How can an owner assess whether a contractor's completion date is realistic?

TLDR: For owner-side project controls leads reviewing a contractor's baseline: check the schedule can calculate a date, ask what supports its durations and milestone assumptions, then find out how likely the date is with a probabilistic forecast before you accept or return it.

The contractor's baseline arrives as a P6 file, the acceptance clock starts, and the finish milestone sits exactly on the contract completion date with no float to spare. Your controls lead now has days to decide whether that date will hold and to put the reasons in writing.

Our view is that the decision should rest on evidence, not on the date in the file. Ask the contractor where its key numbers came from, get an honest measure of how likely the date is, and write your response around what that evidence shows.

How can an owner tell whether a contractor's completion date is realistic?

An owner can tell whether a contractor's completion date is realistic by separating two questions the schedule answers together: could the work finish by that date, and how likely is it to? The critical path method answers the first by calculating the finish as if every activity takes exactly its planned duration. Only a probabilistic forecast answers the second, and acceptance needs both answers.

The US Government Accountability Office shows how far apart the two answers can be. In a worked example in its Schedule Assessment Guide (GAO-16-89G, December 2015), a schedule that calculates a 10 February finish has about a 7 percent chance of finishing on or before that date once realistic duration ranges are applied. A contractor planning for 70 percent certainty would promise 2 March instead.

The gap widens wherever paths converge. GAO calls this merge bias: if two parallel start-up and testing paths each have a 60 percent chance of finishing on time, the milestone they both feed has only a 36 percent chance (0.6 × 0.6), assuming the two paths are independent. Any commissioning or handover milestone fed by several paths behaves the same way, so a finish date can look comfortable on the critical path while depending on many things going right at once.

Merge bias is a property of the calculation, not a sign of careless planning: a well built single-date schedule carries the same optimism. That is why the review below asks for the evidence behind the date rather than starting from fault.

How should an owner review a contractor's baseline schedule before accepting it?

An owner should review the baseline in three steps, each depending on the one before: a forecast can only be as good as the logic it reads, and well-evidenced durations still need combining to show what they mean for the finish.

1. Check the schedule can calculate a date. Confirm it covers the full scope, every activity is linked by logic, calendars match the real working pattern, and any date constraints are few and justified.
2. Ask what supports the durations and milestone assumptions. Request the basis document and duration rationale, and test the assumptions behind key milestones, including third-party and owner-side interfaces.
3. Ask how likely the date is. Obtain a probabilistic forecast of the completion milestone, compare the contract date with the confidence level you need, and focus discussion on the activities with the greatest delay-causing potential.

Step 1: can the schedule calculate a date?

The schedule can only calculate a meaningful date when its scope, logic, calendars and constraints are sound, so check those before discussing the finish. GAO's guide describes a reliable schedule as comprehensive, well constructed, credible and controlled, and its ten best practices give owners a public checklist to work from.

• Scope: every work package in the contract scope, including owner-side activities such as approvals, free-issue equipment and access, appears in the schedule.
• Logic: activities have predecessors and successors, so a delay in one moves the dates of others. Open ends, long lags and leads deserve a written reason.
• Calendars: working days, shift patterns, holidays, possessions, outages or seasonal windows reflect how the work will actually run.
• Constraints: hard date constraints are rare and explained. GAO is direct on this point: "A customer-mandated date, including contractual obligations, does not constitute a legitimate reason to constrain an activity."

That last check matters most for the question you are answering. A finish milestone pinned to the contract date will always show the contract date, whatever the logic behind it says. Remove or relax the constraint in a copy of the file and look at where the logic alone puts the finish. Diagnostic counts of open ends or high float help prioritise, but GAO stresses that its measures have no pass-or-fail tripwires and that the severity of an error matters more than the count.

Step 2: what supports the durations and milestone assumptions?

Durations and milestone assumptions are supported when the contractor can show where each important number came from, and the owner can see which assumptions depend on someone else. GAO recommends that durations use the same rationale, historical data and assumptions as the cost estimate, and that every baseline comes with a basis document explaining ground rules, assumptions, constraints, lags and long durations.

