Collapsed As-Built Analysis: When But-For Persuades and When It Fails

Collapsed as-built analysis - also called but-for analysis - removes one party's delay events from the as-built programme to show when the project would have finished without them. The gap between the actual and collapsed completion dates is the delay attributed. Tribunals find this easy to follow, but every link is rebuilt after the event.

Contractors often reach for a collapsed as-built when the baseline programme has let them down, because the method needs no baseline. That is a sound reason to consider it and a poor reason to choose it: the method swaps an argument about the baseline for an argument about as-built logic, which is often harder to win.

What Is Collapsed As-Built Analysis?

The Society of Construction Law Delay and Disruption Protocol, 2nd edition (February 2017), lists collapsed as-built among six commonly used delay analysis methods. Paragraph 11.6(f) describes it as the extraction of delay events from the as-built programme to provide a hypothesis of what might have happened had those events not occurred. It needs no baseline programme, but it does need a detailed, logic-linked as-built programme.

That last requirement is the catch. The Protocol notes that such a programme rarely exists, so the analyst usually has to add logic to a verified as-built, which it calls a time consuming and complex endeavour. A bare as-built is only a bar chart of actual dates, and a bar chart cannot be collapsed.

The Protocol classes the method as cause and effect, with the critical path and the delay impact both determined retrospectively. AACE International's recommended practice 29R-03 calls it a modeled, subtractive method, also known as but-for analysis or as-built but-for. Either way, the question is commercial: what completion date the project would have achieved without the other side's events.

How a Collapsed As-Built Analysis Works, Step by Step

Building a collapsed as-built analysis follows six steps:

1. Verify the as-built data - every significant activity needs an actual start and finish supported by daily reports, labour and plant returns, dated photographs and inspection records.
2. Add logic, linking activities so the scheduling software recalculates the actual dates exactly - AACE treats that match as a minimum condition before anything is removed.
3. Find the delay events inside the as-built - some are discrete activities such as a suspension, others are buried in extended durations and must be split into productive and delay time.
4. Collapse: remove the selected events, recalculate, and read the earlier completion date that results - the but-for date.
5. Measure and attribute: if a project finished 120 days late and removing the employer's events collapses completion by 70 days, those events caused 70 days of critical delay. At QAR 40,000 a day in liquidated damages, QAR 2.8 million now rests on the model's logic.
6. Test the answer by collapsing the contractor's events too, and check the result against what the site minutes said was driving completion - a contradiction needs explaining before the other side asks.

When Does But-For Analysis Persuade?

Collapsed as-built persuades when four conditions line up:

• The baseline is unreliable or was never accepted.
• The as-built record is detailed enough to carry logic.
• A few dominant events drove the critical path.
• The question is asked after completion, as with a long access delay or a design freeze on an otherwise orderly project.

It also suits questions that are counterfactual by nature. Fluor v Shanghai Zhenhua Heavy Industry (2018) was a damages claim over defectively welded monopiles. The judge accepted that a prospective analysis is correct for an extension of time, but held that the damages question required some form of retrospective analysis. The starting point was the position as if the contract had been properly performed, compared with what actually happened.

That is the shape of a but-for argument, so the method sits comfortably with prolongation cost claims. Under Core Principle 14 of the Protocol, a contractor recovers that cost for a concurrent period only if it can separate the employer's delay costs from its own. A careful collapse can help show the split.

Where Collapsed As-Built Fails: Hindsight, Subjectivity and Cherry-Picking

Three weaknesses limit how far a collapsed as-built can be pushed:

• Hindsight: the logic is written after the event by someone who already knows how the project ended, so a collapse can be arithmetically correct and evidentially worthless.
• Subjectivity: the analyst decides how activities link and how much of an extended duration was delay; change those calls and the but-for date moves, which is why AACE expects every change to be tabulated and justified.
• Cherry-picking: the Protocol's summary table says the method requires selecting delay events to model, and rates cause-and-effect methods as generally more forensically reliable after completion because they consider every potential cause.

Removing an event also leaves the rest of the as-built sequence untouched, yet that sequence was often the contractor's response to the event. In Fluor the judge refused to give a hypothetical scenario more precision than the circumstances justified, observing that it would almost certainly have differed from the actual as-built sequence.

Collapsed As-Built and Concurrent Delay

Concurrency is where the method is most often misread. The Protocol notes a structural limitation: the method measures only incremental delay, because the completion date will not collapse further than the closest near-critical path. Where employer and contractor events run in parallel, removing either alone leaves the other holding the date, so each collapse shows little or nothing.

The problem is rarely the arithmetic. It is that a pure but-for test asks a different question from the extension of time clause. Core Principle 10 of the Protocol says concurrent contractor delay should not reduce any extension due. In Walter Lilly & Co v Mackay [2012] EWHC 1773 (TCC), Akenhead J held the contractor entitled to time for concurrent delay. A one-sided collapse across a period of concurrent delay could show none at all.

Contract wording can change the answer. In North Midland Building v Cyden Homes [2018] EWCA Civ 1744, the Court of Appeal upheld a clause that disregarded Relevant Event delay concurrent with contractor delay when assessing extensions. Where a GCC civil code governs, the contract and the local law decide how shared causation is treated, and the collapse has to follow them.

Collapsed As-Built vs Impacted As-Planned

The two methods are mirror images. Impacted as-planned adds delay events to the logic-linked baseline and reads the forecast effect. Collapsed as-built subtracts them from the logic-linked as-built and reads the hypothetical effect. Both are cause and effect methods in the Protocol's terms, so both depend on which events the analyst models.

The Protocol treats impacted as-planned as the simplest and least expensive method but notes that it does not consider actual progress. Collapsed as-built is the reverse: full hindsight, high cost and logic nobody wrote at the time. Where monthly updates exist, windows analysis or time impact analysis usually rests on firmer ground, and our guide to delay analysis methods sets out the wider choice.

Practical Tips Before You Rely on a Collapsed As-Built

Before relying on a collapsed as-built, take three steps:

1. Audit the as-built record first - if daily reports, labour returns and photographs cannot fix actual dates for the critical activities, the logic will be invented rather than reconstructed, and the opposing expert will notice. A claim readiness score tests the record early.
2. Build the model to survive cross-examination - reproduce every actual date before removing anything, source each link to a document, log every change, and collapse both parties' events to test whether the order of removal changes the answer.
3. Think about the forum - the Protocol recommends agreeing a method before significant after-the-event analysis and warns that failing to consult may count when costs are allocated.

In CALIM's experience across dozens of engagements, the collapsed as-built analyses that hold up are two-sided, document every link and agree with what the site recorded at the time. Not the most sophisticated models. The most transparent ones.

At CALIM, we audit the as-built record before recommending a but-for analysis, then collapse both parties' events so the result stands up to an opposing expert. Contractors facing a disputed extension or a liquidated damages deduction can bring in our EOT claims consultants while the site records are still easy to gather.

Frequently Asked Questions