Physical artifacts on a dark desk: two bound English law reports stacked beside a printed construction programme with a Gantt chart and red critical-path bars highlighted, an antique brass pocket watch resting on the timeline, a fountain pen, documents tied with red barrister's ribbon, a magnifying glass, blueprint corner
Educational Guide 11 min read

Prospective or Retrospective? What Four English Judgments Actually Say About Choosing a Delay Analysis

What four English judgments say about choosing between prospective and retrospective delay analysis, and why the method label is not what a tribunal decides.

Every delay report makes this choice, usually in a methodology section near the front, and often in a single sentence. Do you assess delay as it appeared to a competent person at the time (prospectively, from the programme that was current), or do you look back from the end and work out what actually drove completion?

The received wisdom is that the answer is settled: prospective for extensions of time, retrospective for damages, pick one and defend it. The judgments are more interesting than that. Read together they say something closer to: the label on your method is not what the tribunal is deciding. What it is deciding is whether you have shown what critically delayed the works as they went along, and whether you can evidence it. The choice sits inside a larger construction schedule analysis, and it is the one practitioners most often make by reflex.

Four English judgments are quoted below. Each is linked to the full text on Find Case Law so you can check every paragraph cited.

1. Walter Lilly: the debate is sterile, if the analysis is done properly

Walter Lilly & Company Ltd v Mackay [2012] EWHC 1773 (TCC) is where the argument was put most squarely, and Akenhead J’s answer is the one most often quoted out of context.

He starts by noting the asymmetry that makes the question live at all. The architect granting an extension before practical completion is making a forecast; the tribunal is not. “a court or arbitrator has the advantage when reviewing what extensions were due of knowing what actually happened” (Walter Lilly ¶362).

That advantage does not licence a pure hindsight exercise. In the same passage, dealing with what the contract’s extension of time machinery requires, he says “one cannot therefore do a purely retrospective exercise” (Walter Lilly ¶365), and his reason is specific rather than stylistic. The vice he identifies is picking the last event in a sequence and treating it as the cause of the whole overrun. Instead: “One needs to consider what critically delayed the Works as they went along” (Walter Lilly ¶365).

Then the line that gets quoted on its own. The prospective-versus-retrospective debate was “sterile”, and it is worth reading why he thought so: because “if each approach was done correctly, they should produce the same result” (Walter Lilly ¶380). Note the conditional. That is not a finding that the two methods are interchangeable. It is a finding that both experts accepted they should converge, and that the argument between them was therefore not where the case lay.

Where did the case lay? Immediately afterwards, in the same paragraph: “An underlying problem, accepted as such by both experts, was the absence of a contemporaneous critical path programme” (Walter Lilly ¶380). Had one existed with sound logic, he says, both experts would in all probability have done the prospective exercise in time slices, but it could not readily be done, because there was no such programme.

The experts had already conceded the point between themselves. Their joint statement recorded that “there is no programme of all the works outstanding at that date which could sensibly be used as a baseline in a retrospective programme analysis” (Walter Lilly ¶371).

What the judge preferred, on those facts, was an expert who worked month by month through what was actually impacting progress, having regard to the longest sequence of outstanding work: “This was a wholly logical approach and, indeed is the approach used by most delay experts when there is a usable baseline programme from which to work” (Walter Lilly ¶378). And crucially, that expert did not treat his contemporaneous view as sealed. “He then applied a cross check by reference to what actually happened” (Walter Lilly ¶381). The judge saw no difficulty in that; on the contrary, “it is necessary to have regard to how long individual items actually took to perform” (Walter Lilly ¶378).

So the approach that survived contact with this tribunal was neither pure. It was contemporaneous in structure and retrospective as a check.

2. Fluor: they do not necessarily agree, and the question you are answering decides

Fluor v Shanghai Zhenhua Heavy Industry [2018] EWHC 1 (TCC) pulls the other way on the first point and adds a distinction Walter Lilly does not make.

Accepting the evidence of one of the delay experts, the judge records that a prospective analysis “does not necessarily produce the same answer as an analysis carried out retrospectively” (Fluor ¶275). Set against Walter Lilly ¶380 that is a real difference of emphasis, and it should be presented as one rather than smoothed over: in Walter Lilly both experts accepted the two approaches should converge if done correctly; in Fluor the court accepted evidence that they need not.

The more useful half of the paragraph is what follows. Of the two, “The former is the correct approach when considering matters such as the award of an extension of time” (Fluor ¶275), but that was not the exercise before that court, and so “I agree that some form of retrospective analysis is required” (Fluor ¶275).

That is the distinction worth carrying into practice. The approach is selected by the question, not by preference: valuing an entitlement that the contract machinery frames prospectively is a different exercise from proving what loss a breach actually caused.

Two qualifications keep this honest, and both matter.

First, neither expert’s approach was accepted: “I am not persuaded that the approach adopted by either expert is appropriate” (Fluor ¶276), on the unusual facts of that case, which turned on the effect of a waiver.

Second, and this is the qualification usually left out when Fluor ¶275 is cited: the court did not have to resolve the methodological dispute at all. “I do not need to decide which expert’s approach is correct, or even whether neither is correct” (Fluor ¶280). Whatever ¶275 is, it is not a ruling on which method wins.

The same judgment contains a warning about the practical risk of a prospective analysis left unchecked. Assessing how a claim would have looked to an assessor, the judge observed that “the claim had in fact been advanced on a prospective basis, and not in the light of what actually happened” (Fluor ¶510). The claim was built on 63 days, some nine weeks; the period for which installation was actually prevented was, on the judge’s finding in the same paragraph, some 3.5 weeks. That gap reinforced a suspicion that the claim was inflated. A forecast nobody ever reconciled to the outturn is a soft target.

