Project AssureWhich delay analysis method do your records actually sustain?
Method selection is governed by two things: what the contract requires, and what the contemporaneous record can actually sustain. An analysis whose method the records cannot support will be discounted however sound the underlying entitlement.
Choosing a forensic delay analysis method is a judgement about evidence, not preference. It turns on whether a credible baseline exists, what the contract requires, what state the contemporaneous updates are in, and whether concurrent causes have to be separated. More than one method is often defensible on the same facts, and a report that runs a primary method and a cross-check is usually stronger than one that runs either alone.
What are you trying to prove?
Delay methods answer a different question from disruption or acceleration. This decides whether the selector applies at all.
Is there a credible baseline programme?
Every technique here depends on one. Answer this honestly before anything else.
When is the assessment being made?
Prospective methods model an event as it arises. Retrospective methods measure what the record shows.
What does the contract require?
Asked whichever way you answered the last question, because a prescribed method still binds you after the event.
What state are the contemporaneous records in?
Answer against the weakest part of the delay period, not the best of it. If more than one statement describes your project, pick that one; the selector will offer the other as a cross-check rather than discarding it.
Do employer-risk and contractor-risk events affect the same period of critical delay?
Concurrency drives method choice, because only some methods separate causes period by period.
General guidance, not advice, and no client relationship is created by using it. Method selection is fact-specific and contract-specific. More than one method may be defensible on the same facts, and the appropriate choice should be made by a qualified analyst on the actual records and the governing contract and law. Entitlement consequences, particularly for concurrent delay, are determined by the contract and the applicable law and differ between jurisdictions.
This selector is Project Assure's own synthesis. It is not endorsed by, affiliated with, or derived from the text of the Society of Construction Law or AACE International. Method names are terms of art in general use.
The seven techniques, side by side
Method names and the prospective/retrospective distinction follow the Society of Construction Law Delay and Disruption Protocol, 2nd edition (February 2017). The Protocol identifies six techniques; a straightforward as-planned versus as-built comparison is widely used in practice alongside them and is included here, making seven.
| Method | Viewpoint | What it does | What your records must sustain | Where it is weak |
|---|---|---|---|---|
| Time Impact Analysis | Prospective | A fragnet modelling the delay event is inserted into the programme in force at the time, and the predicted effect on completion is read off. | A well-maintained programme in force at the time of the event, and an event that can be discretely identified and dated. | Data-hungry and sensitive to update quality. It produces a forecast, and a tribunal reviewing a completed project may prefer what actually happened. |
| Impacted As-Planned | Prospective | Delay events are added into the baseline programme to model their theoretical effect. | A sound, logic-linked baseline. | It relies entirely on a sound baseline and ignores what actually happened on site. A tribunal reviewing a completed project can see the as-built, so a model that ignores it reads as academic. |
| Time Slice Windows | Retrospective | The programme is divided into periods; critical-path movement is measured in each window against the updated schedules, and the delay in each window is then attributed to causes from the contemporaneous record. | Contemporaneous programme updates with logic maintained through the delay period. | It can only separate employer and contractor causation if the contemporaneous updates actually captured both, with logic, fragnets and notices that match the events. |
| As-Planned versus As-Built Windows | Retrospective | A windows analysis driven by factual records rather than programme logic, comparing intended against actual progress period by period. | Reliable period-by-period progress and as-built data. | Causation is inferred from the factual record rather than modelled through logic, so it is more contestable where causes overlap. |
| Retrospective Longest Path Analysis | Retrospective | The as-built longest path is traced back from actual completion to identify what in fact drove the finish date. | A reliable, detailed as-built with dated actuals. | It identifies the driving path rather than the cause. It applies no contemporaneous test of criticality, and it is weak on separating employer and contractor responsibility. |
| Collapsed As-Built (but-for) | Retrospective | Delay events are stripped out of the as-built programme to show what would have happened without them. | A complete as-built programme whose logic was recorded contemporaneously. | It requires a fully logic-linked as-built, which is rare and often reconstructed after the fact. It reasons from effect back to cause, applies no contemporaneous criticality test, and the result is sensitive to the order in which events are removed. |
| As-Planned versus As-Built | Retrospective | A visual, factual comparison of the intended programme against what actually happened. | A baseline and an as-built record, both good enough to compare. | It shows correlation, not causation, and is weak where multiple causes overlap. |
Take it with you
The whole selector on one A3 sheet: the six questions, the seven techniques side by side, and what to do when the answer is not a method. Free to print, pin up and share with attribution.
Related reading
Bring this discipline onto your programme.
Method selection is a judgement made on the records in front of you. A principal responds within 24 hours.
Start the conversation