A schedule can be approved, statused, and updated every month and still be worthless the moment it has to defend a claim. Approved means it looked fine in a meeting. Defensible means it survives a forensic analyst who is paid to take it apart. Those are different standards, and most owners only learn the difference after the schedule fails the second one.

Here is the checklist to run before the schedule has to hold up, while there is still time to fix what it finds.

Approved is not the same as defensible

Schedule approval usually means the durations looked reasonable and the finish date matched the contract. It rarely means anyone tested the logic. A schedule passes approval by looking right. It passes a forensic review by being right, which means its critical path reflects how the work will actually sequence and its calculations can be reproduced and trusted. The gap between those two is where claims are lost.

The checklist

Run these against any schedule update before you accept it.

1. The critical path is continuous and physically sensible. It runs from start to finish without gaps, and the activities on it are the ones that actually pace the job. If the critical path runs through something trivial, the logic is wrong.
2. Logic ties reflect real sequencing. Relationships represent how the work happens, not how it was easiest to draw. Start-to-start and finish-to-finish relationships with justified lags appear where the work overlaps.
3. No chains of finish-to-start zero-lag. A schedule where every activity is tied to the next with finish-to-start and no lag is not logic. It is a bar chart pretending to be a network.
4. No open ends. Every activity except the project start and finish has both a predecessor and a successor. Open ends break the critical path calculation.
5. Constraints are minimal, justified, and documented. Hard constraints override logic and can hide the real critical path. Each one should have a reason on the record.
6. Lags are justified. A lag should represent a real cure or cure period, like concrete, not a hidden buffer.
7. Out-of-sequence progress is resolved. When work happens out of the planned order, the schedule has to be reconciled, and you should know whether it runs retained logic or progress override, because the two produce different critical paths.
8. Total float values are reasonable. Widespread negative float, or suspiciously zero float everywhere, signals constraints or manipulation rather than a healthy network.
9. Activity durations are granular. A two-hundred-day activity hides everything happening inside it. Durations short enough to status meaningfully are durations you can defend.
10. Updates are genuinely statused. Actual start and finish dates entered, the data date advanced, progress real and not estimated. A contemporaneous update is the asset every later analysis depends on.

The three that fail most often

If you only have time to check three things, check these.

Finish-to-start zero-lag everywhere. It is the single most common tell that the schedule was drawn, not planned, and its critical path will not survive scrutiny.

Constraints doing the critical path’s job. When a hard constraint, rather than logic, is what holds the finish date, the schedule is not modeling the work. It is asserting an outcome.

Updates that were never really statused. A schedule updated by sliding bars instead of entering actual dates produces a critical path that reflects nobody’s reality, and it cannot anchor a window analysis later.

Why this matters before a claim, not after

A schedule cannot be made defensible retroactively. The moment you try to clean up the logic after a dispute has started, the cleanup itself becomes the issue, and the other side will say so. The defensible schedule is built in the monthly discipline of the updates, long before anyone is thinking about a claim. The owner who enforces that discipline is buying an option on every delay argument that follows. The owner who does not is hoping none ever come.

Frequently asked questions

What makes a construction schedule defensible? Continuous and sensible critical-path logic, relationships that reflect real sequencing, minimal and justified constraints, no open ends, reasonable float, and contemporaneous updates statused with actual dates. In short, a schedule whose calculations can be trusted and reproduced.

How do I review a CPM schedule as an owner? Test the logic, not just the dates. Check for open ends, finish-to-start zero-lag chains, unjustified constraints, and whether the updates were genuinely statused. The checklist above is the order to run them in.

What is an open-ended activity? An activity missing a predecessor or a successor. Open ends distort the critical path calculation and are a basic defect a forensic analyst will find immediately.

How many constraints should a schedule have? As few as possible, each documented. Constraints override logic, so a schedule heavy with hard constraints is usually hiding its real critical path behind them.