Time Impact Analysis vs. Windows Analysis: Choosing a Forensic Delay Method

Time impact analysis vs. windows analysis: picking the right forensic delay method

Time impact analysis and windows analysis are two of the most widely used techniques in forensic schedule analysis, and choosing between them is rarely a matter of preference. The method you select signals how you understand the delay, what data you had, and whether a reviewer, arbitrator, or judge should trust your conclusion. Getting the choice right protects credibility long before anyone examines your numbers.

The industry framework most practitioners lean on is AACE International's Recommended Practice 29R-03. It does not crown a single best method. Instead it organizes techniques into families and gives each a shorthand identifier, so that an expert can name exactly what they did and an opposing expert can test it on the same terms.

How AACE 29R-03 organizes the methods

AACE 29R-03 sorts forensic schedule analysis techniques along a few practical axes. Understanding these axes is more useful than memorizing method names, because they explain why a method behaves the way it does.

• Prospective vs. retrospective: whether the analysis models delay as it was expected to unfold, or reconstructs what actually happened after the fact.
• Observational vs. modeled: whether you observe delay in the schedule updates that already exist, or actively insert or remove activities to model a scenario.
• Additive vs. subtractive: whether you add delay events into a schedule to see their effect, or strip actual events out of an as-built to infer what would have happened without them.
• Static vs. dynamic logic: whether the network is recalculated (CPM re-run) as part of the analysis, or evaluated without recomputation.

Nearly every named technique is just a combination of these traits. Once you can place a method on these axes, its strengths and vulnerabilities become predictable.

Time impact analysis: the prospective, additive approach

Time impact analysis, often called the fragnet insertion method, is prospective and additive. You take a schedule update that reflects status just before a delay event, build a small fragment of network logic representing that event, insert it, and recalculate the CPM. The difference in the projected completion date is the modeled impact.

TIA is the natural choice when a claim needs to be evaluated contemporaneously or near-contemporaneously, and it is frequently the method a contract specifies for time extension requests. Its strength is that it isolates a single event and answers a clean forward-looking question: given where the project stood, what should this event have done to completion? Its weakness is that it is a model, not a record. If the inserted fragnet is speculative, if the update it sits on is unreliable, or if it ignores what actually happened afterward, the result loses persuasive force.

Contemporaneous windows analysis

Windows analysis, sometimes called the contemporaneous period or time-slice method, is observational and retrospective. You divide the project into consecutive windows, usually aligned with schedule updates, and examine what drove the critical path during each period using the schedules that existed at the time. Because it relies on contemporaneous data rather than after-the-fact modeling, a well-executed windows analysis is often regarded as among the most credible techniques available, especially when regular, reliable updates exist.

The tradeoff is data dependence. Windows analysis is only as good as the update history feeding it. Sparse, poorly maintained, or manipulated updates undermine it, and the analyst must still exercise judgment when a window shows shifting or concurrent critical paths.

Collapsed as-built and impacted as-planned

When contemporaneous updates are missing or unreliable, two modeled methods fill the gap from opposite directions.

Collapsed as-built is retrospective and subtractive. You build a detailed as-built network of what actually happened, then remove specific delay activities and recalculate to estimate the completion date that would have occurred without them. It can be powerful when the record is strong, but constructing defensible as-built logic is demanding, and reviewers scrutinize how removals were chosen.

Impacted as-planned is prospective and additive, applied to the baseline. You insert delay events into the original plan and recalculate. It is the simplest to perform and, for the same reason, the least persuasive: it ignores actual progress and assumes the baseline was perfectly executable. It is best reserved for early screening or situations where no other data exists.

How method choice affects credibility

A forensic conclusion is judged not only on its answer but on the fit between method, data, and question. A few principles hold across disputes:

• Prefer methods that use contemporaneous data when reliable updates exist.
• Match the method to the question: forward-looking entitlement favors TIA; retrospective causation favors windows or collapsed as-built.
• Disclose the method by its AACE 29R-03 family so it can be tested transparently.
• Address concurrency and pacing explicitly rather than letting a single method obscure them.

A method chosen because it flatters the result, rather than because it fits the facts, is the fastest way to lose an expert's credibility.

Fragnet produces time impact analysis, contemporaneous windows, collapsed as-built, and impacted-as-planned results as clearly labeled, reproducible outputs, so your method and your reasoning are transparent from the first review to the last.