Delay analysis is often presented as though it produces a single correct answer. That is what clients commissioning it expect. It is not as simple as that.

The same delay event, on the same project, with the same records, can yield materially different results. That is not a flaw in the discipline but the nature of it.

To understand why, we examine the three sources of variation that dictate the outcome of any analysis - capacity, method, and application.

Capacity - the role of the analyst within the bounds of instruction

The objective of a delay analysis is dictated by the role of the person performing it. There is a distinct professional difference between an analyst engaged in an advocacy role and one engaged as an independent expert.

• The advocate builds the strongest defensible case for the instructing party within the bounds of the evidence. They highlight the interpretations of the data that support their client's position. They are always instructed by one of the parties - whether as a shadow expert working in the background to support the broader strategy, or as a named expert representing the instructing party.
  
• The independent expert owes a duty to the tribunal, court or arbitrator. They must reach a conclusion they would hold regardless of who instructed them. They may be instructed by one party, jointly by both, or by the tribunal itself.

Both roles are entirely legitimate. But they are not the same exercise.

Expecting both to land on the same number misunderstands what each has been asked to do, i.e. the advocate to make the strongest honest case, the expert to find the answer they would defend to anyone.

Method - choosing the analytical lens

There is no single correct methodology. The SCL Delay and Disruption Protocol sets out six common methods; AACE International Recommended Practice No. 29R-03 identifies nine. Other methods exist beyond both. The right one depends on the available records, the complexity of the project, the time available, the nature of the delay, and many other factors. In fact being able to select and advocate for the right method to be used is a skill as important as conducting the analysis itself. The methods are not interchangeable, because they interrogate causation from different directions.

Take the four most commonly cited: Impacted As-Planned, Time Impact Analysis, Windows Analysis and Collapsed As-Built.

The additive, prospective methods (Impacted As-Planned and TIA) ask what an event was likely to cause at the time it occurred, on the assumptions reasonable then. These methods ask the analyst to model, theoretically what the most likely outcome would have been.

The retrospective methods (Windows Analysis and Collapsed As-Built) act as a post-mortem. They ask what actually happened, once the full effect of the delay was known.

Run all four across the same project and you should expect a range of outcomes, not a single point.

Application - the human element

Even once capacity is settled and method chosen, the analysis remains a human endeavour. Give the same methodology, the same records and the same programme to three competent analysts, and you will get three different results. Close, in all likelihood, but not identical.

This is not a defect. The leading recommended practice is explicit that forensic schedule analysis is "both a science and an art," that every method involves judgement calls in both preparation and interpretation, and that no method is exact. 

The variation comes from a sequence of professional judgement calls. For example:

• Baseline selection - which programme best reflects reality at the starting point? Which satisfies the criteria of the chosen method, and which is contractually correct? Rarely as obvious as it appears, and a baseline carrying a fundamental error can be stripped of evidential value entirely (Skanska v Egger and more recently Mace v Baltic).
• Logic and fragnet placement - how is the delay event modelled, and where is it inserted? How did the project team reflect it at the time, and were they right to? The small differences in how the fragnet is tied propagate through the whole calculation.
• Window and period selection - where are the slices cut, how many, and how long? Window size is not neutral. The same records, sliced differently, can shift which events appear critical and how delay is apportioned between the parties.
• Critical path interpretation - which path truly drove completion? Was it shaped by genuine sequence, or by modelling choices? Retained logic versus progress override, out-of-sequence handling, constraint and calendar assignments that determine what shows as critical? Does the reported path reflect what actually controlled the works, including near-critical paths a single calculation can hide?
• Concurrency and apportionment - were two effective causes truly concurrent, or merely sequential? Is one dominant? Is the delay shared, apportioned, or netted, and on which approach (Malmaison, or City Inn v Shepherd Construction)? 
• As-built interpretation - when did an activity start and finish? Does "complete" mean the slab was poured or cured? Each call is trivial alone but multiplied across thousands of activities, they set the band of accuracy within which any as-built result can honestly be stated. On a multi-year project that band may be a month, not a day.
  

Each of these choices, and many more, can be perfectly defensible on technical grounds. Each one moves the needle.

The mark of a sound analysis is that every judgement is identified, reasoned, and open to scrutiny.

Control of the output

So where does that leave the contractor relying on the number? 

In my view subjectivity in itself is not the problem. Unexamined subjectivity is.

A competent analyst does not merely run the software or follow the steps of the selected method. They understand and document precisely how each micro-choice drives the macro-result. They can explain their methodology, defend their judgement calls, and challenge the other side's choices on exactly the same basis.

An analyst who does not understand those levers is not offering a defensible opinion. In construction claims, the difference between recovery and a failed claim is rarely the method itself but it is the ability to articulate why that method was chosen, why those specific choices were made, and why they represent the most defensible view of the delays, their impact and entitlement they underpin.

Best, Radek Makar Director | ViViAD radek@viviad.co.uk

ViViAD is an independent consultancy specialising in Planning, Project Controls, Power BI Reporting, and Construction Claims Support. We work across the UK - typically on NEC contracts in defence, nuclear, renewables, and infrastructure - embedding with project teams to bring structure, clarity, and commercial control. Interim or longer-term, we fit around what you need.