When a program finishes late and the parties disagree about why, delay analysis becomes the instrument that allocates time and money. Different methods, applied honestly to the same facts, can produce different answers. Choosing the method is therefore one of the most consequential decisions in any claim, and it is too often made by habit.
The recognized family of methods
Industry guidance, notably AACE International Recommended Practice 29R-03 on forensic schedule analysis and the Society of Construction Law Delay and Disruption Protocol, describes a family of approaches. The most frequently used are:
- As-planned versus as-built: comparing the plan with what actually happened
- Impacted as-planned: inserting delay events into the original plan
- Time impact analysis: modeling each event as a fragnet in the contemporaneous update
- Windows analysis: tracing the critical path period by period through the updates
- Collapsed as-built: removing delays from the as-built record to show the but-for date
Fit the method to the facts
No method is universally correct. The right choice depends on:
- The quality and continuity of contemporaneous updates
- Whether the analysis is prospective or retrospective
- What the contract requires for extension requests
- The number and nature of the delay events
- The forum: negotiation, mediation, arbitration or litigation
TIA done properly
Time impact analysis is the method most contracts specify for prospective extension requests, and the one most often done badly. A sound TIA starts from the last accepted update before the event, statused to its data date. The fragnet models the event as discrete activities with logic tied to the affected work, not a single bar with a convenient duration. The analysis reports the change in completion and in float on the affected paths, and it is submitted while the facts are fresh.
Common failures are predictable: an unaccepted or manipulated update used as the starting point, a fragnet tied to whatever produces the largest impact, mitigation ignored, and submissions made months after the event when the record has already moved on.
The method should follow the facts. When the facts are bent to fit the method, the analysis fails at the first cross-examination.
Windows when the record is strong
Retrospectively, windows analysis is persuasive where reliable updates exist. Each window shows how the critical path moved and what moved it, so the conclusions trace directly back to schedules the parties were managing by at the time. Collapsed as-built can work where updates are poor but as-built records are strong, though it relies heavily on the analyst's reconstruction of logic and draws closer scrutiny for it.
Concurrency changes the outcome
Concurrent delay, where owner and contractor delays affect the critical path in the same period, is among the most contested questions in construction disputes. The method must be able to identify concurrency clearly; the contract and the governing law then determine its consequences. An analysis that cannot see concurrency will be challenged by one that can.
What survives scrutiny
Analyses that hold up share common traits. They are reproducible, so the other side can follow every step. They rest on contemporaneous records. They identify the critical path objectively rather than by assertion. And they acknowledge their own limitations instead of leaving them to be discovered.
The best defense is a good record
The strongest delay position is built long before a dispute: an accepted baseline, disciplined updates at a consistent data date, timely notices and a daily record that says what happened and why. Programs that keep that record rarely need heroic analysis. Programs that do not often find that the truth can no longer be proven.