Ask for that basis first, which spares both sides arguing durations one by one. Productivity rates, crew sizes, procurement lead times and commissioning durations should trace back to a tender estimate, supplier quotations or recorded performance on similar work. Then look hard at the interfaces behind each key milestone: utility connections, regulatory consents, rail possessions, plant outages, long-lead equipment and anything the owner itself has promised to supply. An owner often finds that the riskiest assumptions in a contractor's baseline are the owner's own obligations, and that is a finding worth having before acceptance.

Step 3: how likely is the completion date?

The likelihood of the completion date comes from a probabilistic forecast, which turns duration uncertainty into a range of finish dates with a confidence level attached to each. GAO's guide notes that some programmes treat the 80th percentile as a conservative promise date and others use lower percentiles, and that schedule outcomes tend to skew late, so the owner should decide its tolerance before the numbers arrive.

A short scenario shows why the percentile matters more than the date itself.

Illustrative scenario (assumptions labelled, not a real project). The contract completion date is 30 June 2028 and the submitted baseline also finishes on 30 June 2028 with zero total float. A probabilistic forecast of the same schedule puts the 50 percent confidence (P50) finish at 15 September 2028 and the 80 percent confidence (P80) finish at 8 December 2028.

That puts the P50 finish 77 days after the planned date (31 days in July, 31 in August and 15 in September) and the P80 finish 161 days after it (31 + 31 + 30 + 31 + 30 + 8). The schedule describes a credible sequence; it simply carries no contingency against the uncertainty in its own durations, and that gap is the conversation the owner now needs to have with the contractor.

Many owners and contractors produce that forecast through a quantitative schedule risk analysis (QSRA), in which the team agrees three-point duration ranges in workshops and runs them through a Monte Carlo simulation. QSRA is formal, widely understood and often written into contracts, and its ranges carry the experience of the people who know the project best. Because it takes that collective effort, it is usually refreshed at set points in the project, which is where a second view can complement it between workshops.

This is the point where we can help. With nPlan Insights Pro, we forecast every activity in the submitted schedule using what our models have learnt from how past projects actually performed, map the driving paths most likely to affect the completion milestone, and highlight the activities and themes with the greatest delay-causing potential. Those findings turn a general worry about optimism into a short agenda for the contractor: these paths, these activities, and the evidence that would make us comfortable with them. Mitigation scenarios can then show how much a proposed resequence or earlier procurement would move the forecast. Our article on traditional risk software and nPlan explains how this AI-led schedule risk analysis (AI-SRA) works alongside QSRA.

On the Transpennine Route Upgrade in northern England, the project team built nPlan's outputs into its T-minus reviews at 12, 8 and 4 weeks before major track possessions, alongside its traditional methods, according to nPlan's published Transpennine Route Upgrade case study (October 2024). The same case study records that after one Driving Paths review, the TRU East team raised risks to key possession dates with senior stakeholders. The forecast gives your team an independent view to test in the same way. Your team still validates scope, site conditions and delivery assumptions, reconciles the findings with the risk register, and decides whether the evidence supports acceptance or a revised submission.

When is a different check the better call?

An activity-level forecast is not always the right first tool. If the submission is a summary schedule with missing logic, fix the structure under step 1 first, because no forecast can be better than the network it reads. If there is no detailed schedule yet, a top-down comparison with similar completed projects is more useful, as our piece on reference class forecasting explains. And if you want a fully independent comparator built from the same scope documents, Schedule Studio can draft one for your planners to review. Where the contract prescribes a particular risk analysis, follow it and use other views as a challenge to its results.

What does the contract let an owner do?

The contract decides how much leverage an owner has over a baseline it doubts, so read its scheduling clauses before writing comments. The reference points differ between Great Britain and the US, so each is set out separately below.

What can an owner do under NEC4 in Great Britain?

