Most schedule reviews on capital projects check the wrong things. They confirm the milestones look right and the bars line up with the narrative. They do not test whether the schedule can be relied on as a forecasting instrument. The DCMA 14-point assessment does.
The methodology was developed by the United States Defense Contract Management Agency to give schedule reviews a common, objective baseline. It has since become the de facto quality standard for capital project schedules worldwide. Applied properly, it tells you in an afternoon whether a schedule is structurally sound enough to trust. We have run it against a 30GB+ live schedule database spanning 120+ benchmarked programmes across nine sectors, so we also know how real schedules fail it.
Here is what each check actually tells you, not just the threshold.
The structural checks: can the schedule calculate?
1. Missing logic. No more than 5% of incomplete activities should lack a predecessor or successor. An activity without logic is a date typed into software. When the programme moves, that activity does not. On large mining schedules this is the most common failure, usually concentrated in procurement and commissioning sections built in a hurry.
2. Leads. The target is zero. A lead (negative lag) forces a successor to start before its predecessor finishes, by an offset the software cannot verify. Planners use leads to fake overlap. The honest fix is to split the activity.
3. Lags. No more than 5% of relationships should carry positive lag. Lag is invisible duration. It consumes time, carries no scope, cannot be statused and cannot be resourced. Concrete cure is a legitimate lag. Fifteen days of "coordination" is not.
4. Relationship types. At least 90% of relationships should be finish-to-start. Start-to-start and finish-to-finish pairs have their place, but a schedule dominated by them is usually hiding level-of-effort planning behind apparent detail.
5. Hard constraints. No more than 5% of activities should carry hard constraints. A mandatory finish date does not manage a deadline. It silences the schedule's ability to report that the deadline is in trouble. This is the check most often gamed before a review.
The float and duration checks: is the network honest?
6. High float. No more than 5% of activities should show total float above 44 working days. Excessive float almost always means missing logic upstream, not genuine flexibility.
7. Negative float. The target is zero. Negative float means the plan, as built, cannot meet its own commitments. It is not a footnote to be explained away. It is the schedule stating plainly that the date will be missed without intervention.
8. High duration. No more than 5% of incomplete activities should exceed 44 working days. Long activities resist objective progress measurement. If a single activity spans three update cycles, its percent complete is an opinion.
The data hygiene checks: is the status real?
9. Invalid dates. Zero tolerance. Forecast work in the past and actual dates in the future are both signs that updates are processed mechanically rather than reviewed.
10. Resources. Every activity with duration should carry resources or cost. Without loading, the schedule cannot be tested against the estimate, and earned value becomes guesswork.
11. Missed tasks. No more than 5% of activities should have finished, or be forecast to finish, later than baseline. This is the first check that measures performance rather than construction quality, and it is where deteriorating projects show up first.
The performance checks: is the plan being executed?
12. Critical path test. Push a critical activity out and confirm the project finish moves with it. If it does not, the network is broken, and every critical path report published to date has been decorative.
13. Critical path length index. CPLI measures how efficiently the remaining critical path must run to hit the baseline finish. Below 0.95, the project needs above-plan performance on its most constrained work.
14. Baseline execution index. BEI compares tasks completed against tasks that should have been completed. Below 0.95, the project is quietly falling behind, whatever the milestone dashboard says.
The failure patterns we see repeatedly
Across hundreds of schedules in the database, the failures cluster. Constraints are used to hold contract dates while negative float accumulates behind them. Logic is maintained diligently up to contract award and abandoned afterwards. Progress updates pass through without logic repair, so the critical path drifts from reality one update at a time. And reviews concentrate on checks 1 to 11, which measure construction quality, while ignoring checks 12 to 14, which measure whether the plan is actually being executed.
The assessment has limits. It tells you whether a schedule is mechanically capable of forecasting. It does not tell you whether the durations are achievable. That requires benchmarking against comparable delivered projects, which is a separate discipline.
But run honestly, on every update rather than once at baseline, the 14 checks turn schedule review from an argument about opinions into a conversation about evidence. That is worth an afternoon.
ScheduleIQ automates the full 14-point assessment and benchmarks the result against 120+ comparable programmes drawn from a 30GB+ live schedule database. To see your schedule scored, visit the ScheduleIQ product page or contact us.