Assessing NEC4 compensation events: the five judgement calls that decide how much delay you recover
- Aug 3, 2025
- 12 min read
By Roman Bazelchuk | NEC Accredited Project Manager | APMG Project Planning and Control Founder, NEC Planning Solutions Ltd
Give the same compensation event to two competent planners. Same accepted programme, same dividing date, same facts on the ground. One comes back with three weeks of delay to planned Completion. The other comes back with nine days. Both can defend their model. Both followed clause 63.5. Neither has made an arithmetic error.
The difference between them is not the modelling. It is five judgement calls that sit inside every NEC4 delay assessment, that the contract deliberately leaves open, and that most contractors make silently without realising they have made them. The number at the end of the assessment is downstream of those calls. The project manager who rejects a quotation as "overstated" or "not adequately supported" is almost never disputing the critical path arithmetic. They are disputing a judgement call the contractor never stated, and because it was never stated, the whole assessment gets rejected instead of the specific call being argued.
This article names the five calls, sets out what the contract actually says about each, shows where two competent planners diverge, and gives the defensible position for each one. How to package the finished assessment so the project manager can verify it, the five structural elements of the submission itself, is covered separately in how to structure a time impact assessment under NEC4. That article is about the container. This one is about the decisions that go inside it, which is where the delay you recover is actually decided.
Why two planners assess NEC4 compensation events differently
The contract fixes the frame of the assessment precisely. Clause 63.5 assesses delay as the length of time that, due to the compensation event, planned Completion is later than planned Completion as shown on the accepted programme current at the dividing date. Clause 62.2 requires the quotation to include details of the assessment. The assessment is prospective: a forecast of the event's effect from the dividing date forward, not a reconstruction after the fact. The article on NEC delay analysis and extension of time covers why the prospective principle exists and what it replaced.
What the contract does not fix is the judgement inside the frame. The accepted programme at the dividing date might be five weeks old: how much of what happened in those five weeks enters the baseline before the event is impacted? The event touches one activity: does the assessment model the activity or the path the disturbance propagates through? An activity is part-complete at the dividing date: what happens to its remaining work? The contractor is assumed to react competently and promptly: what does that assumption oblige you to build into the forecast, and what does it not? The changed work carries risk: how much allowance enters the assessment, and what happens to the allowances and float already in the programme?
Five open questions. Five calls. Two planners who answer them differently produce different numbers from identical inputs, and both are following the contract. The contractor who understands this stops treating the assessment as a calculation to be run and starts treating it as a set of positions to be taken and defended. That shift is the single biggest improvement available in most contractors' compensation event recovery, and it costs nothing but discipline.
Call one: how much reality enters the baseline
The accepted programme current at the dividing date is the contractual starting point, and it is almost never current in the ordinary sense of the word. Its data date may be weeks before the dividing date. In the gap, the project moved: progress was made or lost, other compensation events landed, and some delay accrued that is nobody's compensation event at all.
NEC's own practice note is clear that the assessment should recognise what happened between the accepted programme's data date and the dividing date, including other compensation events and delays from events that are not compensation events. You do not pretend the project stood still since the last acceptance. The judgement is how far that recognition goes, and this is where assessments are won and lost before the event is even impacted.
The two failure directions are recognisable. The under-cooked baseline impacts the event onto the accepted programme exactly as it was, ignoring the intervening period. If the project had already drifted, the event's modelled effect lands on a fiction, and the project manager's first challenge writes itself: the baseline does not represent the project at the dividing date. The over-cooked baseline goes the other way: in "updating" the programme to the dividing date, the planner quietly corrects logic errors, resequences work, and tidies float that was always there. Each correction may be individually sensible. Collectively they manufacture a different programme, and the assessment is now measuring the event against something that was never accepted. The project manager is entitled to reject it on that ground alone.
The defensible position sits between the two and is stated, not hidden. Take the accepted programme current at the dividing date. Apply actual progress to the dividing date. Recognise intervening events for what they are, including the contractor's own delay where it exists. Change nothing else: no logic repairs, no resequencing, no float housekeeping. Then impact the event. The baseline statement in the narrative says exactly this in three sentences, and the assessment that opens with it has already answered the most common ground of rejection. Where no accepted programme exists at all, the problem is different in kind, and the article on compensation events without an accepted programme covers that situation separately.
Call two: where the event actually bites
The lazy assessment models the activity the event touches. The defensible assessment models the path the disturbance travels.
A late access date does not just delay the first activity behind it. It moves a chain: the trade that follows, the test that follows the trade, the interface a third party was holding a window open for. Sometimes the chain runs straight down the existing critical path and the propagation is obvious. The harder cases, and the ones where planners genuinely diverge, are the events that land on a near-critical path and consume its float until a new critical path emerges partway through the forecast. One planner models the touched activity, sees float absorbing the impact, and concludes minimal delay. Another traces the propagation, finds the path goes critical in week three of the forecast, and concludes three weeks. Same event. Different answer. The second planner is doing what clause 63.5 actually asks, because the question is the effect on planned Completion, not the effect on the activity.
