Every Extension of Time claim rests on the same question: did this delay actually push out the completion date, and by how much? The answer depends entirely on which delay analysis method you use to get there — and picking the wrong one is one of the most common reasons a claim gets rejected or discounted, regardless of how genuine the underlying delay event was.

There are four methods that come up in almost every EOT dispute: Time Impact Analysis (TIA), Windows Analysis, Collapsed As-Built (CAB), and As-Planned vs As-Built. None of them is universally "correct" — each fits a different set of circumstances, and adjudicators, engineers, and dispute boards expect to see the method matched to the situation, not chosen for convenience.

Why the method you choose matters

A delay analysis method isn't just a calculation technique — it's an argument about causation. TIA argues prospectively: "if we insert this delay into the schedule at the moment it happened, here's what it does to the finish date." Windows Analysis argues retrospectively: "here's what actually happened to the critical path during this specific period, and here's what caused it." These aren't interchangeable framings. A method mismatched to the available evidence — or to what the contract actually requires — invites exactly the kind of challenge that turns a strong delay event into a weak claim.

The four methods at a glance

Method	Direction	Best suited for	What it needs
Time Impact Analysis (TIA)	Prospective (forward-looking)	Single, discrete delay events, assessed close to when they occur	An accepted baseline/update immediately before the delay
Windows Analysis	Retrospective (backward-looking)	Multiple delays over an extended period, especially overlapping ones	A full series of periodic schedule updates
Collapsed As-Built (CAB)	Retrospective	Projects where the baseline was poorly maintained or abandoned	A detailed as-built record, not a live baseline
As-Planned vs As-Built	Retrospective	Low-value or preliminary claims, quick assessments	Just the original plan and the final as-built record

Time Impact Analysis (TIA)

TIA is the method most contracts and adjudicators default to when a delay event needs to be assessed in real time, close to when it happened. You take the schedule update immediately before the delay, insert the delay as a "fragnet" (a small fragment of activities representing the delay event), and recalculate. The shift in the finish date is the time impact — and, in most cases, the basis for the EOT claim itself.

Because it's forward-looking and doesn't require reconstructing what actually happened afterward, TIA is fast relative to the alternatives, and it's the method FIDIC-style notice provisions are effectively built around: you notify the delay, and TIA is the natural way to quantify it while the schedule logic is still fresh and uncontested.

A full standalone guide on TIA — including the fragnet-insertion mechanics — is planned as a follow-up piece linked from here.

Windows Analysis

Windows Analysis takes the opposite direction: instead of predicting forward from one delay event, it divides the whole project duration into a series of "windows" (often tied to the actual periodic schedule updates that were issued) and looks at what really happened to the critical path within each window.

This is the method of choice when a project has multiple delays, especially overlapping or concurrent ones, spread across a long period — situations where a single TIA can't cleanly isolate cause and effect. The tradeoff is data: Windows Analysis needs a complete, credible run of periodic updates. If those updates are missing, inconsistent, or weren't contemporaneously approved, the method loses much of its evidentiary strength.

A full standalone guide on Windows Analysis is planned as a follow-up piece linked from here.

Collapsed As-Built (CAB)

CAB is the method of last resort when the baseline itself can't be trusted — either it was never properly maintained, or it diverged so far from reality during the project that it no longer reflects a credible "as-planned" position. Instead of comparing planned versus actual, CAB works backward from the as-built record: it removes the delay events one at a time from what actually happened, to reconstruct a hypothetical "but-for" completion date — what would have happened if those delays hadn't occurred.

Because it relies entirely on the as-built record rather than a live baseline, CAB tends to be the most labor-intensive of the four methods, and the most exposed to disputes about which delay events genuinely belong on the critical path.

As-Planned vs As-Built

The simplest of the four: lay the original plan next to what actually happened and compare. It's quick, cheap, and doesn't require any of the intermediate schedule updates the other methods depend on — which makes it a reasonable fit for smaller or lower-value claims where the cost of a full TIA or Windows Analysis wouldn't be proportionate.

Its weakness is exactly its simplicity: it doesn't distinguish which party caused which delay, doesn't handle concurrent delays well, and rarely holds up as the primary evidence in a high-value or contested dispute.

Choosing a method for your contract

FIDIC-based contracts, with their strict Sub-Clause 20.1-style notice windows, tend to push claims toward TIA by design — the notice obligation forces assessment close to the delay event, which is exactly when TIA works best. NEC contracts, with their compensation-event mechanism, follow a similar logic. JCT contracts and projects with weaker contemporaneous notice discipline are more likely to end up needing Windows Analysis or CAB after the fact, when the dispute is being reconstructed rather than assessed in real time.

The practical rule: if you have a clean baseline and you're assessing a delay close to when it happened, use TIA. If you're looking back at a project with multiple overlapping delays and a full set of periodic updates, use Windows Analysis. If the baseline itself is unreliable, CAB is often the only credible option left. As-Planned vs As-Built should be reserved for genuinely low-value or preliminary assessments, not as a shortcut around the other three.

Delay Analysis Methods in Construction: TIA, Windows, and Collapsed As-Built Compared