Under NEC4, the Project Manager either accepts the programme or gives reasons for not accepting it, and one permitted reason in clause 31.3 is that the programme "does not represent the Contractor's plans realistically". The other listed reasons are impracticable plans, missing required information and non-compliance with the Scope, as Brodies explains. NEC's own guidance adds that reasons must be given, and that withholding acceptance for a reason not in the contract is a compensation event.

Timing matters as much as the reasons. The CECA NEC4 bulletin (June 2021) notes that after two weeks of silence the Contractor can notify the Project Manager, and after a further week the programme is treated as accepted.

What can an owner do on US federal construction contracts?

On US federal construction contracts, FAR 52.236-15 requires the contractor to submit a schedule within five days after work begins, or another period the Contracting Officer sets, and lets the Contracting Officer withhold progress payments until it arrives. GAO uses its guide to assess federal programme schedules, and AACE International's recommended practice 78R-13 covers original baseline schedule review for the people deciding whether a baseline is acceptable to the owner.

On private US work, the contract's own scheduling specification governs, so read it first.

Whichever contract applies, the same discipline helps: specific reasons, tied to evidence, delivered inside the review window. The two tables below turn each concern into that kind of reason.

What evidence should an owner request before accepting the baseline?

An owner should request the evidence that would close each specific concern, name who supplies it, and agree the review action before acceptance. GAO's guide lists similar items for auditors, including the native schedule file rather than a PDF, the basis document, the duration rationale, the risk analysis and the current risk register.

The evidence only helps if someone on the owner's side acts on it before the acceptance window closes. The second table sets out that action for each concern.

Once each row is closed or carried as a recorded condition, the decision becomes straightforward to write down: accept, accept with conditions the contractor has agreed, or return the programme with reasons tied to specific rows. Either way, both parties start delivery with a shared view of how likely the date is and what would move it.

Frequently asked questions

What if the contractor will not remove a date constraint on the finish?

Ask for the written justification GAO expects for every hard constraint, and review a copy of the file with the constraint relaxed so both parties can see where the logic alone puts the finish. If the constraint stays without a justification, record it as a reason or a condition in your response.

How often should the owner repeat this check after accepting the baseline?

Repeat it whenever a revised programme arrives, and run the forecast again on each update to see whether the likelihood of the completion date is improving or slipping. Under NEC4 the interval is set in the Contract Data, commonly every four weeks. On US federal work, GAO expects each update to use actual progress and come with a narrative of what changed.

What is the difference between a P50 and a P80 completion date?

A P50 completion date is the date the project has a 50 percent chance of finishing on or before, and a P80 date has an 80 percent chance. GAO's Schedule Assessment Guide notes that some programmes treat the 80th percentile as a conservative promise date while others use lower percentiles. The owner should agree which confidence level it needs before the forecast arrives.

Does an AI forecast replace a schedule risk workshop or QSRA?

No. An activity-level forecast drawn from past project performance gives the team a second, independent view to test against the ranges and risk register the QSRA already uses. The project team still validates the assumptions and makes the acceptance decision.

Sources

1. GAO Schedule Assessment Guide: Best Practices for Project Schedules (GAO-16-89G), December 2015
2. Brodies LLP, On what grounds can a Project Manager reject a programme submitted for acceptance by a Contractor?, May 2021
3. NEC Contracts, FAQ: categorising reasons for non-acceptance
4. CECA NEC4 Bulletin No. 6: Importance of an accepted programme every month/period, June 2021
5. FAR 52.236-15 Schedules for Construction Contracts
6. AACE International Recommended Practice 78R-13: Original Baseline Schedule Review, As Applied in Engineering, Procurement, and Construction
7. nPlan, Using AI to de-risk delivery during the Transpennine Route Upgrade project (published case study, October 2024)
8. nPlan Insights Pro
9. nPlan, Traditional risk software vs nPlan
10. nPlan, Reference Class Forecasting (RCF) vs nPlan
11. nPlan, Who Schedule Studio helps, and how
12. nPlan, Why we built Driving Paths