Earned value, and choosing a forecast
Comparing budget to spend cannot tell you whether a project is in trouble. Earned value can — provided you understand which assumption your forecast rests on.
The idea in one line
Earned value is the budgeted cost of the work actually done — what the completed work was supposed to cost.
That single number turns a spend figure into information. Having spent 60% of a budget is meaningless alone: excellent if 80% of the work is complete, alarming if 40% is. Comparing what the finished work was worth against what it actually cost is the whole technique.
The two indices
CPI = EV ÷ AC — how much planned work each unit of money bought. Below
1.0 means you are paying more than budgeted for what you are getting.
SPI = EV ÷ PV — how much of the planned work is done. Below 1.0 means
less has been completed than the plan expected by now.
Both need a non-zero denominator. Early in a project, before anything is spent or scheduled, they are genuinely undefined rather than zero — and a tool that reports 0 or infinity there is giving you a number where it should be giving you a blank.
Four forecasts, four different beliefs
EAC is where judgement enters, and the spread between methods is large — commonly 40% or more on the same project.
BAC ÷ CPI assumes the overrun is systemic: whatever made the work so far cost more will keep doing so. Usually the safest default, because overruns are more often about estimating and rates than about one bad month.
AC + (BAC − EV) assumes the variance is behind you. Defensible when you can name the cause and it has genuinely ended. Wishful when you cannot.
The CPI×SPI version assumes lateness costs money to recover — overtime, extra people. Often realistic on a deadline-driven project, and nearly always the most alarming number.
A tool that shows one of these and calls it "the forecast" has made a judgement about your project and hidden it from you.
SPI breaks down at the end
Schedule performance index converges to 1.0 as a project completes, however late it is. Once all the work is earned, EV equals PV by definition, and the index cheerfully reports 1.0 while you are three months over.
SPI is a cost-based proxy for schedule. It is most useful in the middle of a project and least useful at the end — which is precisely when people reach for it. For schedule, use the critical path.
TCPI is the reality check
TCPI = (BAC − EV) ÷ (BAC − AC) asks what efficiency the remaining
work must achieve to finish on budget.
Take it seriously. A team that has been running at 0.7 will not suddenly run at 1.5. Once TCPI climbs much past 1.1, the budget is gone and the honest conversation is about scope or funding rather than about trying harder.
What this tool will not do
It calculates from the figures you enter and nothing else. It cannot tell you whether those figures are honest, and a precise-looking output built on rough inputs is still rough.