CPI and SPI are the two numbers most people mean when they say "earned value." They are useful because they are unitless ratios: you can compare a pipe item to an excavation item to a whole project without converting anything. They are dangerous for exactly the same reason, because a ratio hides the size of what it describes.
The formulas
SPI = EV ÷ PV (Schedule Performance Index)
Both share the same numerator, Earned Value: the budgeted value of work actually completed. CPI divides it by what that work cost. SPI divides it by what you planned to have completed by now.
| Value | CPI means | SPI means |
|---|---|---|
| 1.00 | Earning exactly a dollar of budgeted work per dollar spent | Progressing exactly at the planned rate |
| Above 1.00 | Under budget on the work performed | Ahead of the planned rate |
| Below 1.00 | Over budget on the work performed | Behind the planned rate |
A useful mental translation for CPI: a CPI of 0.87 means you are getting 87 cents of budgeted work for every dollar you spend. To finish the remaining work at that rate, you will need about 1 ÷ 0.87 = 1.15 dollars for every budgeted dollar left.
Two worked examples
Example 1: an excavation item in trouble
40,000 CY at a $6.50 per CY budget, BAC of $260,000, planned at 4,000 CY per week over 10 weeks. At the end of week 4: 13,000 CY installed, $97,000 spent.
EV = 13,000 CY × $6.50 = $84,500
AC = $97,000
CPI = 84,500 ÷ 97,000 = 0.87
SPI = 84,500 ÷ 104,000 = 0.81
Over cost and behind schedule. The unit cost check confirms it: $97,000 ÷ 13,000 CY = $7.46 per CY actual against a $6.50 budget, and 7.46 ÷ 6.50 = 1.15, which is exactly 1 ÷ CPI. When those two agree, your data is internally consistent.
Example 2: a storm item running well
8,000 LF of 12-inch storm at a $92 per LF budget, BAC of $736,000, planned at 667 LF per week over 12 weeks. At the end of week 6: 4,600 LF installed, $402,000 spent.
EV = 4,600 LF × $92 = $423,200
AC = $402,000
CPI = 423,200 ÷ 402,000 = 1.05
SPI = 423,200 ÷ 368,000 = 1.15
Under budget and ahead of schedule. Actual unit cost is $402,000 ÷ 4,600 = $87.39 per LF against $92 budgeted.
The four combinations and what to do
| Pattern | Likely meaning | Action |
|---|---|---|
| CPI low, SPI low | Genuine production problem, or a bad estimate, or bad data | Verify data first, then check production rate against the bid rate, then consider notice |
| CPI low, SPI high | Buying schedule with money: overtime, extra crews, premium equipment | Confirm it is a deliberate decision, and that the acceleration is recoverable or worth it |
| CPI high, SPI low | Efficient but under-resourced, or work is blocked by others | Look for constraints: access, materials, predecessors, inspection waits. Adding crew may be cheap |
| CPI high, SPI high | Genuinely good, or the budget was padded, or quantities are overstated | Verify the quantities before celebrating. Then check whether the estimate should change for the next bid |
Where the indices mislead you
- SPI drifts toward 1.0 near the end. As a project approaches completion, EV and PV both converge on BAC, so SPI approaches 1.0 mathematically even on a job that finishes late. Late in a job, trust the schedule, not the SPI.
- SPI says nothing about the critical path. You can be ahead on 90 percent of the work and still finish late because the one item that matters is behind. SPI is a volume measure, not a logic measure. Use it alongside the schedule, never instead of it.
- Ratios hide magnitude. An item at CPI 0.60 with a $30,000 budget is a rounding error. The same index on a $2M item is the job. Always look at the dollar variance next to the ratio.
- Timing mismatch inflates CPI. If quantities are captured daily but invoices arrive three weeks later, EV is current and AC is stale, so CPI looks great until the invoices land. Compare like periods or accrue committed cost.
- Aggregation hides the sick item. Roll up far enough and everything looks average. Read the indices at the level you can actually act on, which for most civil work is the cost code.
An index is a diagnosis, not a verdict. The value of CPI and SPI is that they tell you which item to go look at, two weeks before the cost report would have.
How often should you look?
The traditional answer is monthly, because that is how often the accounting close produces actual cost. That cadence is a legacy of how slowly data used to move, not a property of the method. On a 14-week utility job, a monthly index gives you three data points, and the first one arrives when a quarter of the job is gone.
If installed quantities and hours are captured daily, weekly indices are practical and far more actionable, because the correction window is still open. The limiting factor is almost always how quickly actual cost becomes visible, which is why field-captured labor and equipment time matters so much: it is the largest component of cost and the one you can make current immediately.
Where to go next
If any of the terms above were unfamiliar, start with what is earned value in construction. To apply the indices to yardage-based work, see earned value for dirt work. To make weekly indices realistic, see the field data capture guide.
PM Axsus computes these indices from field-captured quantities and hours rather than from month-end reconstruction, which is what makes a weekly cadence workable. See earned value management.