Earned value management in plain English: CPI, SPI and what to do about them
Turn planned value, earned value and actual cost into useful decisions without confusing budget performance with cash flow or time.
By Dr. Hassan Khames EliwaLast updated

Spending less does not necessarily mean doing well
A project can be under budget this month because it has accomplished very little. It can also spend ahead of plan while completing valuable work early. Earned value management separates those situations by comparing the budgeted value of completed work with both the plan and the actual cost.
The important word is earned. You earn budget by completing defined work, not by receiving an invoice, making a payment or announcing that an activity is nearly finished.
Used properly, EVM is an early-warning system. Used carelessly, it produces precise-looking ratios from unreliable progress claims. Start with the work measurement, then calculate the indicators.
The three numbers you need
The terminology comes from established earned value practice, including the US Department of Energy's EVM guidance.
- Planned value, PV: the approved budget for work scheduled to be complete by the status date.
- Earned value, EV: the approved budget for work actually completed by that date.
- Actual cost, AC: the cost incurred for that completed work.
All three must describe the same scope and reporting cutoff. Budget at completion, or BAC, is the total approved performance budget for the work being measured.
Suppose a contractor has a $600,000 work package. At the reporting date, the baseline says $240,000 of work should be complete. Measured progress earns $210,000, while accrued actual cost is $250,000.
| Measure | Amount | Meaning |
|---|---|---|
| BAC | $600,000 | Total performance budget |
| PV | $240,000 | Budgeted work scheduled |
| EV | $210,000 | Budgeted work completed |
| AC | $250,000 | Cost incurred |
These are not four competing estimates of the same thing. They answer different questions.
CPI: are we buying progress efficiently?
Cost performance index is earned value divided by actual cost.
CPI = EV / AC
For this package, CPI = $210,000 / $250,000 = 0.84. Each dollar incurred has earned 84 cents of budgeted work.
A CPI below one signals unfavorable cost performance. Above one is favorable against the budget basis. Neither result proves that the original estimate was realistic, so investigate the underlying quantities and rates.
Cost variance provides the dollar view:
CV = EV − AC = −$40,000
The negative variance means completed work has cost $40,000 more than its budgeted value. It does not mean the final overrun will necessarily be $40,000.
Run these inputs through the earned value calculator to keep the arithmetic consistent while you investigate the causes.
SPI: are we producing work at the planned rate?
Schedule performance index is earned value divided by planned value.
SPI = EV / PV
Here, SPI = $210,000 / $240,000 = 0.875. The package has earned 87.5% of the budgeted value scheduled by this date.
Schedule variance is:
SV = EV − PV = −$30,000
This is a budget-valued measure of schedule performance, not a delay measured in dollars that can be converted directly into days.
A project might complete expensive noncritical work early while a low-cost critical activity slips. Its SPI can look acceptable even while the completion date is deteriorating. Read the critical path schedule alongside EVM.
Traditional SPI also tends toward one at completion because all planned work eventually becomes earned work. A late finish does not disappear just because the final SPI reaches one.
Forecasting the likely final cost
A simple estimate at completion assumes the current cost efficiency continues:
EAC = BAC / CPI
For the example, $600,000 / 0.84 = $714,285.71.
The corresponding estimate to complete is $714,285.71 − $250,000 = $464,285.71. Variance at completion is $600,000 − $714,285.71 = −$114,285.71.
That is a useful warning, not a contractual prediction. The formula assumes the remaining scope experiences the same cost efficiency as the completed scope.
If the overrun came from a genuinely nonrecurring event, a detailed remaining-work estimate may be better. If labor productivity is still poor and overtime is increasing, assuming an immediate return to budget efficiency needs strong evidence.
Prepare a bottom-up estimate for the remaining quantities, labor hours, equipment and commitments. Reconcile it to the statistical forecast rather than silently replacing whichever figure is less comfortable.
How hard would recovery have to be?
The to-complete performance index asks what efficiency the remaining work must achieve to meet a target.
To finish within BAC:
TCPI = (BAC − EV) / (BAC − AC)
The calculation is $390,000 / $350,000 = 1.1143, rounded. The remaining work must earn about $1.11 of budget for every dollar incurred, compared with $0.84 achieved so far.
That is a substantial change in performance. It may be possible, but it needs a credible explanation: different work mix, resolved access constraints, improved crew balance or a corrected procurement problem.
“Recover next month” is not an explanation. Name the operational change, its owner, its implementation date and the evidence that will show whether it worked.
Diagnose before prescribing a recovery plan
Break the variance into manageable work packages. A project-wide CPI can conceal one troubled subcontract, an estimating error or a badly sequenced operation.
Ask practical questions:
- Are installed quantities measured consistently?
- Do actual costs include accruals for work already performed?
- Are material costs recorded before related installation earns value?
- Has unauthorized scope entered the field?
- Are rework and overtime visible rather than spread across unrelated codes?
- Is productivity deteriorating, or is the budget simply inadequate?
Material purchases deserve special attention. Paying for equipment early can distort comparisons if the earning rules recognize it only at installation. Establish a defensible method for procurement milestones before reporting begins.
Do not change measurement rules midway through the job merely to improve the ratios.
Make earned value believable
Choose progress methods suited to the work. Installed quantities work well for repetitive measurable production. Weighted milestones can suit engineering or procurement. Short tasks may use completion-based rules.
Subjective percentages are weaker, especially when unfinished activities sit at “almost complete” for several reporting cycles.
The baseline also needs an approved scope, time-phased budget and controlled changes. Without those, unfavorable performance can be hidden by moving budget or dates after the event.
For uncertain remaining durations, use the PERT three-point estimating calculator to structure optimistic, most likely and pessimistic estimates. Those duration estimates support schedule analysis; they do not replace earned value or establish a completion probability by themselves.
Turn the report into a decision
A useful monthly review ends with actions, not just ratios. Record the cause, the proposed intervention and the forecast consequence.
For example, a crew productivity problem might lead to resequencing access, changing material delivery arrangements or reducing interference between trades. Assign someone to test whether installed output per labor hour improves.
Keep an unchanged baseline for performance measurement unless an approved change justifies revision. Maintain a separate current forecast that reflects what the team now expects.
The baseline tells you what was promised. The forecast tells you what is likely. EVM helps expose the gap, but managing the work is what closes it.
Calculators used in this guide
Frequently asked questions
Does a CPI below one mean the project is losing money?
It means actual cost exceeds the budgeted value of completed work. Profit also depends on contract revenue, approved changes, overhead treatment and the final cost, so CPI alone does not establish profitability.
Can SPI tell me how many days the project is late?
No. Traditional SPI compares budgeted work values, not calendar days. Use the current logic-linked schedule and critical path to assess the forecast completion date.
Should actual cost mean invoices paid?
Usually not. Actual cost should reflect costs incurred for the measured work, including suitable accruals, so invoice and payment timing do not distort the comparison.
What if PV or AC is zero?
SPI is undefined when PV is zero, and CPI is undefined when AC is zero. Investigate the reporting period and data rather than forcing a ratio.
Should I reset the baseline after an overrun?
Not simply to remove an unfavorable variance. Apply the project's formal change-control and any authorized rebaselining procedures, preserving an explanation of historical performance.
Sources and standards
About the author
Dr. Hassan Khames Eliwa holds a PhD in Construction & Built Environment and has more than 20 years of international project controls experience. He is Director of Eliwa Project Controls Limited in Auckland and founder of PMMilestone.org and PMMilestone.com.
Buy me a coffee