Percent Complete vs. Milestone Weights on a Long Engineering Package
A design-build firm tracks a $3.0M detailed-design work package over 18 months. The CAM has been using a simple percent-complete estimate based on 'drawings issued' and reports 45% complete at month 10 with $1.5M spent. The control account manager's engineers say the package is really about 30% done because the issued drawings are preliminary and will require substantial revision. Two other work packages in the same control account use 0/100 milestones and are performing on plan. The project manager likes the current numbers because they show SPI near 1.0. You are asked to validate the earned value approach before the next integrated program review.
1. Evaluate the appropriateness of the percent-complete technique here and recommend a measurement approach.
Read the worked answer
Percent-complete based on a subjective proxy ('drawings issued') is the weakest EV technique and is inappropriate for a long-duration, high-uncertainty package where the proxy does not correlate with real accomplishment. The engineers' 30% estimate versus the CAM's 45% is a red flag: a 15-point swing on a $3.0M package is $450K of potentially overstated earned value.
Recommendation:
- Replace with a weighted milestone or fixed-form (0/100, 50/50) technique tied to objective, verifiable completion criteria (e.g., design review sign-off, issued-for-construction package, customer approval).
- If milestones are too coarse, use units completed or apportioned effort tied to a measurable driver.
- Apply the change prospectively from the data date; do not retroactively rewrite history, but disclose the technique change and its effect.
- Investigate whether the 45% figure was influenced by pressure to show SPI ≈ 1.0 - that is a data-integrity concern.
2. If earned value is restated to 30%, what is the immediate effect on SPI, CPI, and the forecast, and what governance step is required?
Read the worked answer
At $1.5M AC and 30% of $3.0M budget:
- EV = $900K, so CPI = 900/1500 = 0.60 (was 0.90 at 45%).
- SPI depends on PV at month 10; if PV was ~$1.35M, SPI falls from ~1.0 to ~0.67.
- EAC (CPI-based) = BAC/CPI = 3.0/0.60 = $5.0M for this package, versus $3.33M before.
The restatement is a baseline/measurement integrity event, not a routine correction. Required governance:
- Document the technique change and rationale in the control account plan.
- Notify the customer/oversight as a variance and data-integrity disclosure.
- Re-examine the other 0/100 packages for similar proxy issues.
- Update the EAC and re-forecast at the program level - the PM must not simply revert to the flattering number.