On this page
Two people can open the same project and calculate different results if they use different settings. A defensible schedule comparison records the settings with the file, then checks whether they changed between updates.
Oracle’s P6 schedule settings reference documents the controls discussed here. Confirm the documentation for your installed version and preserve the actual configuration used for each run.
Start with out-of-sequence progress
Imagine a fictional finish-to-start relationship from rough-in to wall closure. Wall closure has started while rough-in remains incomplete. The schedule now contains actual progress that does not follow its original relationship.
Retained logic holds the remaining successor work behind the unfinished predecessor logic. Progress override permits progressed work to continue without that same delay from the predecessor relationship. This choice can move the forecast; it does not establish whether the field sequence was acceptable.
- Observed: wall closure started before all rough-in finished.
- Evidence question: was a completed area released separately?
- Model question: should the work be split into areas, or should a documented progress option describe the remaining sequence?
Do not switch settings just to make the project appear on time. Confirm what happened, determine how remaining work can proceed and document the approved modeling choice.
Capture the settings that affect interpretation
| Setting family | Review question |
|---|---|
| Progressed activities | How does unfinished predecessor work affect remaining successor work? |
| Critical activities | Is criticality based on a float threshold or longest path? |
| Total float | Which calculation definition and finish reference are selected? |
| Relationship lag | Which calendar governs the waiting time? |
| External relationships | Are relationships to other projects considered? |
| Expected finish dates | Are entered expected finishes used in the calculation? |
An activity flagged critical by a float threshold is not necessarily the same thing as an activity identified on a longest path. Keep the definition visible whenever a report counts “critical activities.”
Compare settings before explaining variance
Save the source file, calculation log, settings record and result together. When a date changes, first ask whether status, logic, calendars, durations or settings changed. A setting change can alter a result without any crew performing additional work.
If the received schedule does not identify the settings, request them. Label any substitute configuration as an assumption, and keep its result separate from the submitted forecast. A controlled sensitivity run can illustrate a setting’s effect, but it cannot prove which configuration the original author used.
Use XER review to preserve the source and calendar review to understand timing inputs. The Schedule Forensics library brings these checks into one workflow.
Share this article