a
CPI of less than one means it is over budget.
The data from which these indicators are calculated is
available from your scheduling tool. The tool is used to create the project's
Work Breakdown Structure (WBS), project activities are assigned to the work
packages and activities are scheduled with a start and finish date. The
schedule is updated with project information as work is completed. Work that is
finished is marked as 100% complete. Updates beyond that will depend on the
tracking approach you choose.
The simplest approach to partially completed work is to
ignore it. The activity is either 100% complete or 0% complete. This is the
easiest method to administer because it does not require you to make difficult
assessments. This approach may be practicable where work is broken down into
activities that take a day to complete at maximum. In that case you won't be
missing a great deal of work when only including activities that are 100%
complete. If the activity in question is a 4 hour activity discounting it, even
when it is 90% complete only skews the total by a little more than 3.5 hours -
not much in a project with a total of thousands of man hours of work. Skewing
will increase as the number of partially completed activities mounts, or if
those activities account for days or weeks worth of work.
Accounting for partially completed work only becomes an
issue when that work has a planned or forecast completion date before the reporting
date. Activities to be completed after this date will not be included in the
"Budgeted Cost of Work Scheduled" category. Work that has fallen
behind schedule is not always 100% behind and reporting the amount of work
completed on an activity that is behind schedule increases the accuracy of your
reporting.
Where the size and complexity of the project does not allow
for defining activities of under a day in duration, you should consider
reporting partially completed tasks. Knowing the amount of work that has been
completed, compared to the total amount to be done, is tricky. The easiest way
is to have the owner of the task report progress to you in the same way the
tracking tool represents it, as percentage complete. Unfortunately, this will
also yield the least accurate estimate. The owner of the task will want to
present the most optimistic case. You know the scenario, you ask for an
estimate and get a "90%" complete. You ask at the next reporting
period and you get a "95%" complete. The approach that will yield the
most accurate data is one that measures a concrete, verifiable deliverable.
This could be the number of lines of code, the number of tests executed, the
number of feet of flooring installed, or some other metric but it should be
something that is tangible. You should verify that the claimed work has been
done. Have the lines of code been checked into the library? Have they been
counted? Do they meet the criteria established for code? Has the code been unit
tested? etc. Code that has not been checked into the library, tests that have
not been reported, or flooring that is not installed don't get counted. The
method you choose should one that meets the two criteria: verifiable and
measurable, and should add a minimum of overhead to your project to support.
The lengths you go to make your reporting accurate should
also be commensurate with your stakeholders need for accuracy. If the
stakeholders express a need for accuracy to within +/- 1%, they shouldn't
object to your requirement that the extra time spent managing partially
completed code in the library, or tracking tests as they are completed. On the
other hand, if the need for accuracy is within 10% the extra care in validation
is not warranted.
The work to include in your budgeted cost of work scheduled
will include any work with a forecast, or planned completion date falling on or
before the reporting date. It should also include work that was planned to be
partially complete by the reporting date. There is no need to include work that
should be partially completed, unless you plan to add partially completed work
to the Budgeted Cost of Work Performed. For example, where you are 1 week into
an activity scheduled to take 2 weeks, your BCWS would include the 1 week. If
the activity is indeed 50% complete then the BCSP would be incremented by 1
week to offset the 1 week added to the BCWS. A level of completion above or
below 50% would shift the SPI.
The decision on how to handle partially completed work
should be made in concert with the decision on how to handle scheduled work. If
you plan to report on work that is partially completed, before its scheduled
completion date, include the portion of that work that you had planned to
complete by the reporting date. If you do not include partially completed work
due on a future date in your BCWP, then there is no need to include the scheduled
portion of that work in the BCWS (or indeed any work whose due date is in the
future).
The Actual Cost of Work Performed and Budgeted Cost of Work
Performed should be approached in the same way. Don't include partially
completed work, due future, in the ACWP category if you won't be including
partially completed work (due in the future) in the BCWP category.
It is important to set expectations around the accuracy of
your reporting. Educate your stakeholders in the mysteries of performance
reporting and then inform them of the source of your information, how it is
collected and extracted, and how you calculate the various indicators. Don't
neglect to provide your stakeholders with an indication of the accuracy of your
indicators, and don't forget to discount them yourself when assessing your
project progress. Your project progress reporting will become much more
transparent when your audience understands what you are telling them, how you
collected the information, and its accuracy.
|