The judgement call has a second face: separation. The assessment measures the effect due to the compensation event, not the effect of everything else going wrong on the project at the same time. Where the event's path runs through territory already disturbed by other causes, the planner has to hold the causes apart, and the narrative has to show the separation rather than assert it. The assessments that survive scrutiny are the ones where the project manager can follow the disturbance from the event, through named activities and named logic, to the movement in planned Completion, with everything else visibly left where it was.
The practical discipline: before modelling anything, write the causal chain in plain sentences. The event changes this. That delays this operation. That operation feeds this one. Planned Completion moves because of this link. If the chain cannot be written in sentences, the model is not ready, and no amount of scheduling software will fix a chain the planner cannot articulate.
Call three: the dividing date split
The dividing date cuts every activity in the programme into one of three states, and each state gets different treatment. Getting this split wrong is the quietest way to lose an assessment, because the errors look like modelling choices rather than contractual ones.
Activities complete before the dividing date are history. Their actual timing stands. The event may have affected them, and where it did, that effect is recognised, but their dates are not rewritten to make the model cleaner. A planner who adjusts completed work to strengthen the forecast has stopped assessing and started reconstructing, and the project manager who spots one rewritten actual will distrust every forecast in the package.
Activities not started at the dividing date are pure forecast. Their durations change only where the event genuinely changes them: new scope, changed methods, changed conditions, changed logic. The discipline here is restraint. The event is not an invitation to revisit every duration the planner was never comfortable with.
Part-complete activities are where the judgement concentrates. The work done is history; the work remaining is forecast; and the assessment needs a stated basis for the remaining duration. Two planners diverge here constantly: one carries the original remaining duration, one re-derives it from performance to date, and the difference flows straight into planned Completion. Neither is automatically right. What is automatically wrong is silence. The defensible position states the basis: remaining duration held at programme logic because the event does not change the method, or remaining duration re-derived because the event changes the conditions under which the balance of the work is done, with the derivation shown. One sentence per part-complete activity on the affected path. That sentence is the difference between a challengeable assumption and an attackable hole.
Call four: what competent and prompt actually obliges you to assume
The contract assumes the contractor reacts competently and promptly to the event, and assessments are built on that assumption. This is the call that most reliably separates over-claimers from under-claimers, because both misread the same words.
The over-claimer assumes nothing. The event lands, the forecast absorbs it raw, no re-sequencing, no practicable parallel working, no use of the flexibility a competent contractor genuinely has. The project manager reads the assessment as a contractor doing nothing in the face of a problem, and marks it down accordingly, often by making their own assessment under clause 64, which is the worst commercial outcome available.
The under-claimer assumes heroics. Resources doubled, sequences compressed, premium hours worked, the impact mitigated nearly to zero, and the contractor carrying the cost and risk of all of it without instruction. This is not competence. It is uninstructed acceleration dressed as mitigation, and it gives away entitlement the contract never asked the contractor to surrender. The boundary between the two is one of the most consistently misunderstood lines in NEC4, and it has its own treatment in the practical guide to acceleration and mitigation.
The defensible position assumes a competent contractor, not an extraordinary one. Re-sequencing where the logic genuinely allows it. Parallel working where access, resources and safety make it practicable. Reasonable management of the things a contractor manages. And nothing that costs money or risk the event did not create: no uninstructed extra gangs, no compressed testing, no premium time. The mitigation assumptions go into the narrative as named, credible measures with their constraints stated: resources available from, access dependent on, approvals required by. A mitigation plan with visible constraints reads as competence. A forecast with no mitigation reads as opportunism. A forecast mitigated to nothing reads as a contractor who has not understood what they are entitled to.
Call five: what allowance enters, and what survives
The final call is the risk and float treatment, and it decides more disputed days than any other because it is the least understood.
Two movements happen in this call. The first is what enters: the assessment of the changed and new work includes risk allowances for cost and time for matters which have a significant chance of occurring and are at the contractor's risk. Not a blanket percentage smeared across the impact, which is the version project managers strike out on sight, but reasoned provision tied to the specific exposures the event creates: weather on newly external work, commissioning iterations on changed systems, productivity on disrupted operations. Each allowance earns one sentence of justification, the same discipline that earns allowances acceptance in the programme itself.
The second movement is what survives: the provisions already in the programme. NEC guidance is explicit that the contractor's existing time risk allowances are retained in the assessment. The event's impact lands on top of them, because they are provision for the contractor's risk and the event is not the contractor's risk. The same logic preserves terminal float: planned Completion moves, and the Completion Date moves with it, so the gap the contractor built between the two is protected. A planner who lets the event consume declared allowances, or who measures delay to the Completion Date instead of to planned Completion, is surrendering protection the contract grants. The full mechanism, including why hidden risk gets none of this protection, is covered in time risk allowances in NEC.
