Earned Value Indices

Three inputs, everything else derived

Earned value management needs exactly three measurements: what the plan said should be done by now (planned value), what has actually been completed valued at plan rates (earned value), and what has been spent to achieve it (actual cost). Every index and forecast follows from those three.

The cost performance index is the ratio of value earned to money spent. At 0.857 it says that every pound spent has produced 85.7 pence of budgeted work — and projecting that across the remaining budget is where the forecast at completion comes from.

Why SPI converges to 1.0 and becomes useless

The schedule performance index is measured in currency rather than in time, and at completion earned value equals planned value by definition. A project delivered a year late therefore finishes with SPI exactly 1.0.

This is not a flaw to be worked around; it is a property to be understood. SPI is informative in the middle of a project and meaningless at the end, and time performance belongs to the schedule network rather than to an index denominated in money.

TCPI is the reality check

The to-complete performance index asks what efficiency the remaining work must achieve for the project still to land on budget. It is the most useful number EVM produces, because it converts optimism into a testable claim.

If CPI has been 0.86 all project and TCPI is 1.12, the plan is asking you to believe the team will become thirty percent more efficient than it has ever been. That is occasionally true — after a genuine change in method — and usually not.

Every index depends on one soft number

Earned value is an assessment of how much is genuinely complete, and it is the input most easily corrupted. Optimistic percent-complete reporting produces a healthy CPI on a failing project right up until the final month, and this is the standard failure mode of EVM in practice.

The defence is objective measures — units installed, milestones passed, tests witnessed — rather than a judgement of how far along something feels.

The forecast formulas are three different bets

A forecast at completion needs an assumption about the remaining work, and the standard forms each encode a different one. Dividing the budget by the cost index assumes performance to date CONTINUES — appropriate where the overrun is systemic, a wrong rate or a wrong quantity basis.

Adding the remaining budgeted work to what has been spent assumes the overrun was a one-off and the rest will run to plan. That is the optimistic form, and it is the right one only where the cause has been identified and removed.

The third form divides the remaining work by both indices together, assuming cost and schedule pressure both persist. It is the pessimistic form and is usually the closest on troubled projects, because schedule pressure is itself a cost driver — acceleration, overtime and out-of-sequence work all arrive together.

None of these is more correct than the others as arithmetic. Choosing between them is a judgement about the CAUSE of the variance, and a report that quotes a single forecast without saying which assumption produced it has hidden the only interesting decision in the calculation.

Earned schedule is the fix for the index that stops working

The schedule index is a ratio of two money figures, and at completion the earned value necessarily equals the planned value, so the index returns to one however late the project is. It reports on time on the day a project finishes six months late.

Earned schedule replaces the ratio with a TIME one: find the date at which the planned value equalled the value actually earned today, and compare it with today. The result is expressed in time units, it does not converge to one, and it continues to report lateness right through the endgame.

It also removes a unit confusion the money-based index invites. Saying a project is running at 0.9 schedule performance is not a statement about weeks, and it cannot be converted into one without the plan's shape, whereas a time-based measure is directly a number of weeks behind. Where the schedule index is reported here, the page says which of the two it is.

The measurement method decides the answer more than the arithmetic does

Earned value needs a rule for how much credit an incomplete task has earned, and the choice of rule moves the result far more than any index formula. Crediting nothing until a task is complete produces a conservative, lumpy curve. Crediting half at start and half at completion smooths it and flatters early progress.

Percentage estimates are the most common and the least reliable, because they are a judgement made by the person being measured. Where the work has countable units — metres of pipe, square metres of cladding, numbers of doors — units-completed removes that judgement entirely and is the method worth insisting on.

Effort that has no deliverable of its own, such as supervision or site management, cannot be earned at all in the usual sense and is conventionally credited in line with elapsed time. That means part of every project's earned value is a measure of the calendar rather than of progress, and a project whose level-of-effort share is large will always look closer to plan than it is.

What the indices cannot see

Every index on this page is a ratio to a BASELINE, so all of them inherit whatever was wrong with the baseline. A plan that was never achievable produces variances that describe the plan rather than the performance, and no amount of index discipline recovers from it.

Quality is invisible to the method. Work that has been done, valued and paid for counts as earned whether or not it will have to be done again, so rework not yet discovered shows as progress. This is the systematic reason earned value looks healthy on projects that are storing up defects, and it is why the indices are read alongside inspection and test records rather than instead of them.

And the money is not the same as the cash. Earned value works in the values of the plan, while cash flow depends on payment terms, retention and certification timing — so a project can be performing well on every index and still be short of cash, or the reverse. The two reports answer different questions and the pages here answer the first one.