Which EAC to Report When the Variance Is Not Uniform
At the month-9 review of a 24-month, $12.0M EPC contract, the control account shows PV = $4.80M, EV = $4.32M, AC = $4.68M. The project manager wants to report EAC = BAC/CPI and tell the client the job is '10% behind schedule.' You know two things he does not. First, $400K of the AC is a one-time rework event caused by a vendor's incorrect weld map; it is closed and not expected to recur. Second, the critical path runs through a permitting activity that has absorbed 5 weeks of delay unrelated to production, while the remaining 15 months of work are repetitive and well understood.
1. Compute CPI, SPI, and the BAC/CPI estimate at completion.
Read the worked answer
CPI = EV / AC = 4.32 / 4.68 = 0.923
SPI = EV / PV = 4.32 / 4.80 = 0.900
EAC (CPI-based) = AC + (BAC - EV) / CPI = 4.68 + (12.00 - 4.32) / 0.923 = 4.68 + 8.32 = $13.00M
VAC = BAC - EAC = -$1.00M
Note the arithmetic is only as good as the assumption behind it: this EAC says the entire remaining scope will be executed at the blended efficiency experienced to date. That assumption is what you have to test, not just the division.
2. Recompute using the known one-time event and decide which forecast you present.
Read the worked answer
Strip the closed, non-recurring $400K of rework from AC:
- Adjusted AC = 4.68 - 0.40 = $4.28M
- Adjusted CPI = 4.32 / 4.28 = 1.009
- Adjusted EAC = 4.68 + 7.68 / 1.009 = 4.68 + 7.61 = $12.29M
So the defensible answer is a range of roughly $12.3M to $13.0M, not a single figure.
Which one you lead with depends on evidence, and you should say so:
- If the rework was genuinely a one-off vendor error with a corrected weld map and no other systemic driver, the adjusted figure is more realistic and the $13.0M over-penalizes the forecast.
- But removing a variance requires proof, not assertion. Ask: was the root cause corrected? Did it change the weld mix or inspection rate going forward? If the same vendor or the same QC gap persists, the anomaly is systemic and the unadjusted CPI wins.
Best practice: present $13.0M as the upper bound, a bottom-up ETC for the remaining 15 months of repetitive work as the lower bound, and state the assumption that separates them. A single top-down number with a known anomaly buried in it will be challenged and will lose credibility.
3. Critique the planned statement that the job is '10% behind schedule.'
Read the worked answer
SPI = 0.900 does not mean 10% behind on the calendar, and saying so to a client is a claim you cannot support.
Why:
- SPI is cost-weighted, not time-weighted. It compares earned value to planned value in dollars, not days. A dollar-weighted 10% shortfall can correspond to any duration slip, large or small.
- SPI converges to 1.0 at completion regardless of how late the job finishes, so it is a poor completion-date forecaster at any point in the curve.
- It mixes float-bearing and critical work. Work slipping on activities with 5 days of float does not move the finish date at all.
- The real driver is visible elsewhere: a 5-week permitting delay on the critical path. That is the schedule fact that matters, and SPI will not surface it.
What to do instead: calculate earned schedule (time-based SV(t) and SPI(t)) to express slip in duration terms, and run a schedule risk analysis on the critical path so the forecast completion date carries a confidence level. Report cost and schedule separately - CPI 0.92 and a 5-week critical-path delay are two different problems with two different recovery plans.