The defensible position, stated in the narrative: allowances on the changed work itemised and justified; existing allowances identified and retained; delay measured to planned Completion with terminal float preserved. Three lines. Most rejected assessments cannot produce them.
Turning the calls into a defensible submission
The pattern across all five calls is the same. The contract leaves the judgement open, two competent planners can answer it differently, and the assessment that survives is the one that states its answer instead of burying it.
This is the practical move that changes outcomes: every call goes into the narrative as an explicit, numbered assumption. The baseline statement. The causal chain. The remaining-work basis for each part-complete activity on the path. The mitigation measures and their constraints. The allowance treatment. Stated assumptions convert the assessment from an opinion the project manager must accept or reject wholesale into a position they can test at specific points. That sounds like an invitation to challenge, and it is, deliberately. A challenge to one named assumption is a conversation that ends in an agreed adjustment. A rejection of an unexplained number is a cycle of resubmission that ends, too often, in a clause 64 assessment at a fraction of the entitlement. The contractor who exposes the calls keeps control of the argument. The contractor who hides them hands the argument, and usually the assessment, to the other side of the table.
Once the five calls are made and stated, the package itself, the dividing date sheet, the programme versions, the fragnet, the impacted programme and the narrative, is assembled the way the time impact assessment structure guide sets out. Calls first, container second. Most contractors do it in the other order, which is how technically competent assessments end up rejected for reasons nobody in the planning team can quite name.
Free download: The Five Judgement Calls checklist.
A one-page, desk-side pre-submission checklist covering all five calls, the check items under each, and the assumption statements the narrative needs. Built to be printed and worked through before any NEC4 delay assessment leaves the building. No email required.
The number is not the assessment
Clause 63.5 gives every contractor the same method. The accepted programme at the dividing date, the event impacted, the movement in planned Completion measured. If the method were the whole assessment, two competent planners would never disagree, and project managers would have nothing to reject but arithmetic.
They disagree because the method used to assess NEC4 compensation events is a frame around five open judgements. The judgements are where the delay is actually decided. How much reality enters the baseline. Where the event bites. How the dividing date splits the work. What competence obliges you to assume. What allowance enters and what survives. The contractor who makes the five calls deliberately, defensibly and visibly recovers what the event justifies, on a timeline the project can live with. The contractor who makes them silently produces a number, and a number with hidden assumptions is not an assessment. It is an opening offer, and the project manager prices it like one.
FAQ
How is delay assessed for an NEC4 compensation event?
Under clause 63.5, the delay to the Completion Date is the length of time that, due to the compensation event, planned Completion is later than planned Completion shown on the accepted programme current at the dividing date. The assessment is prospective: a forecast of the event's effect from the dividing date forward, built on the accepted programme, not a retrospective reconstruction after the works are complete.
What is the dividing date in NEC4?
The contractual snapshot point for the assessment. For compensation events arising from an instruction or other communication, it is the date of that communication; for other events, the date the event was notified. The assessment starts from the accepted programme current at that date, with progress and intervening events recognised up to it, and forecasts forward from there.
Does the assessment have to assume mitigation?
It assumes the contractor reacts competently and promptly to the event: re-sequencing where logic allows, practicable parallel working, reasonable management. It does not assume uninstructed acceleration, which is additional resource, premium time or compressed sequences at the contractor's cost. Confusing the two either gives away entitlement or produces a forecast the project manager will not accept.
Can a valid compensation event still produce a rejected delay assessment?
Routinely. The event's validity and the assessment's quality are separate questions. Assessments fail on the judgement calls inside them: a baseline that does not represent the dividing date, an impact modelled on an activity instead of a path, silent remaining-duration assumptions, missing or heroic mitigation, blanket risk percentages. Stating each call as an explicit assumption is what converts a contestable number into a defensible position.
What risk allowances go into a compensation event assessment?
Allowances for cost and time for matters with a significant chance of occurring that are at the contractor's risk, tied to the specific exposures the changed work creates and justified individually rather than applied as a blanket percentage. Separately, the contractor's existing time risk allowances in the programme are retained: the event's impact is assessed on top of them, not absorbed into them, and terminal float is preserved.
About the author
Roman Bazelchuk is the Founder of NEC Planning Solutions Ltd, a UK project planning and controls consultancy supporting contractors with NEC programme compliance, compensation event assessments and live project controls. He is an NEC Accredited Project Manager and holds the APMG Project Planning and Control qualification, with a BSc in Mechanical Engineering and postgraduate training in Planning and Control.
NEC Planning Solutions provides contract-aware planning support through a QA-governed delivery model, helping project teams keep programmes accepted, current and commercially useful from tender through to live delivery.
Quotations coming back rejected, or the project manager assessing your events for you?
If delay assessments are cycling through revise-and-resubmit, if clause 64 assessments are landing at a fraction of the entitlement, or if the team can run the model but cannot defend the assumptions inside it, specialist compensation event support builds assessments around the five judgement calls and the evidence that carries them.
