Few management techniques have been declared dead as often as earned value. Too bureaucratic, people say, too slow, too detached from how construction actually moves. Yet in most of the cases we see, earned value did not fail. It was fed opinions and asked to produce facts.
The idea is simple
Earned value compares three things: the value of the work planned to date, the value of the work actually done, and what that work cost. From them come schedule and cost variance, the schedule and cost performance indices, and an independent estimate of the cost at completion. Properly implemented, it answers the questions a sponsor cares about: are we doing the work we planned, are we getting the value we are paying for, and where will we land?
Where it stops telling the truth
- Percent complete reported by the people being measured, with no rules of credit
- Budgets and schedules built separately, so planned value describes neither
- Baselines re-planned so often that variance means only that the baseline moved
- Front-loaded measurement that earns value on mobilization and design and leaves the hard work unvalued
- Reports produced for compliance and never used to make a decision
Each of these produces indices that look healthy until late in the program, then collapse. The collapse is then blamed on the method.
An SPI of 0.98 means nothing if the 98 percent was estimated by the person whose performance it measures.
Measure what is installed
For construction, the most reliable foundation is physical quantity: linear meters of pipe tested, cubic yards placed, panels terminated, systems turned over. Define the rules of credit before work starts, with weighted milestones for steps that cannot be counted. Audit claimed progress against the field. When quantities drive earned value, the indices start to agree with what a superintendent would say on a site walk.
Watch the schedule index late in the program
SPI calculated in dollars converges to 1.0 as a program finishes, whether it finishes on time or not. Late in a program, it stops being a useful time measure. Earned schedule, which expresses performance in units of time, and a direct reading of the critical path are better guides to whether the date will hold.
Control the baseline
Every legitimate change should flow through a documented baseline change, with the reason recorded. Every illegitimate one, a re-plan to hide variance, should be refused. A baseline that moves without governance is not a baseline; it is a running commentary.
What the executive needs
The earned value report a sponsor actually needs fits on one page: are we ahead or behind, over or under, what is the realistic estimate at completion, and how does it compare with what the team is reporting? When the independent estimate and the team's forecast diverge, that gap is the most important line on the page, because it is where optimism is hiding.
For federal programs, alignment with the EIA-748 guidelines is often required. For private owners and lenders, the same principles apply with lighter governance. In both cases the test is the same: does the number tell you something you did not want to hear early enough to act on it?