3. Adyard: when the facts are clear enough, the choice does not bite

In Adyard Abu Dhabi v SD Marine Services [2011] EWHC 848 (Comm), the claimant’s own delay expert accepted in cross-examination that “there was no actual delay, whether viewed prospectively or retrospectively” (Adyard ¶293) caused by the items relied on. That accorded with what was happening on site (the project was already in irretrievable critical delay, ¶294), and the causation claim failed (¶295).

That is a small point with a large practical consequence. The methodological argument is expensive, and on facts like these it changes nothing. It is worth asking early whether the case you are working on is one where the answer differs by method at all, because if it does not, the argument is cost without benefit.

The same judgment contains a passage that any practitioner running a time impact analysis should read. Considering what the SCL Protocol contemplates, the judge accepted evidence that “any extension to the Contract Completion Date is determined by impacting an updated programme showing progress” (Adyard ¶291), that is, the programme current at the time, with progress to date in it. That is a precise description of the time-impact method, and it carries a demand most methodology sections skip past: the programme you impact must be the one that was live when the event struck, updated for actual progress, not the baseline re-run at leisure after the works are over. An impact bolted onto a stale or idealised programme is not the exercise the court described.

The court also recorded that the Protocol is not in general use in contracts and had not been approved in any reported case, and so was of little assistance on the legal causation issues before it (Adyard ¶289–¶290). Both halves of that are worth knowing: what the method is supposed to do, and how little weight the document describing it carried in that court. A methodology built on the Protocol alone stands on a document a court has already said it can set to one side.

Where this sits against the published commentary

Fluor was written up promptly and well, and the reading above is not a contrarian one. It is worth saying where it agrees and where it goes further.

CMS reached the same view of Walter Lilly’s “sterile” remark that this piece does: that it does not travel. Writing days after Fluor was handed down, Matthew Taylor, Shona Frame, Robbie Leckie and Aidan Steensma noted that the statement “has proved to be controversial, with commentators pointing out that a retrospective approach, by definition, takes into account matters which a prospective approach does not” (CMS, 15 January 2018). They also attribute Akenhead J’s remark partly to the preferred expert’s prospective method having been reality-checked against what actually happened, which is the reading of ¶378 and ¶381 set out above.

Lamb Chambers, writing on the same judgment, agrees that the approach follows the question but draws a directional conclusion: “The Judgment therefore lends weight to the retrospective approach” (Lamb Chambers). They then leave open whether that becomes the standard in similarly complex cases.

That is where this piece is more cautious, and the reason is ¶280. Five paragraphs after the passage everyone quotes, the court records: “I do not need to decide which expert’s approach is correct, or even whether neither is correct” (Fluor ¶280). A judgment that expressly declines to decide the methodological question is thin ground for concluding that it favours one method. Neither commentary above engages with that paragraph. That is the gap this piece is trying to fill: not a rival reading of ¶275, but a reminder that ¶280 caps how far ¶275 can be pushed.

None of which makes the commentary wrong on the substance. On method selection, all three readings agree. This is a caution about how much weight ¶275 will carry when the next expert cites it.

4. Saga Cruises: the situation at the time

Saga Cruises BDF Ltd v Fincantieri SPA [2016] EWHC 1875 (Comm) states the underlying requirement compactly, in a concurrency context: “Causation in fact must be proved based on the situation at the time as regards delay” (Saga ¶251).

That sentence is the thread running through all four. Whatever you call your method, the tribunal wants to know what was in fact driving completion at each point in the works, and it wants that proved rather than modelled.

What this leaves a delay analyst with

These four judgments do not resolve into a preference, and it would be wrong to report them as one. Held together, they support four working propositions:

1. The question selects the approach. An extension of time under contractual machinery is framed prospectively (Fluor ¶275). A claim about what loss was actually caused is not. Say in your methodology section which exercise you are performing, and why that dictates the approach, rather than which standard you are complying with.

2. Neither approach is used pure by the experts who get preferred. The analysis that survived in Walter Lilly was contemporaneous in structure with an explicit reality check against outturn (¶378, ¶381). A prospective number never reconciled to what happened invites the inference drawn in Fluor ¶510.

3. The contemporaneous record decides more cases than the method does. The determinative finding in Walter Lilly was the absence of a contemporaneous critical path programme (¶380), not a ruling on methodology. Where the programme record is thin, no choice of method rescues the analysis; where it is good, the choice matters less than the argument suggests.

4. A methodology section that only cites standards is exposed. The SCL Protocol was described in Adyard as not in general use and not approved in any reported case (¶289–¶290). Standards tell you what a method is; they do not tell you what a tribunal has done with it. Cite both.

The most common mistake in a methodology section is to argue for a method in the abstract. On the evidence of these four judgments, no tribunal is waiting to be persuaded that time impact analysis or as-planned versus as-built is superior in principle. They are waiting to be shown what critically delayed the works, as it went along, out of the contemporaneous record.

About this article

The collection behind it is incomplete, and this selection is narrower still. Ten English judgments have been read for this project. Four are discussed here, chosen because they address the choice of analytical approach directly. Nothing here should be read as a survey of English authority on delay analysis; there are judgments not in the collection that bear on it.

Where the judgments conflict, the conflict is presented rather than resolved. Walter Lilly ¶380 and Fluor ¶275 sit differently on whether the two approaches converge. Both are cited; neither is treated as having settled the point.

AI assisted the extraction from these judgments; a person checked every citation before publication.

Read the judgments. Quotations here are short and illustrative. The full judgments are public:

Not legal advice. This is commentary on published judgments for delay practitioners. It is not legal advice and must not be relied on as such. Whether any of it applies to your contract or your dispute is a question for a lawyer.